Optical antennas play the role of transmitting and receiving coupled optical signals in optical wireless communication, and a pair of multipoint transmitting antennas can expand optical wireless communication applications. Summarize the development of one-to-multipoint optical antennas at home and abroad, and introduces the basic principles of optical fiber array structure, three-concentric sphere structure and rotating paraboloid one-to-multipoint optical antenna. Finally, the development prospects of this field are foreseen. Optical wireless communication is an emerging communication technology with advantages of high speed, large capacity and good security. However, the traditional optical wireless communication system has the limitation of single-to-single communication mode, which can’t satisfy the demand of multi-user simultaneous communication. Optical antenna plays the role of transmitting and receiving coupled optical signals in optical wireless communication, and a pair of multipoint transmitting antennas can expand the applications of optical wireless communication. The development of one-to-multipoint optical antennas at home and abroad is summarized, and the basic principle of one-to-multipoint optical antennas is introduced. Several existing one-to-multipoint optical antennas design schemes are discussed. Finally, the development trend of this field and the future development direction are prospected.

Terahertz (THz) technology, as a very important interdisciplinary field, has brought profound impacts to fields such as environmental detection, biomedicine, and space communication. However, due to the impact of THz waveguide transmission performance, the comprehensive development of THz technology is still limited. Photonic Crystal Fiber (PCF), as a new type of THz waveguide, has infinite single mode characteristics, flexible and adjustable dispersion, controllable nonlinearity, high birefringence, and other characteristics, greatly promoting the development of THz technology. This article first summarizes the development history, light guiding mechanism, and optical characteristics of Solid Core PCF and Hollow Core PCF based on the structural characteristics of PCF; Secondly, the problems and corresponding solutions of PCF used for THz waveguides were summarized; Then, the application of PCF in THz time-domain spectroscopy, THz transmission and communication, and THz fiber coupled sensing was emphasized; Finally, a summary and outlook were made on the application of PCF in THz technology.

In the field of geomagnetic measurement, self-exciting dual-cell optical pumping magnetometers have become an important technology for their fast response and high accuracy. Among them, self-exciting dual-cell optical pumping magnetometers excludes phase-shifting circuits, and has simple circuit. The laser frequency stabilization method included a laser temperature control circuit, a feedback circuit, and a laser current source modulation and control circuit. The study applied the frequency stabilization method of injection current modulation to the dual-cell optical pump magnetometer structure. The proposed method reduces the power consumption and volume, with the temperature control accuracy of ±0.002 ℃ and the wavelength stability of 4.36×10-10 within 10 seconds. The research explores the possibility of integrating and miniaturizing self-exciting dual-cell optical pumping magnetometers.

As the problem of climate change is becoming more and more prominent, President Xi Jinping has made the important instruction of carbon peak and carbon neutral, so the research on carbon dioxide detection is of great significance. For the need of simultaneous detection of carbon dioxide and water vapor, this paper designs a frequency division multiplexing TDLAS laser drive system. The system adopts STM32H7A3 as the main controller, and uses its internal advanced timer and multi-channel DAC to generate multiple amplitude-frequency-adjustable voltage signals, which are matched with the external V-I conversion circuit and fiber optic combiner to develop the frequency division multiplexing TDLAS laser driving system. The system is equipped with a FLASH memory chip and corresponding host computer software, which has the functions of flexibly setting system parameters, storing and loading control data, and protecting the data from being lost in case of power failure. After testing, the output current signal waveform of the system is stable with low frequency error, which can be applied to TDLAS system to drive multiple lasers at the same time.

In the industrial production of metallurgy, petrochemical, chemical and other fields, a large amount of flammable, explosive, toxic and harmful multi-component hazardous gases (such as CO, CO2, SO2, NO, NH3, etc.) are generated, which can cause accidents. However, the existing online monitoring and warning systems cannot meet the requirements of high integration, low interference and high sensitivity. This paper designs corresponding circuits and algorithms for the factors that affect the detection results, such as weak signal, light source modulation, gas signal cross interference, temperature control, etc., and conducts experimental verification. The experimental results show that the weak signal extraction technology based on the lock-in amplifier circuit, the PWM wave modulation technology based on the MOS tube driving circuit, and the concentration inversion model based on the fusion algorithm proposed in this paper can realize the online monitoring of five kinds of hazardous gases (CO, CO2, SO2, NO, NH3), with a detection limit of 0.01%, avoid the cross interference between various gases, and are suitable for the online analysis and warning system of hazardous gases in industrial production, which have a broad application prospect.

