Topics
Contents APPLICATION TECHNOLOGY, 50 Article(s)
Improved Res-UNet-based vascular segmentation of retinal images
Han YANG, Baicheng LI, and Lingling CHEN

Accurate retinal vascular segmentation supports the treatment of diseases such as diabetes and hypertension. Because of the complex vascular structure of the eye, the complexity of the pathological features leads to many limitations in the accuracy and speed of vascular segmentation. To overcome this problem, an improved U-net segmentation method is proposed, which replaces the convolution module in the U-net network decoder and encoder with a residual module, using a non-local attention module to connect the encoder and decoder. The network model enhances the correlation of pixel information and the ability to extract features without increasing the number of parameters. Finally, the DRIVE dataset was used for comparison and evaluation with the original U-net network, and the model achieved 0.9679、0.9896、0.8245 and 0.8281 of feature detection accuracy, specificity, sensitivity and Dice coefficient on the test set, respectively. The experimental results demonstrate that the proposed network model can perform accurate vascular segmentation of the retina.

Optical Instruments
Aug. 25, 2023, Vol. 45 Issue 4 24 (2023)
Design of roughness detection system based on transfer learning and model fusion
Qiang ZHANG, Zhiwen HUANG, and Jianmin ZHU

According to the problem of low recognition accuracy of traditional roughness measurement methods, a roughness detection method based on transfer learning and model fusion was proposed. Firstly, the CCD module in the roughness detection system was used to collect the workpiece surface images and construct a data set. Secondly, through the migration fine-tuning VGGNet-19, Inception-V3 and DenseNet121 multi-model fusion, a suitable roughness detection model is obtained by multi-model fusion. Finally, the data set is used for network training to extract the texture details from the images and achieve accurate recognition of the roughness level. The experimental results show that 15 different roughness level images from turning, milling and grinding are used, and the recognition accuracy of the system can reach up to 91%. The results show that the proposed system can effectively realize the automatic detection of roughness grade.

Optical Instruments
Aug. 25, 2023, Vol. 45 Issue 4 17 (2023)
Preparation of red solid carbon quantum dots and its application in fingerprint detection
Zhijin YANG, Zhaoxia HAN, Zhihong ZHANG, Shuhui DING, Dawei ZHANG, Ruijin HONG, Chunxian TAO, Hui LIN, and Xueke XU

Fluorescent carbon quantum dots (CQDs) are a new type of photoluminescent nanomaterials. Due to their stable luminescent properties, abundant surface functional groups, safety, non-toxicity, good biocompatibility and low cost, they have great application prospects in the field of latent fingerprint detection and recognition. Latent fingerprints refer to the traces of human finger secretions left on the solid contact surfaces that are difficult to distinguish with the naked eyes, which require effective physical or chemical methods to effectively display and extract. So far, there are few reports on the development of latent fingerprints using CQDs and their precise identification in combination with computer technology. In this paper, red CQDs were successfully synthesized by one-step solvothermal method using o-phenylenediamine as the precursor and zinc oxalate as the modifier. The prepared red CQDs were mixed with polyvinylpyrrolidone, dried and ground to prepare red solid luminescent CQDs with uniform dispersion and quantum yield up to 27%, which were successfully applied to enhance the detection of latent fingerprints on various substrates. In order to accurately evaluate the similarity between the developed latent fingerprint and the target reference fingerprint, the structural similarity algorithm was used for similarity analysis. The matching degree of the latent fingerprint on the tin foil was as high as 90.5%, indicating that the combination of red solid fluorescent CQDs with digital processing programs can effectively develop and accurately identify latent fingerprints, which has great application prospect in the field of criminal investigation.

Optical Instruments
Jun. 25, 2023, Vol. 45 Issue 3 37 (2023)
Calibration and correction of 3D video based on short baseline binocular endoscope imaging system
Chaocai ZHANG, Wei ZHANG, Defeng ZHOU, Xiaoxiao WEI, and Xinjun WAN

In order to solve the problem that the video depth and stereoscopic sense of the video obtained by the short baseline binocular endoscopic imaging system are weak in the naked eye 3D display device, by analyzing the parameters of the binocular endoscope and the parallax of the image pair in the stereoscopic video, a stereo video correction and parallax adjustment method based on the short baseline binocular endoscopic imaging system is proposed. Firstly, the camera of the binocular endoscope system is calibrated to obtain the camera parameters and the position parameters between cameras; Secondly, the camera video is corrected by using the obtained parameters, and then the image parallax is adjusted according to the parameter requirements of the naked eye 3D display device for the video source, so as to finally obtain the real-time display stereoscopic video of the binocular endoscope system which meets the requirements of the naked eye 3D stereoscopic display device and is suitable for human eyes. The feasibility of the method is verified by experiments. A short baseline binocular endoscope imaging system with a baseline distance of 8mm is actually built. The original parallax range is (0,64) pixels, which can reach (-30,30) after parallax adjustment. The two-way parallel video processing is 25 frames/s for real-time display. Matching with the naked eye 3D stereoscopic display system designed in the laboratory, real-time naked eye 3D imaging of medical endoscope with obvious stereoscopic feeling can be realized.

Optical Instruments
Jun. 25, 2023, Vol. 45 Issue 3 30 (2023)
Automatic recognition of water source microorganisms based on particle swarm optimization algorithm
Xingang MIN, Shaoqi HUANG, Shaojie YOU, and Bo DAI

The detection of water source microorganisms is of great significance to the biosafety of water source and so on. However, the traditional methods such as microscopic observation are inefficient and need professional personnel. Therefore, an automatic recognition method of micro-organisms in water source is proposed. Water samples were collected and a microorganisms image set was made. Automatic and semi-automatic image segmentation algorithms were proposed to extract the target microorganisms area, and 6 features were extracted. The model optimization problem of water microorganisms classification process was studied. First, the parameters of a few features were optimized. Then, all the features were fused, and a microorganisms recognition model of support vector machine optimized by particle swarm optimization (PSO-SVM) was established and compared with other recognition algorithms. The results show that, compared with the other 3 recognition algorithms, PSO-SVM can recognize different kinds of microorganisms more effectively, with an average recognition rate of 97.08%.

