最新刊期
卷 39 ， 期 10 ， 2024
- “聚氨酯丙烯酸酯基聚合物分散液晶研究，为柔性显示器件提供新方案。”摘要：Polymer dispersed liquid crystal （PDLC） is a composite material formed by uniform dispersion of liquid crystal domains within a polymer matrix. Because of its simple preparation and no need for polarizer and orientation layer， it has a wide range of application prospects in the fields of smart windows， flexible displays and wearable devices. In order to broaden its application scenarios in flexible display devices， this paper uses polyurethane acrylate （PUA） oligomers as the polymer matrix to prepare PDLC devices by ultraviolet photopolymerization. The effect of the molecular structure of PUA on the morphology and properties of polymer dispersed liquid crystals is investigated. The results show that the speed of the polymer monomer polymerization with acrylate groups for capping is fast， forming a polymer mesh-like structure， and the size of the polymer mesh decreases gradually with the change of the molecular structure from aliphatic to aromatic， resulting in an increase in the driving voltage of PDLC. The hydrogen bonding formed between the molecular chains by utilizing the urethane urethane bonding enhances the adhesion with the plastic substrate， thus obtaining PDLC devices with excellent overall performance. Compared with the aromatic PUA， the PDLC samples prepared from aliphatic PUA are characterized by lower driving voltage （Vth=9.14 V， Vsat=35.50 V）， higher contrast （CR=73.25）， better adhesion （20.11 N/cm2）， and smaller changes in the optoelectronic properties before and after bending at a bending radius of 4 cm.关键词：polymer dispersed liquid crystals;polyurethane acrylates;molecular structure;optoelectronic properties;mechanical properties
- 9
- |
- 4
- |
- 0
发布时间：2024-10-09
Material Physics
- “在半导体领域，研究人员通过射频磁控溅射法制备MgZnO薄膜，发现真空退火处理能显著提升器件性能，场效应迁移率增至0.29 cm2·V-1·s-1，阈值电压降至2.28 V，电流开关比高达1.68×10^6，为提高TFT电学稳定性提供新思路。”摘要：To investigate the influence of annealing atmosphere on the performance of MgZnO-TFT， MgZnO thin films were prepared by radio frequency magnetron sputtering and used as the channel layer to construct a bottom-gate top-contact structure MgZnO-TFT device. The MgZnO thin films were subjected to annealing treatment at 500 ℃ for 1 h in four different atmospheres， including air， vacuum， oxygen and nitrogen. Atomic force microscopy （AFM） and X-ray photoelectron spectroscopy （XPS） techniques were used to characterize and analyze the thin films. The results show that the MgZnO thin film quality is better after annealing in vacuum atmosphere， and the device performance is the best with a field-effect mobility of 0.29 cm2·V-1·s-1， a threshold voltage of 2.28 V， a subthreshold swing of 3.6 V·dec-1 and a current switching ratio of 1.68×106. The analysis suggests that this may be due to the fact that annealing in vacuum atmosphere can effectively isolate external interference to a certain extent and avoid the generation of defects in the active layer thin film. At the same time， we studied and tested the stability of positive bias stress （PBS） and negative bias stress （NBS） of the device， and the TFT showed good stability under different gate bias stress conditions. When the positive bias pressure is 10 V and the stress time is 3 000 s， the threshold voltage drift of MgZnO-TFT optimized under vacuum atmosphere decreases from 1.38 V to 0.54 V compared with the ZnO-TFT. The results indicate that doping magnesium element into zinc oxide to prepare MgZnO thin films as the active layer of TFT has a certain degree of improvement on the electrical stability of TFT devices.关键词：MgZnO-TFT;annealing atmosphere;XPS analysis;stability
- 12
- |
- 7
- |
- 0
发布时间：2024-10-09 -
Compound eye camera array system with large field of view， high-resolution and achromatic aberration AI导读