To achieve a uniform lighting environment suitable for cockpit man-machine training in the target area (4 m×6 m), a design method utilizing area splice of a high-power LED array was proposed. In order to address the issue of uneven illuminance distribution in traditional rectangular and circular arrays, the arrangement was optimized through the employment of an improved array distribution and an enhanced particle swarm optimization algorithm. A luminance uniformity of 91.17% was exhibited by the improved single-module array based on simulation analysis. Due to the influence of installation errors, the actual test light uniformity was measured at 89%. After the optimization of the multi-module LED array through splicing, an illumination uniformity of 85.83% and a light energy utilization of 87.40% were indicated by the simulation results. The improved multi-module array, designed using the area splicing method, is indicated to be more suitable for simulating large area lighting environments. The characteristics of high illuminance uniformity and a high light energy utilization rate are possessed by the multi-module array. These findings hold significant reference value for practical applications.

This paper introduces the design of a multi wavelength semiconductor laser driving system, which adopts a hybrid driving method of voltage controlled current source and PicoLAS module. The system has the advantages of high efficiency, small size, simple and reliable circuit, easy to use for system integration, and provides more choices and references for the driving design of semiconductor lasers.

This paper proposes a method combining segmentation and Gaussian fitting based on the analysis of image features to extract the center of marker under the perspective of infrared camera. Firstly, the ROI rough positioning of the marker is carried out based on the gray difference of the marker in the image, and the influence of background reflection on the segmentation is eliminated. Then, the K-means clustering algorithm is used to segment the target image. Finally, the improved Gaussian fitting extraction method is used to obtain the center coordinate of the marker spot by removing the central redundant information. The experimental results show that compared with the centroid method, Hough transform method, and circular fitting method, the algorithm in this paper has a stability error of less than 0.1 pixels for center extraction, with high repeatability and stable accuracy. Even with different degrees of occlusion, it still maintains high accuracy, which is suitable for the center extraction of marker in the positioning system under the perspective of infrared camera.

Laser induced damage to optical thin film components is a bottleneck that limits the development of lasers towards high-power and high-energy. Therefore, rapid detection of optical thin film damage has become an urgent problem to be solved. To improve the accuracy and efficiency of damage identification and classification for optical thin film components, a deep learning based damage image classification model training method is proposed. Collect images of laser irradiated oxide thin film damage, extract feature information such as RGB values, grayscale, texture, shape, etc. of the damaged area through preprocessing such as noise removal and image enhancement, and input BP neural network training for recognition. Due to the limited number of datasets and computational errors, the classification results did not meet the expected values, Therefore, transfer learning was used to train the data set. The results showed that transfer learning was better than BP neural network in terms of accuracy and sensitivity, with an accuracy rate of 90%, The depth transfer learning technology is applied to the damage identification of optical thin film components, which provides a new idea to solve the laser induced damage identification of optical thin films.

Aiming at the shape measurement requirements of nano-step objects, a partially coherent optical offaxis image plane interferometry system is designed and constructed. Phase unwrapping, as a key step of interferometry, requires fast speed, high precision and strong adaptability. In this paper, the characteristics of five common phase unwrapping algorithms are analyzed, and a fusion algorithm based on region segmentation is proposed based on the morphology characteristics of nanosteps. This algorithm solves the quality of each segment by combining the light intensity map with the tangent line, and uses the powerless least square method and the tangent method to phase expansion in the high and low quality regions respectively, so as to meet the requirements of real-time measurement and accuracy. The experimental results show that, compared with other algorithms, the proposed algorithm has high precision, fast solving speed, anti-interference to residual handicap, and has certain application value to the measurement of nanometer step topography under partial coherent light interference.

Aiming at the problems of complex algorithm structure and large number of parameters existing in road lane visual detection technology under all-weather conditions, a lane detection method based on depth-separable convolution and residual attention module is proposed, and the LPINet network model is established. We use depth-separable convolution to reduce the size of the input images, design three bottleneck residual units with different structures to reduce the number of network parameters, and introduce the ECANet attention mechanism, which can increase the weight of important feature channels, to improve the lane detection accuracy. The experimental results on Tusimple dataset and GZUCDS self-built dataset show that the LPINet network lane detection accuracy can reach 96.62% in sunny scenarios, and the number of model parameters is reduced to 1.64 MB, which realizes the lightweight design. We carried out exploratory researches in complex scenes such as foggy, rainy, night and tunnel, and the accuracy of lane detection reaches 93.86%, which proves the effectiveness of our method.