Optical Instruments
Apr. 25, 2023, Vol. 45 Issue 2 8 (2023)
Design of ABS sensor output sinusoidal signal measurement device based on STM32F4
Mengyuan REN, Fangling HUANG, and Rongfu ZHANG

The automobile wheel speed sensor is one of the key components of anti-lock braking system (ABS), and its performance directly determines the level of ABS performance. The output of the magnetoelectric ABS speed sensor is approximately a sinusoidal signal, and the performance of the sensor can be obtained by analyzing the amplitude, frequency and effective value of the sinusoidal signal. This work takes STM32F407 as the main control board and uses fast Fourier transform as the main tool to design a digitally controlled sinusoidal signal measurement device to measure various parameters. It has been verified by experiments that the device can accurately measure the frequency, effective value, peak value and distortion of the input signal within 5% error and restore the signal waveform. At the same time, in order to facilitate observation, this design uses the bluetooth communication module to synchronously upload the measurement data to the mobile phone.

Optical Instruments
Apr. 25, 2023, Vol. 45 Issue 2 36 (2023)
Fruit damage detection and classification based on attention mechanism
Jie ZHANG, Chunlei XIA, Rongfu ZHANG, Julaiti HALIZHATI, and Yi LIU

For the daily essential food of people, automatic damage detection and automatic classification are essential for the increasing consumption of fruit. In view of this demand, automatic detection of fruit damage has become a hot topic in recent years. In this paper, the application of convolutional neural network, an existing deep learning technology, in fruit feature extraction and classification was discussed. A method based on ResNet34 as the backbone network and the introduction of attention mechanism SE and CBAM module was proposed to realize the detection and basic classification of fruit damage. The method was verified on fruit fresh and rotten for classification data set, and compared with VGG16, GoogLeNet, MobileNetV2 and other common networks. The accuracy of fruit damage detection and classification is improved. The classification accuracy reaches 98.8%. By adding the new apple data set, the performance of the model is further improved, compared with the original network ResNet34, and the generalization of the model is effectively improved, which provides a reference for the complex multi-feature classification of actual fruit images.

Optical Instruments
Apr. 25, 2023, Vol. 45 Issue 2 26 (2023)
Optical fiber sensing vibration signal recognition based on lightweight network
Lingling CHEN, Baicheng LI, Dawei ZHANG, Han YANG, and Chunbo WU

Based on the application of distributed optical fiber sensing system in the field of perimeter security monitoring, there are problems such as slow response speed and low recognition rate. Although the recognition rate of the traditional convolutional neural network is very high, its huge amount of parameters makes industrial deployment difficult and the recognition response speed is slow. This paper introduces the lightweight convolutional neural network MobileNet, which uses depth-separable convolution to replace the traditional convolution, which greatly reduces the amount of model parameters. This paper uses MobileNet as the benchmark network to implement a one-dimensional lightweight network based on MobileNet-18 Φ-OTDR perimeter intrusion event recognition, compared the network recognition rate and recognition speed under different structures through experiments, and selected MobileNet-18 as the best model under the condition that the accuracy of the model would not be greatly reduced. In the experiment, six perimeter fiber intrusion signals of climbing, cutting, wind blowing, lifting, pulling and walking were collected. Among the six types of fiber intrusion signal recognition, MobileNet-18 achieved a recognition rate of 98.33% and a response time of 9.27 ms.

Optical Instruments
Apr. 25, 2023, Vol. 45 Issue 2 18 (2023)
Face image frontalization method for face expression analysis
Xuedian ZHANG, Zhongjun CHEN, and Xiaofei QIN

In the process of face expression analysis, head pose changes often cause asymmetry in face information, and it is difficult to obtain features that are robust to pose by traditional operations related to cropping and aligning face images only. In order to obtain structured features of faces, a face image frontalization processing method is proposed in the paper. The method maps the detected face landmarks to a new two-dimensional space for frontalization of landmarks, then restores the frontalized landmarks to the original image as new landmarks, and guides the image deformation from the original landmarks to the new landmarks by moving least squares to obtain the frontalized face image. The face images are preprocessed on public RAF-DB and ExpW face expression datasets using the proposed processing method, models are trained in VGG16 and ResNet50 deep learning networks for face expression classification tasks. The effectiveness of the frontalization method in the paper for face expression analysis is evaluated by the accuracy of the classification tasks. The experimental results show that this proposed method outperforms traditional pre-processing methods of deep learning in face expression analysis and can effectively improve the quality of face information.

Optical Instruments
Feb. 25, 2023, Vol. 45 Issue 1 8 (2023)
Application of XGBoost machine learning in error compensation of photoelectric encoder
Yingzheng LI, Zhibin LI, Lei JIN, Zhenzhen HU, Yefei KANG, and Gengbai LI

The error of photoelectric encoder detection system is mainly affected by the angle measurement error of the reference photoelectric encoder, data acquisition error and coaxial error. The angle measurement error can be compensated. In this paper, an algorithm based on extreme gradient boosting (XGBoost) machine learning is designed to compensate the error of the reference photoelectric encoder. After compensation, the static accuracy is improved by 35 times. The standard deviation is decreased from 3.62" to 0.13", and the maximum error value is reduced from 5.53" to 0.39". Compared with the traditional back progagation (BP) neural network algorithm and radial basis function (RBF) neural network algorithm, XGBoost's compensation is better than the others. XGBoost machine learning algorithm compensation effectively reduces the measurement error of the reference photoelectric encoder and improves the detection accuracy of the photoelectric encoder detection system.

Optical Instruments
Feb. 25, 2023, Vol. 45 Issue 1 32 (2023)
Longitudinal Raman signal detection based on cylindrical vector beam
Wanxiang XU, and Jian LIN

A method for achieving the Raman signal excited by the longitudinal component of radially polarized light is proposed in this paper to solve the problem that ordinary polarized Raman signal cannot detect samples with vertical molecular orientation. In this method, the focused azimuthally polarized light and radially polarized light are used to excite the sample, and then the Raman signals generated by the two excitation lights are calculated to obtain the Raman signals excited by the pure longitudinal light field. We use Si(0 0 1) samples and Si(1 1 0) samples to verify the feasibility of this method. This work has certain significance for measuring longitudinally oriented molecules and longitudinal molecular vibration modes by Raman scattering.

Optical Instruments
Feb. 25, 2023, Vol. 45 Issue 1 25 (2023)
Virtual staining techniques for cellular microscopic imaging
Hao ZHANG, Bo DAI, and Dawei ZHANG

Cell microscopic imaging is an important tool for cell phenotype detection. Traditional fluorescence imaging techniques are widely used in the cell imaging. However, the fluorescence imaging instruments have complex structure and high cost. Besides, staining could cause damage to cells. To address this problem, this paper proposes a virtual staining technique that performs strict alignment of bright field and fluorescence image datasets using a multimodal alignment algorithm, and improves network architecture, loss function, post-processing, and hardware adaptation for training optimization. The staining conversion bias is calculated by the evaluation criteria of the virtual staining. The method presented in this paper could simplify the fluorescence imaging equipment and eliminates the need for various staining operations, which could reduce the burden of research and diagnostic processes for biologists and pathologists.