“在视觉成像领域，研究者提出了一种新型复眼相机阵列系统，通过仿生视觉模型和多图像处理技术，实现了大视场、高分辨率、无色差的图像获取，为复杂场景下的视觉系统提供了新方案。”摘要：In order to overcome the limitations of single-camera platform due to the trade-off between high resolution and large field of view， this paper proposed a compound eye camera array system capable of capturing wide field of view， high-resolution and achromatic images. Combined with the theory of human visual imaging， a non-uniform sampling biomimetic vision model was established to reduce data redundancy at the input of system. A smooth weighted fusion algorithm for multi-image was designed to stitch the large field of view image seamlessly. Additionally， a multi-level iterative histogram matching color correction algorithm was proposed to enhance color consistency among different sub-cameras. The overlapping regions between adjacent sub-cameras were utilized to perform multi-image super-resolution reconstruction on ROI （Regions of Interest）. Ultimately， a curved camera array system was built up， enabling the acquisition of large-field of view high-resolution images without color deviation. Experimental results demonstrate that the proposed system achieves a 9-fold increase in both field of view and resolution compared to a single sub-camera， while achieving a 4-fold super-resolution reconstruction of ROI. Additionally， compared to typical color correction algorithms HM， the PSNR and HSM metrics of our proposed method are improved by 4.1% and 15.9% respectively. These results satisfy the demands of multi-camera systems for large field of view， high resolution and color-accurate images in complex scenarios.关键词：camera array;image stitching;color consistency;histogram matching;super-resolution reconstruction- 113
- |
- 31
- |
- 0
发布时间：2024-10-09 - “在高精度装调设备领域，专家设计了基于OpenCV的图像处理算法，通过主动调焦和算法优化，实现了4.54 μrad的图像校准精度，提升了装调精度。”摘要：In order to ensure the successful application of dual-wavelength fiber-optic light in high-precision alignment equipment， an image processing algorithm based on OpenCV （Open-Source Computer Vision Library） was designed. The active adjustment exposure mode is used to actively adjust the focus of the CCD （Charge-coupled Device） camera， OpenCV is used to analyze typical image processing algorithms. Through computer analysis of algorithms， adjustment of parameters， fitting of images and drawing of center coordinates， an ideal image is output. After detection and analysis， the average accuracy of image calibration is 4.54 μrad. This algorithm can capture， identify， and track targets in real time， and print the center coordinates recognized by the computer onto the output image， thus improving the accuracy of the installation and adjustment of the dual-wavelength optical fiber guide light and realizing dual-wavelength optical fiber guidance.关键词：OpenCV;dual-wavelength optical fiber guided light;high-precision alignment;image processing;least square method
- 38
- |
- 26
- |
- 0
发布时间：2024-10-09
Device Physics and Device Preparation
- “在智能抄表领域，研究者通过改进YOLOv5算法，显著提升了机械字轮电表偏转字符的识别准确率，实验显示准确率高达99.4%。”摘要：Due to the high cost and harsh environmental limitations of traditional manual meter reading， intelligent meter reading has become the future development direction. The common types of electric meters can be mainly divided into mechanical word-wheel and liquid crystal electric meters. Among them， due to the deflection problem of mechanical word-wheel electric meters， there is a lack of character feature information in the process of character recognition， which leads to a low accuracy of this electricity meter type recognition. In order to solve this problem， this paper modifies the backbone network of YOLOv5 recognition algorithm， which improves the recognition effect of the algorithm on deflection characters of mechanical word-wheel electricity meter. Firstly， CBAM attention mechanism is introduced into the network model， which improves the feature extraction ability of the network model for deflected characters. Secondly， the Focus operation is replaced by a 6×6 convolution， and the original SPP pooling structure is replaced by a faster SPPF pooling structure to improve the operation speed of the algorithm. In order to test the recognition effect of the model， 329 deflection character samples of electric meters are collected for experiments， and the overall recognition accuracy can reach 99.4%. At the same time， 1 500 samples of liquid crystal electric meters are collected to test the generalization of the model， and the recognition accuracy reaches 99.6%. The experimental results show that this method solves the problem of low recognition rate of deflected characters， and verifies that the recognition model has strong generalization.关键词：reading recognition of the electricity meter;YOLOv5;cbam attention mechanism;pooling structure;deflect character
- 37
- |
- 27
- |
- 0