Aiming at the problems of low detection accuracy and insufficient generalization ability caused by inconspicuous color features and irregular size in crack images collected by optical sensors in road inspection systems, it is propose improved YOLOv5s crack detection algorithm. The Global Attention Mechanism (GAM) fused with Depthwise Separable Convolution (DSC) is introduced into the backbone feature extraction network to obtain rich cross-dimensional features while reducing the complexity of attention. The recognition ability of cracks is enhanced; the spatial pyramid soft pooling network (Spatial Pyramid Softpool, SPSF) is used to preserve multi-dimensional semantics through Softpool pooling to reduce information dispersion and improve the accuracy of bounding box regression; in the neck feature enhancement network, Downsampling is performed with Atrous Depth Separable Convolution (Atrous DSC), which enhances the aggregation ability of deep and shallow information by expanding the receptive field, and improves the generalization of crack identification. After experiments on the self-made road crack data set, compared with YOLOv5s, the mAP of improved algorithm is increased by 2.2%, which effectively improves the accuracy of road crack detection and the generalization ability of crack recognition under different backgrounds.

Aiming at the shortcomings of traditional PCB board inspection methods, such as low detection efficiency and low detection accuracy, a PCB board defect detection method with improved YOLOv5 model was proposed. In order to improve the precision of small target defect detection, the BiFPN based network connection method is constructed, which makes full use of the feature information of different scales. In order to better capture the position of target defects, we introduced Coordinate Attention mechanism to make model positioning and target capture more accurate. The experimental results show that compared with the original YOLOv5 model, the mean average accuracy of the proposed method for detecting PCB surface defects is improved by 3.2%.

When a mobile robot performs tasks in an unknown environment, it needs to accurately perceive the map environment and other object positions to determine its own position. In order to improve the global positioning accuracy of mobile robots, a binocular vision global positioning method for mobile robots under laser radar SLAM is proposed. Based on the distance measurement principle of the binocular vision system, any camera in the binocular camera is selected to establish the main coordinate system, and a binocular camera model is constructed. The parameters in the binocular camera are calibrated by combining the chessboard diagram and Zhengyou Zhang calibration method. On this basis, the scale of the feature points in a single frame image is calculated, and the binocular vision initialization is completed. The weight function is used to calculate the weighted value and distance value between any feature point and line or face, estimate the position of the mobile robot and introduce a non iterative distortion compensation method to achieve global positioning of the mobile robot. The test results show that the minimum error between the positioning results of the proposed method and the actual motion position is 0.1 m. This method can accurately locate the position of the mobile robot and guide it to better perform tasks.

The theory of compressed sensing shows that the original signal can be recovered by low sampling rate, so it is often used in the field of optical imaging. In order to solve the problem of large amount of data and heavy computational burden when compressed sensing is used to reconstruct images, a blocking compressed sensing method is proposed. In this paper, we propose the compressed sensing methods of column block and mixed blocking. The block by column mode reduces the requirements of block division, and the mixed block mode effectively improve the effect of compressed sensing. Through simulation experiments, it can be verified that the method proposed in this paper effectively improves the quality of image reconstruction, especially the mixed block method, which significantly improves the speed and quality of image reconstruction.

Aiming at the problems of high complexity and poor feature extraction and presentation performance of ESRGAN model, a super-resolution reconstruction algorithm based on Light weight Generative Adversarial Network (LwGAN) is proposed. The Improved Residual Dense Block (IRDB) is used as the base block to construct the high order feature extraction part of the generated network, extract rich and diversified features, and establish the feature channel and long-distance location relationship. In addition to reducing the number of model parameters, the feature extraction and presentation performance of the model are improved. The experimental results on UC MERCED and NWPU-RESISC45 datasets show that compared with ESRGAN, LwGAN obtains larger peak signal-to-noise ratio and structural similarity, significantly improves the performance of super-resolution reconstruction of remote sensing images, and the visualization results show that the reconstructed images recover more texture detail information, while the number of model parameters is only about one-third of that of the original ESRGAN, which significantly improves the operation efficiency of the model and lays the foundation for subsequent analysis and processing of remote sensing images.