Optical Instruments
Feb. 25, 2023, Vol. 45 Issue 1 18 (2023)
Femtosecond fiber laser oscillator by nonlinear polarization rotation mode locking technique
Jianing CHEN, Han YU, Lu SI, Junyu CHEN, Tao JIAN, and Shuai YUAN

Nonlinear polarization rotation (NPR) mode-locked lasers have attracted numerous attention due to their compact structure and high reliability. Based on this mold locking principle, an ytterbium-doped fiber femtosecond laser was designed and built. When the bidirectional pumping power reached 380 mW, the laser was mode-locked with a fundamental frequency repetition rate of 22.8 MHz. The output pulse was centered at 1030 nm with 224 fs for the pulse duration, 180 mW for the average power, 8 nJ for pulse energy, and 40 nm for the bandwidth at 10 dB. The beam had a good quality with signal-to-noise ratio greater than 50 dB. The laser can realize self-start mode locking with a ring laser structure. Besides, the third order passive harmonic mode locking with 68.5 MHz for the repeat frequency was observed by increasing the pump power to 1.6 W. Due to the advantages of line width, pulse width and pulse energy, the laser has application significance in spectral measurement, Raman imaging and other fields.

Optical Instruments
Feb. 25, 2023, Vol. 45 Issue 1 1 (2023)
Fabrication of resistive memory devices based on graphene oxide
Min CHEN, and Qiming ZHANG

Resistive memory devices are new type of non-volatile memory devices that can be switched between high resistance state (HRS) and low resistance state (LRS) under the action of an external electric field. The selection and interaction of electrode materials and active layer materials are the main factors for realizing the resistive switching characteristics of devices. Graphene is a two-dimensional (2D) material with excellent electrical conductivity and high ductility. Reduction of graphene oxide (GO) by laser processing is an excellent method to obtain graphene efficiently. The preparation process of traditional memory devices is complicated, which is not conducive to large-scale processing and manufacturing. Using metal Au and reduced graphene oxide (rGO) as electrodes, and GO as the active layer for device fabrication, the resistive switching function of the memory device is well realized. The simple and efficient fabrication method provides a reference for the large-scale and highly integrated production of resistive memory devices.

Optical Instruments
Dec. 25, 2022, Vol. 44 Issue 6 8 (2022)
Terahertz metalens with extended focal depth and polarization controllable
Yang ZHU, and Xiaofei ZANG

In this letter, a pure geometric phase based all-dielectric metalens is designed and aimed to enhance the tolerance of imaging in the longitudinal direction. Numerical simulation for the designed metasurface is carried out using the pure geometric phase spin decoupling design method combined with the finite difference time domain (FDTD) method. The calculated results show that designed metasurface enables the characteristics of controllable polarization and extended focal length, achieving a focal depth of 8 mm along the propagation direction, compared with 4.5 mm for a conventional metalens. Furthermore, we also design and numerically validate a terahertz metalens that can generate two focal points with extended focal length, and the polarization states of the two extended focal points are orthogonal to each other, demonstrating the multiplexing functionality of the designed metalens. The unique and effective approach may open a new avenue for tomography and information encryption.

Optical Instruments
Dec. 25, 2022, Vol. 44 Issue 6 23 (2022)
Optical surgical navigator based on binocular vision and mixed reality
Shao DONG, Xinjun WAN, Zhiqiang FANG, and Lizhengyi SHEN

The current surgical navigation system requires the doctor’s sight to switch back and forth between the display screen and the patient, which can easily cause visual fatigue and affect the safety of the operation. In response, an optical surgical navigator based on mixed reality was introduced herein, which enabled the navigation information to be directly fused into the surgical area. The navigator was based on mixed reality (MR) glasses and a binocular vision tracker system. A novel virtual-real registration scheme was proposed, which established the mapping relationship between the virtual model coordinate system and the real world based on the tracker measurement. The wireless communication between the data processing unit and the MR glasses was realized, and the virtual model can be overlapped with the patient’s surgical area for real-time navigation. The accuracy of the optical navigator was verified with the virtual-real fusion display experiment based on a skull model sample. The results show that the dislocation error between the virtual model and the real sample is about 4.5 mm, and the display refreshing rate reaches about 15 frames per second, which can verify the feasibility of the proposed scheme and its potential for practical clinical application.

Optical Instruments
Dec. 25, 2022, Vol. 44 Issue 6 14 (2022)
Recognition of PCB solder joints and core wire ends
Hanlin LIU, and Rongfu ZHANG

Intelligent miniaturized medical devices have attracted more and more attention in the medical industry. These products are mainly composed of some micro electronic components, in which the point line connection structure of the device core chip needs to be completed by high-precision welding. Therefore, the higher recognition accuracy of solder joint and welded core wire is required. Whether they can be accurate and effective identification directly affects the final quality of welding. In order to complete the recognition of solder joint and core wire in the welding process with high quality, this study mainly uses the method of electronic microscope instrument combined with host computer VS17 + OpenCV to complete the image processing, and recognizes the solder joint and core wire end of the collected image. The color and geometric features of solder joint and core wire end are taken as the analysis object. After preprocessing, the region of interest is highlighted through their respective feature analysis. The segmentation process of solder joint and core wire end is completed through specific color threshold selection method and contrast enhancement algorithm. The recognition accuracy error of measured solder joint and core wire end is required within 0.1 mm. The experimental results show that the proposed identification algorithm of printed circuit board (PCB) solder joint and core wire end can effectively identify the position of solder joint and core wire end in the image and display its pixel coordinate value. After data sorting and analysis, the identification accuracy error of this algorithm is controlled within the tolerable range.

