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.

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.

The method of simulating cone lens with spatial light modulator is the most widely used method to generate perfect vortex beams at present. However, the ring width of the generated beam can not be controlled in real time by this method. Here, we propose an approach to generate and control the perfect vortex beams based on computer-generated phase holography in Fourier space. Through theoretical analysis and experimental measurements, We demonstrate that the method presented in this paper enables real-time control of topological charge, radius, and thickness of the generated ring beam by using a coded phase mask. The method offers simple optical path, high beam quality, and simple operation with only one of optical path ajustment. Furthermore, the proposed method can also be used to generate and control other abnormal perfect vortex beams, such as elliptic perfect vortex beams. The experimental results are in good agreement with the theoretical results. Thus, the proposed method based on phase-only computer-generated hologram presents a versatile and simple way of generating and controlling perfect vortex beams.

ortex beam carries orbital angular momentum and exhibits a helical phase, with a Poynting vector oriented at an angle to the optical axis. The angular component of the vortex beam's Poynting vector causes a rotational Doppler effect, which allows direct measurement of rotational velocity. The linear and rotational Doppler effects of vortex light can be used to detect both linear and rotational velocities of an object, providing a comprehensive view of its compound motion. On this basis, the target is detected using the vortex light superimposed with different orbital angular momentum modes. These different orbital angular momentum modes cause different frequency shifts, which, when combined with the frequency shifts resulting from the linear Doppler effect, allow us to measure the specific velocity components of the compound motion. In this paper, we propose a measurement model for compound motion using superimposed vortex beams and analyze how different motion states affect the detection results.

Spectral processing holds immense significance in the research and application of optics. Various devices and instruments have been developed for specific tasks, including spectral filtering, shaping, analysis, and wavelength demultiplexing. However, advanced multitasking spectral processing equipment has been lacking. In this work, we have designed a diffraction neural network for spectral filtering, comprising a phase-modulated diffraction layer and a detection layer. During the training process, wavelength parameters were incorporated to achieve the processing of broadband signals. By designing a tailored loss function, we gained control over the output spectrum. Taking the broadband signal in the visible light band as an example, we achieved both single and dual passband spectral filtering, with adjustable central band widths and relative intensities. This research demonstrates that optical diffraction neural networks can effectively handle broadband spectra, laying the foundation for tackling more complex spectral processing tasks in the future.

As a major category of plant dyeing, tea dyeing has a deep cultural heritage while having good environmental protection performance. In order to accurately describe the spectral changes of tea staining, this work studied the relationship between the spectral reflectance of rice paper dyed with tea and the tea concentration. First, a spectrophotometer was used to measure the spectral reflectance of rice paper in the 400 to 700 nm band which was stained by tea leaves. Prediction models were constructed by the spectral reflectance of rice paper and tea concentration based on the partial least squares regression model, BP neural network and continuous projection algorithm(SPA) selected feature band, respectively. Then the spectral reflectance was used as an input variable to predict the tea concentration. The results show that the partial least squares method, BP neural network and continuous projection algorithm select characteristic bands to establish a model to predict tea concentration through the spectral reflectance of tea dyed rice paper, which has high robustness and reliability. SPA-BP neural network model has the best performance: the average prediction accuracy rate is 98.40%, the coefficient of determination is 0.9910, and the root mean square error is 0.843 3. This shows that it is feasible to predict tea concentration through spectral data of tea dyed rice paper.

In order to improve the sensitivity of low refraction chemical monitoring, a surface plasmon resonance sensor based on photonic crystal fiber with gold film in open loop was designed in this paper. The effects of open-loop radius, pore size and thickness of metal film on the sensitivity of the sensor were studied systematically by using COMSOL Multiphysics 5.6. Finally, a low refractive index sensor with a refractive index detection range of 1.26-1.31 was designed in the working band of 2 800-4 700 nm. The average sensitivity of the sensor is as high as 22 500 nm/RIU, and the highest sensitivity is 33 000 nm/RIU. It has a good application prospect in the detection of low refractive index substances such as sevoflurane, haloether and fluorine-containing organic compounds.