To address the problems of single extracted feature information and missing image details in the image super-resolution reconstruction process, this paper proposes a new generative adversarial network (DAMFA-GAN) to obtain more realistic and natural reconstructed images. In terms of generator, a Dynamic attention-Multi-scale feature aggregation (DAMFA) incorporating a dynamic attention mechanism is used to obtain multi-scale high-frequency information of each upsampled feature in low-resolution images to improve the quality of the reconstructed images; in terms of discriminator, the ConvTrans Encoder module is designed to enhance the feature information extraction capability to improve the accuracy of discrimination. Experimental results on the Set5, Set14, BSD100 and Urban100 datasets showed that DAMFA-GAN improved the peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) by an average of 0.50 dB and 0.015 2 respectively compared to SRGAN. At the same time, the high-frequency details and visual effects of super-resolution reconstructed images are also significantly improved.

In view of the inherent coherent speckle noise in Synthetic-aperture radar (SAR) images, which affects the accuracy and accuracy of change detection, this paper proposes a change detection method for SAR images based on difference map construction and fusion. This method preprocesses SAR images through L-SRAD hybrid filtering, uses wavelet fusion algorithm based on edge pre-detection to achieve the fusion of logarithmic hyperbolic cosine ratio difference map DCLR and neighborhood ratio difference map DNR, and combines FCM algorithm and CWNN Convolutional neural network to detect changes in the fusion difference map. The FCM algorithm pre-classifies the fused difference map into three clusters, selects appropriate pre-classification results as training samples to train the CWNN model, and finally uses the CWNN model to perform secondary classification on the pre-classification results to obtain the final change detection map. Comparative experiments were conducted on the Bern dataset, and the experimental results showed that this method has strong change detection ability, with a change detection accuracy of 99.67%.

A low brightness laser image detail information enhancement method based on visual communication is proposed to address the issues of low brightness laser images, resulting in missing internal detail information and decreased image applicability. From the perspective of visual communication, the low illumination laser image is converted into Lab color mode. The laser image is decomposed into high-frequency and low-frequency components using the Curvet transform. The high-frequency components are enhanced using a detail enhancement network model. The low-frequency components are enhanced using a low light level image enhancement method based on illuminance map estimation, and the enhanced high-frequency and low-frequency components are fused. The experimental results show that, After using this method to enhance the detail information of the selected images, the information entropy of each image is higher than 8.35, and the contrast and correlation coefficients are higher than 0.846 and 0.815, respectively. The enhanced images are more consistent with human visual characteristics.

A multi-level fusion method for laser images based on visual communication technology was designed to achieve outstanding visual communication effects. Firstly, the improved single scale Retinex algorithm is used to extract the reflection image of the original laser image, and the multi-scale color image obtained through reconstruction using the Gaussian Laplace algorithm is used to enhance the original laser image. Then, a deep stacked convolutional neural network is used to obtain high and low frequency images, and high-frequency images are fused based on the maximum local variance. Low frequency images are fused based on the comparison between matching degree and threshold, The final experimental results show that when the number of stacked CNN is 4, the visual communication effect of the fused laser image is the best. The enhanced laser image has rich local detail information and good color fullness, and the maximum grayscale frequency of the fused image is only 0.015.

The echo characteristics of partially coherent Gaussian Schell-model vortex beams are studied by using the generalized Huygens-Fresnel principle and random Gaussian rough surface model. The cross-spectral density function and intensity analytical formula of partially coherent vortex beams scattered by random rough surface are derived by special integral, and the simulation analysis is carried out. The results show that the intensity autocorrelation function curve decreases more and more steeply with the increase of topological charge, beam waist radius and coherence length. In strong turbulence environment, the attenuation of autocorrelation function is more and more serious. The inner scale will affect the autocorrelation function of light intensity, and it will decay seriously with the increase of the inner scale, while the outer scale has little effect on it. As the coherence length of the rough target increases, the half-peak width of the intensity autocorrelation function gradually increases.

The filtering process of spaceborne LiDAR data is susceptible to interference from complex backgrounds, gross errors, noise points, and other issues, resulting in a significant decrease in filtering effectiveness. Therefore, a multi-scale filtering method for spaceborne LiDAR data based on improved DBSCAN is studied. The improved DBSCAN algorithm is used to cluster spaceborne LiDAR data, label noise points, and extract point cloud data features using a hemispherical neighborhood algorithm. Based on the extracted point cloud data features, a regular grid is constructed, and the coarse points and noise points in the point cloud data are removed through the multipath effect of the grid, completing multi-scale filtering of spaceborne LiDAR data. The experimental results show that the proposed method has low multi-scale filtering error and good filtering effect for spaceborne LiDAR data, and has high practical application value.