Optical Instruments
Oct. 25, 2022, Vol. 44 Issue 5 42 (2022)
Preparation and application of Au-Cu bimetallic surface-enhanced Raman scattering substrate
Hao MA, and Ling ZHANG

Bimetallic materials combine the characteristics of two kinds of metal and exhibit superior properties, which is widely concerned in the field of material applications. Preparing high performance functional materials at low cost is the key to the promotion and application of bimetal. Using Cu30Mn70 alloy as precursor, gold coated nanoporous copper substrate (Au@NPC) was fabricated by free corrosion and subsequent chemical plating. In addition to limiting the oxidation of copper in air, the gold coating reduces the aperture distance between ligaments, increases the electromagnetic coupling effect between adjacent ligaments, and make Au@NPC have stronger local electromagnetic field enhancement characteristics, which is better than the surface enhanced Raman scattering characteristics of nanoporous copper. Gold-copper bimetallic materials can be used as surface Raman enhanced scattering substrates with excellent stability at low cost.

Optical Instruments
Oct. 25, 2022, Vol. 44 Issue 5 35 (2022)
Multiscale hypergraph convolutional network for skeleton-based action recognition
Xiaofei QIN, Ying ZHAO, Yijie ZHANG, Ruijie DU, Hanwen QIAN, Meng CHEN, Wenqi ZHANG, and Xuedian ZHANG

Action recognition is one of the basic tasks of computer vision. The skeleton sequence contains most of the action information, so skeleton-based action recognition has attracted a lot of research attention. Mathematically, the human skeleton is a natural graph, so graph convolution is widely used in action recognition. But ordinary graph convolution only aggregates low-order information between pairwise nodes, and cannot model high-order complex relationships between multiple nodes. To solve this problem, a multiscale hypergraph convolutional network is proposed, which aggregates richer information in the two dimensions of space and time, so as to improve the accuracy of action recognition. The multiscale hypergraph convolutional network has an encoder-decoder structure. The encoder uses the hypergraph convolution module to aggregate relevant information between multiple nodes in the hyperedge, and the decoder uses the hypergraph fusion module to restore the original skeleton structure. In addition, a multiscale temporal graph convolution model based on dilated convolution is designed, which is used to better aggregate the temporal-dimension motion information. The experimental results on NTU-RGB+D and Kinetics datasets verify the effectiveness of this algorithm.

Optical Instruments
Aug. 25, 2022, Vol. 44 Issue 4 39 (2022)
Design and implementation of high-speed balanced homodyne detector
Xin TENG, Huimin SHE, and Yan LIANG

Balanced homodyne detectors have the disadvantages such as small common mode rejection ratio (CMRR) and low response bandwidth, etc. To address the problems, we designed a scheme to test the response characteristics of the photodetector (PIN photodiode) in the detector and selected devices with similar performance for balanced detection to improve the CMRR. Furthermore, the scheme of cascading transimpedance amplification and proportional operational amplifier were proposed to extract the photocurrent signal, which solved the problem of the limitation of the gain bandwidth product of the primary sampling amplifier and improved the response bandwidth of the detector while ensuring the sensitivity. As a result, we achieved 300-MHz balanced homodyne detection and the CMRR went up to 66 dB. We also detected the quantum shot noise in the optical signal. These achievements provide an effective technical scheme for high-speed sensitive detection.

Optical Instruments
Aug. 25, 2022, Vol. 44 Issue 4 32 (2022)
Image quality enhancement method for capsule endoscope system
Xingqi WANG, and Bo YANG

Aiming at the problems of the current capsule endoscope, such as low resolution, limited field of view, and great influence by noise, a system solution was proposed. First, by introducing Q-type aspheric surface to correct aberrations, an endoscope imaging lens with a full field of view of 160°, a relative aperture of F#3.0, and a total system length of 4.3mm was finally obtained. MTF values at 140 lp/mm were all greater than 0.3. The imaging quality of an optical system depends not only on the performance of the lens, but also on the image sensor, especially in low-light environments. The noise model was obtained by analyzing the noise characteristics of each stage when the sensor was working. Using the established noise model, the capsule endoscope image data set was synthesized, and the neural network model was trained. The test results of the algorithm model show that the proposed comprehensive solution can effectively improve the imaging quality of capsule endoscope system.

Optical Instruments
Aug. 25, 2022, Vol. 44 Issue 4 26 (2022)
A dual-branch network for action recognition
Xiaofei QIN, Rui CAI, Meng CHEN, Wenqi ZHANG, Changxiang HE, and Xuedian ZHANG

Action recognition has always been an important task in the field of computer vision. There are mainly two tasks based on RGB video and human skeleton. The mainstream methods are 3D convolutional neural network and graph convolutional neural network. For the data modality of human skeleton, this work designs a graph convolutional neural network based on the self-attention mechanism. The algorithm can achieve advanced performance on skeleton-based action recognition tasks. In addition, a method is proposed to use deep supervision methods to supervise the intermediate features of video and human skeleton, which improves the coupling of the two data features and further improves network efficiency. The network structure of this algorithm is simple, and only 3.37×107 parameters are used to achieve an accuracy of 95.6% on the NTU-RGBD60 (CS) dataset.

Optical Instruments
Aug. 25, 2022, Vol. 44 Issue 4 16 (2022)
Achromatic metalens compund lens
Min ZHANG, Yongzheng LU, Jia PENG, and Jing WEN

Metalens has the prospective to replace the traditional lens for multifunctional sub-wavelength imaging due to its thinness and ease of integration. The beam deflection angle of the metalens increases with the increase of wavelength and produces opposite dispersion compared to the traditional refractive lens. This form of dispersion is also called "abnormal dispersion" or "negative dispersion". Theoretically, the negative dispersion of the metalens and the positive dispersion of traditional refractive optical devices can be used to fully offset the chromatic aberration of the optical system. Consequently, this study demonstrates the design of an achromatic metalens compound lens based on photoresist material by employing Pancharatmane-Berry (PB) phase metalens as the first lens and traditional spherical lens as the second lens, through finite difference time domain method. The focusing characteristics of the lens at λ = 780~980 nm are investigated by FDTD Solutions (a numerical simulation software) and an exceptional achromatic effect of photoresist metalens compound lens is realized. Compared with the traditional achromatic metalens, the achromatic design is simple and efficient which provides a certain significance for achromatic imaging in the specified bandwidth.