发布时间：2024-10-09 - “在工业检测领域，研究者提出了一种改进的YOLOv7算法，通过增加小目标检测层和无参注意力机制，显著提升了小目标焊点缺陷的检测精度，为边缘设备检测提供了新方案。”摘要：Aiming at the problems of the existing small target solder joint defect detection methods， such as error detection and leakage detection， an improved YOLOv7 small target solder joint defect detection algorithm was proposed. Considering the small size of solder joints， a small target detection layer and detection head were added to extract more shallow feature information. The non-parametric attention mechanism （SimAM） was introduced to assign 3D weights to the feature graphs to improve the feature extraction ability of the model.Partial Convolution （PConv） was used to reconstruct ELAN modules to reduce redundant operations and memory access， and GiraffeDet was used to integrate different scale features at the neck to improve the lightweight of the model. Finally， the NWD（Normalized Wasserstein Distance） loss function was used to improve the original CIoU loss function， which sped up the convergence of the model and improved the detection accuracy of small targets.Experimental results show that the average detection accuracy of the improved YOLOv7 algorithm reaches 90.3%， which is 5.1% higher than that of the original algorithm. The recall rate is 3.2% higher， the number of parameters is 36.3% lower， and the convergence speed has been greatly improved. This algorithm provides a reference for detecting small target solder joint defects in edge equipment.关键词：image processing;defect detection;YOLOv7;SimAM;lightweight;NWD
- 81
- |
- 30
- |
- 0
发布时间：2024-10-09 -
Design and implementation of piecewise approximation bilateral filtering algorithm based on FPGA AI导读
“在图像处理领域，研究者设计了分段近似双边滤波算法，有效降低硬件资源消耗，为图像滤波应用提供新方案。”摘要：In order to enhance the quality of image display and reduce the consumption of hardware resources in the bilateral filtering algorithm based on field programmable logic devices （FPGAs）， a piecewise approximation bilateral filtering algorithm is designed. The piecewise approximation reduces the storage capacity of the range domain and the bit width of the output data in bilateral filtering， thus reducing the computational complexity and hardware resource consumption. The bilateral filtering algorithm of piecewise approximation is implemented on the FPGA platform of Zynq-7000 and Sparten-7， and the performance of filtering and edge preserving under different Gaussian noise and its optimal range standard deviation is studied. The results show that the performance is comparable to that of the traditional bilateral filtering algorithm， the use of look-up table （LUT） and digital signal processing module （DSP） is reduced by 9.9% and 71.1% compared to the conventional bilateral filtering algorithm， and the power consumption is only 0.128 W. The algorithm is suitable for image filtering application scenarios with limited hardware resources.关键词：image processing;bilateral filtering algorithm;FPGA;piecewise approximation- 48
- |
- 47
- |
- 0
发布时间：2024-10-09 - “在卫星遥感图像目标检测领域，研究者提出了一种新方法，通过滑动窗口分割和小目标检测器结合，显著提升了检测速度和准确率。”摘要：The high resolution of satellite remote sensing images and the small relative size of the target within the image make it difficult to ensure both detection accuracy and operation speed. In order to solve the problem of target detection in high pixel remote sensing images， this paper proposes a detection method that combines sliding window segmentation and a small target detector. Firstly， the image is segmented into multiple subgraphs using the sliding window method， the sliding step is slightly smaller than the size of the window to make each subgraph have a certain overlap between them， and a larger segmentation window is used to reduce the number of subgraphs segmented. After that， the subgraphs are compressed and the compressed images are processed using a target detection algorithm to reduce the running time of the algorithm. Finally， the detection results are merged and a non-maximization suppression strategy is used to remove the targets that are repeatedly detected in the overlapping parts. In terms of detection algorithm， based on YOLOv8n， this paper uses SPD convolutional kernel and NWD to improve the network structure， and adjusts the feature pyramid structure to improve the algorithm’s performance in detecting small targets， which enables the algorithm to adapt to compressed subgraphs at larger sizes in order to reduce the number of image segmentation and improve the detection speed. The experiment proves that on the vehicle detection dataset with an average image resolution of 4 000×4 000， the average accuracy of the method for target detection is 55.7%， and the average computation time per image is 47.5 ms. The accuracy is improved by 17% compared to YOLOv8n，15% compared to YOLOv5s and 7% compared to YOLOv6s. The operation efficiency of the proposed method meets the real-time requirement， which is capable of detecting targets in satellite remote sensing images in real time with higher accuracy.关键词：object detection;Remote sensing images;YOLOv8;Small target detection;sliding window method