Denoising has long posed a significant challenge for security cameras, even with the denoising algorithms, on which modern security cameras heavily rely. These algorithms often lead to a degradation of image edges, especially in low illuminance conditions. To tackle this problem, a low illuminance image denoising algorithm is proposed, which aims to minimize the loss of edge information while reducing noise based on pixel value deviations. In light of the human visual system, we construct an image pyramid from the pixel values, allowing us to analyze local and global pixel value differences and mitigate the noisy influence on pixels of medium value. Experimental results show that the proposed algorithm removes the noise signal in nighttime road images, preserves edge details, enhances the overall image quality compared to the classic wavelet denoising algorithm, and improves the signal-to-noise ratio according to the Imatest SNR test.

UV Raman spectroscopy has the characteristics of high intensity Raman scattering and no fluorescence interference; visible Raman spectroscopy has the advantages of low wave number and high resolution detection. In order to combine the advantages of two excitation wavelengths, two symmetrically distributed Czerny-Turner optical paths focusing on one detector are designed for a dual-channel Raman spectrometer. Through the selection of components and calculation of the initial structure, the energy loss on the image plane can be avoided by compensating for the astigmatism in the arc-vector direction without adding extra components. Using the software Zemax, the dual-channel spectra were modeled and optimized separately, and the simultaneous detection of the spectra of 400~5000 cm-1 (266 nm excitation) and 50~3500 cm-1 (633 nm excitation) was finally realized. The rationality and feasibility of the design were verified by the root mean square radius, spot diagram, and modulation transfer function. The results show that the resolution of the two sets of spectrometers can reach 8 cm-1 and 5 cm-1, respectively. The design has the advantages of high resolution, low wave number, multi-wavelength excitation, and integration.

The solar blind ultraviolet optical detector accurately captures the ultraviolet radiation generated by corona discharge. In this paper, a fixed focal solar blind optical system was designed by using Zemax. The system comprises five standard spherical lens and has a working wavelength range of 240-280 nm, the F number of 2.5, the focal distance of 50 mm, a field angle of 20.4°, an optical length of 72 mm and matches with a ultraviolet image intensifier with a target diameter of 18 mm. The full field of modulation transfer function is greater than 0.8 at the spatial frequency of 20 lp/mm, the maximum distortion is less than 0.5％. The defocusing phenomenon of the system was analyzed across a temperature range of -20℃ to 60℃. The thermal defocusing amount was calculated at different temperatures, and a Passive mechanical compensation method was employed to correct thermal dispersion caused by temperature changes. Finally, through Monte Carlo simulation analysis, the system is given a reasonable tolerance allocation. The analysis results show that the system still maintains good imaging quality even after processing and assembly.

To solve the short working distance and small measurement range of all-fiber interferometer, we proposed an all-fiber interferometry system based on sinusoidal phase modulation. An electro-optical modulator is combined in reference arm to generate modulated reference beam, which interferes with the target reflected beam received by the collimator. After extracting orthogonal components from interference signals according to phase-locked amplification principle, an observation model composed of orthogonal components is first established. The amplitude and bias of the orthogonal components are iteratively estimated and corrected by Kalman filtering to reduce the amplitude drift and additional bias caused by phase delay, modulation depth drift, parasitic interference, etc. Then the inverse tangent method is used to demodulate the phase difference between the reflected light and the reference light. The phase solution simulation and displacement measurement experiments of analog interference signals are carried out. Simulation and experiment verify the validity of the proposed method. The experimental results show that the displacement measurement system can achieve a working range of 20 cm and a measurement accuracy of 10 nm.

This paper investigated the microscopic imaging and detection of subsurface metal micro/nano structures using Terahertz scattering-type near-field optical microscopy (THz s-SNOM). For the first time, a self-built THz s-SNOM system was employed to measure the Terahertz near-field of gold micro wires covered with hexagonal boron nitride (h-BN) film on the surface, resulting in near-field microscopy images with nanometer-level spatial resolution and high contrast. Combined with full-wave numerical simulation, the spatial resolution, near-field scattering signal intensity, and imaging contrast of THz s-SNOM for detecting subsurface metal micro/nano structures were analyzed. The study shows that THz s-SNOM has excellent subsurface microscopic imaging and detection capabilities and can be applied to the subsurface structure characterization and defect detection of micro/nano electronic devices.