Optical Instruments
Jun. 25, 2022, Vol. 44 Issue 3 37 (2022)
163 MHz/786 fs high repetition rate Yb-doped femtosecond fiber laser oscillator
Jiawei SHEN, Yong WANG, Lu SI, Jianing CHEN, Li ZHOU, and Shuai YUAN

High repetition frequency fiber laser plays an important role in the optical frequency comb, industrial processing, ultra-high speed optical sampling and other fields due to its advantages of good beam quality, low power consumption, and high integration. A 163 MHz fundamental repetition rate soliton mode-locked fiber femtosecond laser oscillator is designed and constructed based on nonlinear polarization rotation (NPR) technique. The laser can produce a maximum power of 200 mW at a pump power of 450 mW, with a 30 nm bandwidth and a 786 fs output pulse width duration. By further analyzing the relationship between the pumping power and the output power, it is found that the best operation condition for the laser is at 400~450 mW for the pump power. At the same times, the laser can realize self-starting mode-locking. The proposed laser is of great potential in the application fields like precision spectroscopy and astronomical exploration because of its compact optical frequency comb, high image quality and high imaging speed.

Optical Instruments
Jun. 25, 2022, Vol. 44 Issue 3 31 (2022)
Nanoporous gold island fabrication and SERS property investigation
Junjie ZHOU, and Ling ZHANG

Surface-enhanced Raman scattering (SERS), which combines nanostructured materials with Raman spectroscopy, solves the problem of low sensitivity of traditional Raman scattering technique and provides a new technical means for trace detection. High SERS active substrate is the key to the application of Raman spectroscopy for trace detection. The island nanoporous gold SERS substrate was prepared by dealloying precursor alloy films which sputtered on silicon wafer, and the obtained substrate posses much smaller ratio between pore size and gold ligament size compared with traditional nanoporous gold, resulting in much stronger local electromagnetic field due to improved coupling effect between gold ligaments. The detection limit of the substrate can reach about 10-10 mol·L-1, and the relative intensity of SERS spectrum has a good linear relationship with the concentration, and the dynamic range can reach 3 orders of magnitude. Meanwhile, with uniform structure the substrate exhibits stable performance, and the preparation process shows good repeatability.

Optical Instruments
Jun. 25, 2022, Vol. 44 Issue 3 23 (2022)
Study on metal spring based terahertz waveguide
Zhanghua DONG, and Jiayu ZHAO

In order to understand the transmission effect of the metal spring waveguide on terahertz wave, the transmission characteristics of terahertz wave by springs with different helix spacings have been studied experimentally. The experimental results show that when the pitch is larger as 3.5/4.4 mm (with wire diameter of 0.8 mm, outer diameter of 12 mm and length of 14 cm), the metal spring waveguide can propagate terahertz wave in a larger bandwidth with the additional ability of polarization maintaining. The terahertz transmission bandwidth of metal springs with pitch of 3.5/4.4 mm are both about 0.9 THz, and the lowest transmission loss are about 0.2 cm-1 and 0.27 cm-1 respectively at their peak frequencies. In addition, it has been proved that the terahertz mode was confined inside the air core rather than being guided by the metal helix. The above results are helpful for applications of the metal spring waveguide in the field of terahertz technology.

Optical Instruments
Feb. 25, 2022, Vol. 44 Issue 1 29 (2022)
High-speed photon-number resolving detection with MPPC
Ying SUN, and Yan LIANG

In order to improve the multi-pixel photon counter (MPPC) high-speed detection and improve the photon-number to distinguish features, we use the gated suppression MPPC, and combine the self-balancing and low-pass filtering technology to suppress the capacitive-spike noise of MPPC down to the thermal noise level, realizing the linear extraction and rapid recovery detection of photogenerated avalanche signal. The experiment result shows that the 200 MHz gated MPPC achieves the effective resolution of 14 photons of the 40 MHz repetition frequency lase with the average detected photon number is up to 6.8 per pulse. Compared with the passive suppression mode, the photon number resolution in the 200 MHz sinusoidal gated mode is obviously improved, which provides a reference for high-speed photon-number resolving detection.

Optical Instruments
Feb. 25, 2022, Vol. 44 Issue 1 22 (2022)
Simulations of reconfigurable optical switching based on Ge2Sb2Se4Te1
Zhangjian MU, Liying LI, Jia DU, Nan CHEN, and Xuejing LIU

Optical switch is an important component of integrated optical circuit. In this paper, an on-chip 2×2 directional coupler reconfigurable optical switch based on one kind of chalcogenide phase change materials (Ge2Sb2Se4Te1) used in the L and C bands was proposed. The switch can be switched by changing the phase state. The optical switching device was designed by means of Mode and FDTD Solutions in the simulation software Lumerical. The results showed that the insertion losses (ILs) were more than -0.36 dB and the crosstalks (CTs) were less than -24 dB in the amorphous state of Ge 2Sb2Se4Te1 with a coupling length of 24.9 μm in the band from 1500 nm to 1625 nm. The ILs were more than 0.44 dB and the CTs were less than -30.46 dB in the crystalline state. Multi-physics simulation software COMSOL was used to simulate 532 nm wavelength laser heating Ge 2Sb2Se4Te1. The results showed that a 25 ns short Gaussian pulse with a peak power of 45 mW could transform the material from a crystalline state to an amorphous state. Applying multiple Gaussian pulse arrays with a peak power of 20 mW, a period of 1 μs and a duty cycle of 0.03% could bring it to the crystalline state. Simulation results showed that the designed optical switch can quickly switch optical path in the communication band by laser heating.

Optical Instruments
Dec. 25, 2021, Vol. 43 Issue 6 6 (2021)
Single-frame image eyeball tracking based on deformable convolution
Jian WANG, and Rongfu ZHANG

In order to improve the accuracy of the eye tracking algorithm and ensure a certain image processing speed, this paper proposes to combine the deformable convolution method to improve the feature distribution extraction level. The fixed-size sampling in the standard convolution makes it difficult for the learning network to adapt to the geometric deformation of the image. In order to solve this limitation, the deformation modeling ability of deformable convolution is used to add a certain offset variable to the position of each sampling point in the convolution kernel. So as to achieve the extraction of potential features, the single frame of the original image is described. According to the current research, the deformable convolution has made preliminary applications in the field of computer vision. After comparing with the advanced eyeball positioning tracking detection network experiment, the accuracy of the deformable convolutional YOLO network can reach 0.685, and the average image processing speed can reach 42 frames per second, which is better than the original YOLO network and the advanced eyeball and location tracking detection network.

Optical Instruments
Dec. 25, 2021, Vol. 43 Issue 6 26 (2021)
Information hiding in plain-text images based on multi-objective optimization
Chanyu JIANG, Zixin ZHANG, and Chuan QIN

In traditional information hiding algorithms, embedding rate and image visual quality are mutually restricted, so the relationship between them cannot be balanced. Therefore, in this paper, a high fidelity scheme combining reversible information hiding method and irreversible information hiding method is proposed, which takes advantage of the high embedding rate of the irreversible information hiding method and the high peak signal-to-noise ratio (PSNR) of the reversible information hiding method in the aspect of image visual quality, so as to embed more secret information with less distortion. In addition, the scheme can also find the relative optimal embedding rate, the balance point between the restored image PSNR and the PSNR containing the dense image according to the specific application scene, so as to adapt to different application requirements. Experimental results show that the proposed scheme has higher embedding capacity and better image visual quality than the existing schemes.