- 30
- |
- 15
- |
- 0
发布时间：2024-10-09 - “在遥感图像变化检测领域，研究者提出了一种融合CNN和Transformer的混合模型，有效提升了检测精度和效率。”摘要：Modern high-resolution remote sensing images have achieved remarkable results in change detection with the aid of convolutional neural network （CNN）. However， the limited receptive field of convolution operations leads to insufficient learning of global context and long-distance spatial relationships. While visual Transformers effectively capture dependencies in remote features， their handling of details in image changes is insufficient， resulting in limited spatial localization capabilities and low computational efficiency. To address these issues， this paper proposes a multi-level cross-layer linear fusion end-to-end encoding-decoding hybrid CNN-Transformer change detection model based on dilated spatial pyramid pooling， combining the advantages of visual Transformers and CNN.Firstly， image features are extracted using Siamese CNN， refined through dilated pyramid pooling to better capture detailed feature information. Secondly， the extracted attributes are converted into visual words， and a Transformer encoder models the compact visual words， feeding the learned context-rich labels back into visual space through a Transformer decoder to reinforce the original features. Thirdly， CNN features are fused with the features from Transformer encoding-decoding through skip connections， facilitating the fusion of position and semantic information by connecting features of different resolutions through upsampling. Finally， a difference enhancement module generates difference feature maps containing rich change information.Comprehensive experiments conducted on four publicly accessible remote sensing datasets， including LEVIR， CDD， DSIFN and WHUCD，confirm the efficacy of the proposed approach. Compared with other cutting-edge techniques for detecting changes， the model presented in this paper achieves superior classification performance， effectively addressing issues such as under-segmentation， over-segmentation and rough edge segmentation in change detection results.关键词：Remote sensing images;change detection;convolutional neural network;Transformer;atrous spatial pyramid pooling
- 45
- |
- 75
- |
- 0
发布时间：2024-10-09 - “在图像重建领域，研究者构建了以金字塔方差池化模块为核心的生成网络，有效提升了图像放大4倍时的峰值信噪比和结构相似性，增强了图像细节的真实感。”摘要：To reduce the impact of high-frequency information loss on image reconstruction and further enhance the mining of feature information， a generation network is constructed with the pyramid variance pooling module as the core. Firstly， the network uses different levels of variance pooling to mine the feature information contained in low-resolution images， and combines the pyramid structure to obtain the context information of different scales and different sub-regions， so as to further enrich the amount of feature information. Then， the dense connection structure is used to enhance the correlation of feature information to improve the expressive ability of the network. Finally， the group normalization operation is introduced to strengthen the convergence of the network. The experimental results show that compared with other methods on the open test sets of Set5， Set14， and DIV2K100， the peak signal-to-noise ratio increases by an average of 0.509 dB and the structural similarity increases by an average of 0.016 when the amplification factor is 4. The proposed model not only improves the peak signal-to-noise ratio and structural similarity to a certain extent， but also has more realistic details in the visual effect of the reconstructed image.关键词：image super-resolution;generative adversarial network;Variance pooling;dense connection
- 50
- |
- 4
- |
- 0