In order to improve the security of wireless communication system transmission, a scheme of Reconfigure Intelligent Surface (RIS) assisted Generalized Spatial Modulation (GSM) is proposed to realize Physical Layer Security (PLS) transmission. Firstly, after the signal is transmitted from the transmitter, the RIS reflection unit is grouped, and then different RIS reflection unit groups are activated by GSM to reflect the signal to the receiver. At the same time, artificial noise (AN) is added to the transmitted signal to expand the channel difference between legitimate users and eavesdroppers. Reduce information leakage and ensure transmission security. In addition, this scheme realizes GSM transmission of additional index bits through RIS cooperation, which can effectively improve the transmission efficiency of the system. Finally, the theoretical analysis and experimental simulation of the system bit error rate and secrecy rate are carried out. The results show that the scheme can not only reduce the bit error rate of the legitimate receiver, but also effectively improve the security of the system transmission. Compared with traditional schemes, when the bit error rate is negative cubic of 10, the SNR has a performance gain of about 2.2 dB. When the SNR is 20 dB, the average secrecy rate is increased by about 200%.

To further improve the accuracy of flower classification, a new network model was proposed based on bilinear convolutional neural network, RepVGG and attention mechanism. Firstly, RepVGG network was used to replace the original feature extraction network VGG to improve the ability to extract the main features of flowers. Then, channel attention and spatial attention mechanisms were introduced into the two RepVGG networks respectively, and the high-dimensional bilinear features generated by the cross-product of the two RepVGG networks were used to extract the fine-grained features of flowers. Finally, the RepVGG layers are transformed into single-way structures by structure reparameterization to improve the speed of model reasoning. Experimental results show that on the enhanced Oxford-102 data set, the inference speed and classification accuracy of the new model are greatly improved compared with the original model and the common model, and the classification accuracy is also improved compared with that before the introduction of attention.

In time-varying environments, traditional models only remove some communication crosstalk and have high transmission losses. Therefore, a machine learning channel modeling method for free space optical communication systems in time-varying environments is proposed. The histogram statistical method is used to calculate the peak values of the dataset, select the shortest communication path, obtain the optimal channel parameters through genetic algorithm, calculate the communication impedance based on the optimal parameters through machine learning, obtain capacitance and conductivity values, adjust the weights to remove communication crosstalk, convert the antenna domain into beam domain based on the channel size characteristics, and build a free space optical communication channel in a time-varying environment. The simulation experimental results show that the average transmission loss of the model in this paper reaches 148 dB, which is relatively low and has high application value.

Wireless ultraviolet communication system is vulnerable to atmospheric scattering and other factors, resulting in low accuracy of communication channel capacity estimation. To this end, a channel capacity estimation method for wireless ultraviolet multiple-input multiple-output (MIMO) communication system is proposed. According to the channel characteristics, the channel capacity estimation model of the wireless ultraviolet MIMO communication system is constructed, the single scattering transmission characteristics of wireless ultraviolet light, the space time coding is improved by using the symbol complement method, and the received energy consumption of the symbols in the advanced detector is set, so that each symbol is kept within the standard power value range, and the channel detection coding of the communication system is completed. The path loss ratio of wireless UV communication is calculated according to different transmission and reception angles, and the final channel capacity of the communication system is estimated by Poisson random distribution. The experimental results show that the transmission rate of the proposed method can be as high as 8 Mbps, the bit error rate is always less than 10-3 bit/s, the estimation accuracy is more than 90%, the estimation time is less than 4 ms, and the estimation effect is good, which can provide a theoretical basis for the development of the communication industry.

Optical communication systems are susceptible to various factors. Traditional methods for detecting abnormal data in optical communication systems have high error rates and detection efficiency. In order to obtain ideal abnormal data detection results in optical communication systems, a clustering analysis based feature extraction method for abnormal data detection in optical communication systems was designed. Firstly, a data transmission model for optical communication systems is designed, and clustering algorithms are used to extract abnormal data features. Then, a degree learning algorithm is used to establish an abnormal data detection model for optical communication systems, and genetic algorithms are used to optimize deep learning algorithms. Finally, a simulation experiment for abnormal data detection in optical communication systems is conducted, and the results show that the accuracy of the proposed method for detecting abnormal data in optical communication systems exceeds 98%, The detection time of abnormal data in the optical communication system is 21.6 ms, which has certain practical application value.