Optical Instruments
Dec. 25, 2021, Vol. 43 Issue 6 19 (2021)
Taper multimode interference coupler based on terahertz
Hongxiang ZHANG, and Jingya XIE

With the rapid development of communication technology, the bottleneck appears in the traditional microelectronics industry. More and more people hope that integrated optical circuit can achieve new breakthroughs. Silicon photonics has attracted much attention because of its advantages in materials and manufacturing technology. Among them, the silicon coupler, as an important silicon passive device, is the key to achieve beam coupling and beam splitting on chip. Multimode interference (MMI) couplers are commonly used as integrated optical devices because of their advantages such as low loss, large process tolerance and large bandwidth. In this paper, based on the self-imaging principle of MMI, guided mode propagation analysis (G-MPA) is used to analyze the mode field distribution in MMI, and a taper multimode interference coupler based on 470 μm high resistance silicon wafer in terahertz band is successfully designed. The coupling efficiency of 93.8% is achieved by FDTD simulation and optimization of its parameters.

Optical Instruments
Dec. 25, 2021, Vol. 43 Issue 6 13 (2021)
Ignition process visualization of pulverized coal particles
Liang PENG, Wu ZHOU, Ronggen ZHENG, Wen DAI, and Xiaoshu CAI

Based on the McKenna burner, a dual light path imaging system was developed to synchronously collect the backlight projection profile image and the self-luminous light intensity distribution during the combustion of pulverized coal particles. Experiments were conducted on bituminous coal and anthracite particles separately, volatile ignition, soot formation, coke ignition were observed simultaneously, and corresponding analysis was conducted. It was found that the two coal types were mainly in homogeneous and heterogeneous ignition pattern, respectively. The separation phenomenon of solids and volatiles due to the interphase velocity difference between phases was observed during the burning of coal particles, and an image processing program was applied to calculate the corresponding motion speed. The study provides a reference for further quantitative analysis of the ignition and combustion characteristics of pulverized coal particles.

Optical Instruments
Feb. 25, 2021, Vol. 43 Issue 1 34 (2021)
Label free guided-mode resonance biosensor for DNA detection
Guangyu SUN, Shuhua CAO, Rui LI, and Qi WANG

To design a bio-sensor for DNA detecting, we proposed a polarization free transmission type guided mode resonance biosensor with high sensitivity in visible wavelength was proposed for DNA detection with the FDTD software simulation based on the rigorous coupled wave theory and equivalent theory. When ssDNA was detected, the resonance wavelength was 481.46 nm. When dsDNA was detected, the resonance wavelength was 490.11 nm. When no DNA is present in the detection area, no transmitted light appears. Hence, it can be used to distinguish between single and double-stranded DNA. This bio-sensor can maintain the biological activity of the sample and avoid physical or chemical damage of the sample, which is convenient for multiple analysis and detection, and owns potential applications in the field of biological detection.

Optical Instruments
Feb. 25, 2021, Vol. 43 Issue 1 28 (2021)
Research on the segmentation of optic disc and cup based on modified U-Net
Qian MAO, Minshan JIANG, and Jing WEI

In the diagnosis of glaucoma, segmentation of optic cup and optic disc based on digital fundus image is a common diagnostic method. In order to segment the cup and disc accurately, we proposed a segmentation method based on the improved U-Net. Compared with the traditional U-Net, a residual block was used to improve the down sampling part, and convolution part was used to improve the skip connection, so that the network could obtain more sufficient feature information. The Dice and IOU of the optic disc segmentation model and the optic cup segmentation model on DRISHTI-GS data set reached 98.3% and 97.2%, 93.2% and 88.5%.

Optical Instruments
Feb. 25, 2021, Vol. 43 Issue 1 21 (2021)
Optimization of target tracking algorithm based on region proposal Siamese network
Xiaofei QIN, Yipeng ZHANG, Haosheng CHEN, Xia LI, and Zhiyuan HE

In order to estimate the target scale in real time, avoid multi-scale test and improve the speed and accuracy of target tracking, a new optimized target tracking algorithm is proposed. By introducing the regional proposal network with good tracking effect into the common siamese network, and introducing the strip pooling module and the efficient channel attention module in the algorithm, we can deal with the scale difference of objects and the severe deformation in the tracking process. The proposed algorithm achieves 0.833 accuracy and 0.658 success rate on OTB100 dataset, 0.411 EAO index on VOT2016 dataset, and 0.275 EAO index on VOT2019 dataset.

Optical Instruments
Feb. 25, 2021, Vol. 43 Issue 1 14 (2021)
Terahertz tunable filter design
Jian SONG

In order to realize terahertz tunable filtering wave, a terahertz tunable filter based on flexible materials was designed. By twisting the Teflon waveguide to form a ring resonator, the band-stop filter function in the 160 GHz to 200 GHz frequency band is realized. Changing the length of the resonant cavity can achieve the tuning of the free spectral range (FSR) and the filtering frequency. In the experiment, the switching of the free spectral range between 1.9 GHz and 2.8 GHz and the corresponding filtering characteristics are tested. Research results show that when the length of the resonant cavity is fixed, the stopband can be adjusted by changing the bending radius. The terahertz ring resonator using flexible materials can be used for tunable filtering and has a higher degree of freedom.

Optical Instruments
Oct. 25, 2020, Vol. 42 Issue 5 7 (2020)
Near-field separated MIMO radar echo amplitude calibration method based on beam pointing adjustment
Qiliang WANG, and Ping LI

The echo amplitude of near-field separated multiple-input multiple-output ( MIMO) radar is different due to the different angle of each channel relative to the target. Therefore, a method of echo amplitude calibration based on beam pointing adjustment is proposed. By using the conversion relationship between the antenna beam pointing and the echo amplitude change, the indirect adjustment of the antenna echo amplitude is realized by adjusting the beam pointing so that the difference of echo amplitude between channels could be calibrated. The proposed method is verified by using a conventional geometrically separated MIMO array. The results show that the imaging performance of near-field separated MIMO radar can be improved significantly after the calibration of echo amplitude difference.

Optical Instruments
Oct. 25, 2020, Vol. 42 Issue 5 43 (2020)
Identifying diabetic retinopathy based on deep transfer learning
Yuming YAN, Feng LI, Deming LUO, Siyuan YIN, Xiaotian FU, Zheng LIU, and Lei YAN

In order to identify diabetic retinopathy (DR) in retinal fundus images automatically, to reduce the workload of ophthalmologists, and to develop an assistant tool in detecting and diagnosing retinal diseases, automated detection of DR images which uses deep transfer learning approach based on the Inception-v3 model is proposed. In the Inception-v3 model trained by ImageNet datasets, the parameters of the previous layers were fixed while the last fully-connected layer of the model was retrained by fine-tuning on the dataset collected by ourselves. Experimental results manifested the performance of the proposed approach providing better predictions and highly reliable detection without specifying lesion-based features, and it could help make automated screening for early DR based on retinal fundus images in addition to assisting ophthalmologists in making a referral decision.

Optical Instruments
Oct. 25, 2020, Vol. 42 Issue 5 33 (2020)
Terahertz spectroscopy for accurate identification of ginseng and panaxquinquefolium
Tianyi KOU

Since ginsenosides contained in the two herbs have different fingerprint spectra in the terahertz band, a method based on ginsenoside terahertz spectrum was proposed for the identification of ginseng and panax quinquefolium. The terahertz spectra of ginseng and panax quinquefolium were identified by MATLAB software, and the ginseng and panax quinquefolium were distinguished by the standard principal component analysis (PCA) according to their terahertz spectral characteristics. The results show that the principal component analysis based on terahertz spectrum can distinguish ginseng and panax quinquefolium accurately, and it can also be applied to other similar substances.

Optical Instruments
Oct. 25, 2020, Vol. 42 Issue 5 27 (2020)
Generation of Bessel beam array using broadband metasurfaces
Lei CHEN, Jing LU, and Jing WEN

Bessel beam arrays can be generated by loading a phase map on a spatial light modulator (SLM) or a microaxicon-like structures fabricated by lithography. However, the pixel size of a typical SLM is more than one order of magnitude larger than the wavelength of visible light, which limits the available range of phase gradient. And the tip of the micro-axicon-like structures fabricated by lithography is not a standard cone, which affects the quality of the Bessel beam. In order to overcome these shortcomings, a device that can generate Bessel beam array (at a wavelength of 700 nm, NA = 0.3) is designed by loading a complex phase map onto the dielectric metasurface. The device can work in a broadband, the polarization conversion efficiencies of one nanopost remains higher than 57% at the wavelength range from λ=590 nm to λ=800 nm. This device (thickness of 380 nm and diameter of only 40 μm) was simulated by the three-dimensional finite difference time domain (FDTD). The array beams generated were perpendicular to the metasurface device. The proposed Bessel beam array generator has a thickness of nanometer level and a diameter of several tens of micrometers, which has great application prospects in the field of integrated optics in the future.

Optical Instruments
Oct. 25, 2020, Vol. 42 Issue 5 20 (2020)
Silicon doping impact on magnetotransport in InAs/GaSb type-II superlattices
Zhichao YING, Zhiyong SONG, Aiying CHEN, Tie LIN, and Shixiong KANG

Aiming at the characteristics of infrared detector fabrication and improving the physical properties of the materials, the magnetic transport properties of silicon-doped InAs/GaSb type II superlattice films (grown by MOCVD) at temperatures ranging from 12 K to 300 K was researched. The principle of the Hall effect was used to calculate the mobility and carrier concentration of the sample at various temperatures. Electrical measurements were performed using the Vanderberg method. At low temperatures, weak localization (WL) of the thin film is observed, and it is found that the doping of silicon gave it better WL stability. The three-dimensional Kawabata model is used to fit the weak local effects to obtain the value of phase coherence length. It explains the advantage of doped n-type silicon for the quantum localization of InAs/GaSb two types of superlattices, which provides a useful reference for the development of infrared detection devices.

Optical Instruments
Oct. 25, 2020, Vol. 42 Issue 5 12 (2020)
Design of mode conversion between terahertz dielectric waveguide and metal waveguide

To realize the mode conversion between the terahertz(THz) dielectric waveguide and the metal waveguide, we design a low-loss, high coupling efficiency tapered waveguide coupler. We have simulated and optimized the transmission efficiency of this structure by the finite-difference time-domain(FDTD) method, and the transmission loss of the fundamental mode is extremely low when the frequency is 170-220 GHz, and TM fundamental mode near lossless transmission at 190-200 GHz. This functional device can be used for coupling of integrated THz chips.

Optical Instruments
Oct. 25, 2020, Vol. 42 Issue 5 1 (2020)
Efficient determination of water content in potato leaves based on spectroscopy technology
Xufeng YU, Hongmei LI, Wei ZHUO, and Jie FENG

The determination of moisture content in potato leaves using spectral technique was studied in this paper. Spectral signatures of one hundred and ten fresh potato leaves in the wavelengths of 900-2100 nm were acquired by the spectral device. Then, the moisture content was measured by the drying method.The near-infrared reflection spectrum information was corrected by the Savitzky-Golay (SG) smoothing, multiplicative scatter correction (MSC) and standard normal variable (SNV) correction. The quantitative relationship between spectral information and moisture was built by partial least squares regression (PLSR) and BP neural network respectively. The effective wavelength was identified by regression coefficients (RC) and corrected by three pretreatment methods. Then the PLSR and BP neural network models were built respectively. The results showed that for full wavelengths-based models, MSC-BP model performed the best with the coefficient of determination (R2) of 0.9791 and the root mean square error (RMSE) of 0.3723 in the prediction. For selected wavelengths-based models, it was the SG-BP model that obtained the optimal result. The R2 value was 0.965 8 and the RMSE value was 0.475 9 in the prediction. This experiment verified that the prediction results of the model established by combining the characteristic band with BP neural network were not different from those of the model established by the whole band, so it could greatly reduce the computation and improve the efficiency.