In order to accurately monitor the status of abnormal nodes in massive IoT data and improve the security of wireless networks, a method for monitoring the status of abnormal nodes in IoT using laser infrared technology is proposed. The laser infrared positioning method based on angle measurement is used to determine the approximate range area of abnormal nodes of the Internet of Things. This area is used as the collection area to carry out abnormal node feature data collection and dimension reduction. The collected abnormal node data is denoised through singular value decomposition, and the abnormal status of nodes is judged by combining random matrix theory and average spectral radius, so as to realize the monitoring of abnormal node status of the Internet of Things. The experimental results show that the proposed method has a maximum positioning error value of only 4.3%, a monitoring time of less than 79.85ms, and a packet loss rate of less than 0.22%, indicating good monitoring capability.

Industrial robots are widely used in many fields, and offline programming has become an important development direction. Due to the high requirement of the absolute positioning accuracy of the robot itself for off-line programming, in order to improve the absolute positioning accuracy, a gait kinematics parameter identification method of industrial robot guided by laser vision is designed. Based on the linear structure laser measurement technology, a laser vision sensor composed of industrial intelligent camera, photosensitive element, filter and line laser is designed. The sensor is used to measure the industrial robot to obtain its three-dimensional information. The laser vision sensor is guided by laser vision through two steps of rough guidance and fine guidance to improve efficiency and reduce aiming error. The identification model of gait kinematics parameter error of industrial robot is solved iteratively by the least square method to obtain the accurate solution of the equations, thus completing the identification of gait kinematics parameters. The test results show that the gait kinematics parameter identification error of this method is less than 0.02 mm, the solution can be completed within 30 iterations.

In order to solve the problems of high energy consumption and low scheduling efficiency of fiber optic communication transmission network nodes in the Internet of Things environment, a scheduling method for complex fiber optic communication transmission network nodes in the Internet of Things environment is proposed. In the context of the Internet of Things, analyze the distribution of nodes in fiber optic communication transmission network nodes, and based on dynamic alliance theory, construct a complex fiber optic communication transmission network node scheduling model with the goal of achieving the fastest scheduling speed and minimum investment. Through a fusion algorithm combining ant colony and particle swarm optimization, all searches are carried out to obtain the optimal scheduling scheme and achieve node scheduling in fiber optic communication transmission networks. The experimental results show that the average energy consumption of the proposed method’s nodes is below 6.0 mJ, the LBF value is always above 0.20, and the node scheduling time is always less than 200 ms, showing good scheduling performance.

In this paper, a new periodic metasurface has been devised based on the sandwich-type nanopillar structure of dielectric and absorber layers, which exhibits strong resonant absorption effect, and achieves the advantages of wide gamut, high saturation and high resolution in the visible light band. The finite difference time domain (FDTD) method is used to simulate the optical response of periodic nanoarrays when the period and the thickness of the dielectric layer are adjusted, and the mapping relationship from structural parameters to color characteristics is established based on the reflection spectrum. The results show that the synergistic adjustment of the period and the thickness of the dielectric layer generates more abundant structural colors. The period significantly affects the hue by changing the main wavelength, and the thickness of the dielectric layer helps to achieve high monochromatic color by optimizing spectral shape; the structure has high spectral reflectance, narrow half-peak width, and wide main coverage of the main wavelength, and the color gamut is expanded to 156.8% of sRGB; the optimized high-quality RGB color has a hue change of less than 0.075π at ±40° incidence. The study is of great value in display imaging, nano-printing, high-resolution printing and other fields.

With the advancement of technology, traditional surface-mounted fiber optic gratings can no longer meet the requirements for monitoring wind turbine systems. Therefore, this paper proposes a method for monitoring the damage of carbon fiber composite materials in wind turbine blades using an embedded Fiber Bragg Grating (FBG) sensing network. Firstly, a three-dimensional model of the carbon fiber composite material test specimen is created, and finite element simulation is carried out to determine the placement of FBGs. Subsequently, FBGs are embedded into the test specimens and mechanical performance tests are conducted to monitor the strain values and related damage at each sensing point. Finally, an ultrasonic non-destructive testing is performed to verify the test specimens. Experimental results demonstrate that embedded FBGs can dynamically monitor the internal damage of carbon fiber composite materials in real-time, thus accumulating experience for structural health monitoring of wind turbine systems.

Retinal blood vessels are small and complex. When segmentating retinal blood vessels, noise, fracture and undersegmentation often occur. To solve this problem, a lightweight network named LRU-Net based on local feature enhancement is proposed to capture more features of small blood vessels. Firstly, a feature extraction module is added to the channel attention module to extract secondary features from input features so as to obtain more detailed features. Secondly, a feature fusion module is designed, which can fuse the high and low features more effectively in the decoder, and strengthen the final feature representation. Finally, a context aggregation module is designed to extract multi-scale information with different resolutions of the deepest features, and then splicing it to make the input features into the upper sampling more detailed. Experimental results on FIVES and OCTA-500 data sets show that compared with U-Net, the proposed method not only achieves lightweight, but also improves the accuracy and Dice coefficient of retinal vessel segmentation to a certain extent.

In the case of low resolution packaging images in weak light, the identification degree of packaging recognition and detection is not high, and contrast enhancement processing is required. A method for enhancing the contrast design of packaging images in weak light based on curved wave transformation is proposed. Construct an image spatial feature information detection model, decompose the image using a panchromatic sharpening method at multiple scales, reorganize the panchromatic sharpening features using a curved wave transformation method, and combine the dictionary set sparse representation method to achieve low dimensional subspace representation. Construct a contrast enhancement model for weakly illuminated packaging images based on the detailed representation structure of the subspace. The simulation results show that using this method to enhance the contrast of weak light packaging images improves the level of image recognition and the ability to express detailed features. The recognition error is low, with an average of 0.045, and the peak signal to noise ratio is high, with an average of 42.134 dB, the average image enhancement time is 239.75 ms.

In order to optimize the planning and design of public spaces such as landscapes and indoor spaces, and reduce spatial redundancy costs, a three-dimensional imaging method for indoor spatial layout based on light field reconstruction is proposed. The laser light field image analysis model of the indoor public space landscape layout is constructed by using the global illumination image. According to the lighting conditions and different surface material attributes of the indoor space layout, the three-dimensional optical flow field reconstruction model of the indoor space layout is established. The joint evaluation of the indoor public space landscape layout is conducted through the subjective perception and objective indicators, The method of analyzing the network structure model using mirror reflection rendering is used to decompose the three-dimensional details of indoor spatial layout. Through the reconstruction results of the light field, richer object information is obtained. Combined with the two-photon correlation imaging method, a three-dimensional imaging method for indoor spatial layout is constructed. Tests have shown that this method can effectively achieve the layout of indoor public space landscapes and improve the overall planning level of indoor landscapes. Its spatial redundancy cost is 4.173 m2, which can effectively reduce the spatial redundancy cost.

The mechanical arm components generally have high working strength and are prone to various defects. If they are not found and handled in time, their working quality will be easily affected. Therefore, it is very necessary to detect the defects of mechanical arm components. Under this background, a defect detection method of mechanical arm components based on triangular laser is studied. In this study, the triangular laser method is used to scan the mechanical arm components to obtain the component image and carry out the graying and denoising processing. The corner feature of the image is extracted by USAN method and two texture features are extracted by gray level co-occurrence matrix. The feature parameters are normalized and fuzzed, and the defect type is determined by calculating the proximity degree to complete the defect detection of the mechanical arm components. The results show that under the application of the method, the under-segmentation rate is the smallest, which shows that the research is more effective for the defects of the mechanical arm components, and the detection accuracy of the method is higher.

The shape of digital media images is often irregular, which increases the difficulty of collecting small details and leads to a decrease in the quality of their 3D reconstruction. Therefore, a 3D reconstruction method for digital media images based on laser holographic projection is proposed. Using CCD cameras and laser equipment to collect digital media images, the phase and amplitude of each point on the object light are obtained based on laser holographic projection and diffraction theory, and a laser holographic projection image containing small information is obtained. In view of the speckle noise generated in the projection process, the compound wave plate and Frosted glass are used to modulate the phase of the object light, suppress the coherence of the light source, and remove the noise on the image. The bounding box algorithm is used to obtain the three-dimensional spatial coordinates of the light entering the object, and the Euler Lagrange equation is used to establish a three-dimensional reconstruction data field for the image, achieving the three-dimensional reconstruction of digital media images. The experimental results show that the proposed method can effectively remove noise during the reconstruction process, ensure the clarity of the reconstructed image, reduce the error of 3D reconstruction of digital media images, and shorten the reconstruction time. The maximum reconstruction mean square error of the proposed method does not exceed 0.5, and the reconstruction time is within 10 seconds.