Optical Instruments
Sep. 23, 2020, Vol. 42 Issue 4 7 (2020)
Improvement of brightness uniformity of out-coupling grating in holographic waveguide
Wanqiu WANG, Xuechang REN, Qinghong LU, and Kaihang LIU

At present, there is a problem of uneven brightness of the out-coupling grating of the holographic waveguide display system in the field of AR. In this paper, a holographic grating with a diffractive efficiency gradient is designed and fabricated to further improve the brightness uniformity of out-coupling grating in the display optical path. Based on theoretical analysis and discussion, this paper proposes a scheme of moving the shutter to realize the change of the exposure time of the holographic plate by divisional region, so that the diffraction efficiency of the obtained holographic grating has a gradual characteristicThe final experimental results are studied and analyzed. The results show that the uniformity of the exit pupil brightness of the coupled grating in the system increases from 28.57% to 57.14%. Therefore, the photographic optical path and exposure method of the holographic optical element used in this paper can effectively improve the uniformity of the exit pupil brightness of the holographic waveguide coupling out grating.

Optical Instruments
Sep. 23, 2020, Vol. 42 Issue 4 14 (2020)
Supercontinuum generation technique for ultrafast laser filamentation in nanoparticle-doped water
Yuan NIE, Shuai YUAN, Hui XU, and Yingsheng DU

It is well known that femtosecond laser filamentation can be initiated by ultrashort laser pulses in condensed media. The supercontinuum generation was studied with the generated white light ranging from 400 to 950 nm in nanoparticle-doped water. We summarized the motion of the generated bubbles. The velocity of the bubble reached 0.16 m/s, and it means that the water flow can also reach the corresponding speed. This could be a potential technique for microfluidic chip fabrications.

Optical Instruments
Sep. 23, 2020, Vol. 42 Issue 4 1 (2020)
Analysis of influence factors on long-term stability of optical coherence tomography
Yongkang WANG, Kang XU, Gang ZHENG, Cheng WANG, and Dawei ZHANG

In view of the current situation that the optical coherence tomography (OCT) scanner has poor stability in clinical use, is easy to cause signal attenuation, and has short maintenance cycle, this paper collects clinical signal attenuation fault information by sampling from the mainstream 4-generation frequency-domain optical coherence tomography equipment in the market. Combining with the principle of OCT, we analyze the influence of internal components of OCT on stability by statistical principle, and the weight of each influence factor on the stability of OCT through analytic hierarchy process (AHP). According to the analysis results, the main factor affecting the long-term stability of OCT is that the spectrometer is locally heated to cause mechanical deformation, which leads to signal attenuation. At last, according to the analytic hierarchy process, we briefly discuss the methods to improve the long-term stability of OCT.

Optical Instruments
May. 19, 2020, Vol. 41 Issue 6 32 (2019)
Comparative study on spatial and spectral of hyperspectral potato leaf late blight
Xinye WANG, Jie FENG, and Xinting LI

In order to detect potato late blight quickly and compare the difference of spatial spectrum information, hyperspectral imaging technology was used to compare the spatial spectrum of potato late blight in order to find the best discriminant method. A hyperspectral camera was used to collect the hyperspectral images of 0?6 days of disease infection. At the same time, the hyperspectral data of typical late blight diseases on the 6th day were selected as the research object. Second-order derivative combined with principal component analysis and second-order principal component analysis were used to extract features from spectral and spatial aspects respectively. Then, K-nearest neighbor classification algorithm, BP neural network and decision tree algorithm were established based on the reflectance of characteristic band and the gray value of principal component image to identify diseases in different periods. The recognition rate of the model was 96.6% based on the gray value of the secondary principal component image and BP neural network. The experimental results showed that the model based on the gray value of the secondary principal component image and BP neural network had good effect on the identification of potato late blight. The three models based on the gray value of the principal component image reduce the redundancy of the band and improve the recognition rate, which provides a reference for researching and developing real-time on-line testing equipment and instruments..

Optical Instruments
May. 19, 2020, Vol. 41 Issue 6 26 (2019)
The study on the automatic localization of the seed point in ultrasound breast tumor images
Huaiyu FAN, Junshan MA, Yutang LIU, and Caihong DU

Segmentation of ultrasound breast tumors using seed region growth is a common method of computer-aided diagnosis. In order to realize automatic and fast locating of seed points, the demand of real-time online image segmentation is satisfied. According to the structural characteristics of ultrasound breast tumor image, and integrating the gray and spatial factors of the image, an algorithm based on iterative quadtree decomposition is proposed to transform the image splitting which meets a specific threshold into searching for seed regions, so as to realize automatic seed location. The verification results of 105 ultrasound breast tumor images show that this method not only satisfies the requirement of automatic location of the seed point within the tumor, with an accuracy of 94.28%. The average time consumed is 2.97 s. This method is also requires few parameters to be adjusted. Compared with other methods, the proposed method is more efficient.

Optical Instruments
May. 19, 2020, Vol. 41 Issue 6 20 (2019)
Terahertz time-domain spectrometer withmulti-modules and its application
Shichang LI, Jiayu ZHAO, Yan PENG, and Yiming ZHU

A multi-modules all-fiber terahertz spectrum system has been designed and built to meet the requirement of accurate identification for the different forms of flammable and explosive dangerous articles in the anti-terrorism goods detection. After that, two different forms of flammable and explosive dangerous goods have been detected to test the system. The results of the test clearly showed that all of the dangerous goods detected by the terahertz spectrum system have obvious characteristic absorption peaks in the range of 0.1~2 THz. On the basis of the detecting results which have been mentioned above, the flammable and explosive dangerous goods may be identified. What's more, it also offers a possible way for public safety inspection.

Optical Instruments
May. 19, 2020, Vol. 41 Issue 6 14 (2019)
Cross-center detection based on deep learning
Huamin WU, Moyu YANG, Xiaoxue HUANG, Caiquan JI, Weijie WANG, Rongfu ZHANG, and Nan CHEN

Crossline center detection is an important part of reflective method for measuring lens center deviation. The detection accuracy of the cross center determines the measurement accuracy of the lens center-offset to some extent. Aiming at the image with irregular edge, poor contrast and low signal-to-noise ratio, a cross-line center detection algorithm based on depth convolution neural network is proposed. The idea of the algorithm is that the convolution neural network can solve the problem that the traditional algorithm is limited to extracting the line and corner features of the edge of the cross image to a certain extent, and realize the recognition and location of the overall features of the cross image. This can relatively reduce the impact of image noise on the location of the cross image center, so as to achieve the accurate location of the cross image center in the case of poor image quality. The experimental results show that the proposed algorithm can get the center of the cross line accurately under the conditions of irregular edges, poor contrast and low signal-to-noise ratio.

Optical Instruments
May. 19, 2020, Vol. 41 Issue 5 38 (2019)
Please enter the answer below before you can view the full text.
6+4=
Submit