In order to achieve ideal 3D visualization of buildings and provide effective technical means for building measurement, a 3D visualization system of buildings based on laser point cloud data was designed. Firstly, a laser scanner is used to collect building laser point cloud data, which is processed through redundancy removal, distortion compensation, and filtering. Then, the laser point cloud data is transformed into coordinates, and the 3D Max modeling software is used to establish a three-dimensional visualization model of the building. Finally, its performance is analyzed through experiments, and the results show that the system has strong compatibility and laser point cloud data preprocessing capabilities, The accuracy of building 3D reconstruction exceeds 93%, and the time of building 3D reconstruction is controlled within 4 seconds, which has higher efficiency of building 3D reconstruction.

In order to improve the design effect of product anti-counterfeiting packaging, a design method of product anti-counterfeiting packaging based on laser interference lithography was proposed. Multi-beam interference information is controlled by fiber-coupled laser transmission, based on fiber waveguide beam splitting forming technology, integrated processing of beam propagation path, based on beam amplitude, phase and polarization joint modulation method, the anti-counterfeiting state characteristic of product anti-counterfeiting packaging is extracted, and the laser interference lithography model is constructed based on amplitude or phase microstructure diffraction unit array transmission method. Multi-beam laser anti-counterfeiting of four beams, five beams and six beams is realized. The test results show that the lowest tampering rate of the proposed method is 0.05%, the decryption time is short, all about 30 ms, MSE is 0.673%, PSNR is 40.40 dB, and SSIM is 0.975. This method is used to design the product anti-counterfeiting packaging, which improves the laser anti-counterfeiting ability of the product and reduces the tampering rate.

In order to solve the problems of high detection error rate and long detection time in the detection process of complex background infrared dim dim targets, a detection method of complex background infrared dim dim targets based on full convolutional network is proposed. This paper analyzes the research progress of complex background infrared dim small targets detection, finds out the defects of different methods, collects infrared images, extracts target detection features, and uses the full convolutional network to design a classifier for dim small targets detection, and realizes the detection of complex background infrared dim small targets. The experimental results show that the accuracy of this method is more than 97%, and it has high practical application value.

In order to ensure the quality and service life of mechanical parts, it is necessary to complete the machining defect identification of mechanical parts. Therefore, a method of machining defect identification of mechanical parts based on laser scanning imaging is proposed. The image acquisition module mainly takes laser scanning imaging system as the core to complete the imaging of mechanical parts. The difference algorithm and resolution generation adversarial network model are used to reconstruct the acquired images. The kernel principal component analysis method was used to extract the characteristic values of machining defect fringes in laser images, and the extracted results were input into support vector machine to identify machining defects of mechanical parts. The test results show that the method has the highest recognition accuracy of 97%, the lowest recognition error of 1%, and the fastest recognition efficiency of 3s. It can effectively deal with noise points in images, improve image clarity, and realize machining defect recognition of mechanical parts.

UAV lidar point cloud has many features, and the first matching takes a long time, so it is difficult to carry out the second matching. Therefore, the UAV lidar point cloud matching method based on two-dimensional normal distribution is studied. Collect the UAV lidar point cloud image, preprocess the image through the rotation translation method and Bilateral filter method, use the two-dimensional normal distribution algorithm and the dynamic time warping algorithm to complete the point cloud feature extraction, use the initial transformation matrix estimation algorithm to rough match the point cloud, and then use the near point iteration algorithm to fast and fine match the point cloud, and realize the UAV lidar point cloud fast matching through two matching. The experimental results show that the proposed method has good denoising effect on unmanned aerial vehicle LiDAR point cloud images, short point cloud matching time, and a matching deviation of only 0.04 m-0.15 m. The matching accuracy has met the relevant expectations.

The current pulse laser acoustic signal detection methods are highly susceptible to environmental noise, resulting in unsatisfactory detection accuracy. Propose a pulse laser acoustic signal detection method based on spatial transformation technology. Firstly, the wavelet threshold method is used to remove noise from the pulse laser acoustic signal; Secondly, using spherical Fourier space transformation technology to locate the sound source in the acoustic signal and obtain the accurate position of the acoustic signal; Finally, the ratio feature, specific gravity feature, and frequency feature of the acoustic signal are extracted separately, and based on this, SVM algorithm is used to classify and recognize them, in order to achieve effective detection of pulsed laser acoustic signals. The experimental results show that the acoustic signal detection accuracy of the proposed method exceeds 96%.