发布时间：2024-10-09 - “在图像超分辨率领域，研究者提出了一种结合残差学习和层注意力的轻量级算法RLAN，有效提升了图像重建质量，减少了伪影。”摘要：Convolutional neural networks （CNNs） have shown great performance in image super-resolution （SISR） problems. However， most super-resolution studies use complex layer connection strategies to improve feature utilization， which makes the depth and the number of parameters of the network increase continuously， and makes it hard to deploy on mobile terminals. Aiming at this problem， a lightweight image super-resolution network combining residual learning and layer attention is proposed to extract and aggregate important features more efficiently. Firstly， a 3×3 convolutional layer is used for shallow feature extraction. In the nonlinear mapping part， the improved local residual feature blocks （RLFB） are stacked for local feature learning， and the layer attention module （LAM） is introduced to further improve the effect of feature aggregation by using the hierarchical features on the residual branch. Finally， the pixel attention reconstruction block （PARB） is used for image reconstruction to improve the reconstruction quality with a small parameter cost. Compared with the NTIRE 2022 champion RLFN， RLAN finally achieves superior performance with only 373k parameters， and the average PSNR and SSIM on the four datasets are improved by 0.35 dB and 0.001 4， respectively. The comprehensive experiments demonstrate that RLAN can accurately restore SR images and effectively reduce the artifacts at the edges.关键词：image super-resolution;convolutional neural network;residual learning;attention mechanism
- 55
- |
- 20
- |
- 0
发布时间：2024-10-09 - “在三维重建领域，研究者提出了基于NeRF的框架，有效解决遮挡问题，为新视图合成提供新思路。”摘要：Single-view 3D reconstruction aims to restore the three-dimensional geometry of an object or scene based on a single 2D image.The limited information provided by a single view often results in occlusion， leading to blurred image features and inhibiting the accurate recovery of object appearance details. This paper introduces a framework， called neural radiation field （NeRF）， that effectively addresses the occlusion problem by utilizing both global and local context information of the image. The proposed approach employs the Vision Transformer to capture long-range correlations and learn global features from the image. The Vision Transformer is combined with the SE channel attention mechanism module to prevent information loss and redundancy across multiple layers. Additionally， a convolutional neural network is utilized to extract pixel-aligned local image features. The dilated convolution of the dilated pyramid pooling structure is employed to increase the receptive field， extract multi-scale context information， and provide more details for restoring occluded areas. Lastly， a density feature aggregation module based on the Transformer architecture is designed to minimize inaccuracies in density prediction due to occlusion. Experimental results on the ShapeNet-NMR dataset demonstrate the method's ability to produce new views with enhanced details and exhibit strong generalization capabilities when applied to unseen objects.关键词：Neural radiation field;Shading;Null convolution;Transformer
- 31
- |
- 2
- |
- 0
发布时间：2024-10-09 - “在水下图像增强领域，研究者提出了一种新方法，通过改进颜色线模型和优化算法，有效提升了图像清晰度和颜色准确性。”摘要：To solve the degradation problems such as hazing and color distortion of images captured by underwater devices， an underwater image enhancement method based on improved color-line model is proposed. Firstly， an underwater background light estimation method based on improved quadtree subdivision is proposed to eliminate the interference of underwater factors and obtain more accurate background light estimates. Then， a local transmittance optimization model based on the color-line law is established， and a new Gauss-Seidel type inertial proximal alternating linearized minimization algorithm （GiPALM） non-convex optimization method is designed to solve the transmittance， which improves the convergence speed of the model and yields a more accurate transmittance estimation at the same time. Finally， based on the background light and transmittance estimation to obtain the recovered image， linear stretching is further used to correct the color information of the image to obtain an underwater enhanced picture that conforms to the sensory vision of the human eye. The experimental results show that our method is superior to other comparative algorithms in terms of qualitative evaluation， quantitative evaluation， color accuracy and application testing. Our method shows excellent performance and effectively improves the clarity and visibility of underwater images.关键词：Underwater image;Color-line model;Non-convex optimization method;Dehazed;image enhancement
- 121
- |
- 34
- |
- 0
发布时间：2024-10-09 - “在图像语义分割领域，研究者提出了一种新方法，通过双重聚合和自合并网络，有效提升了小样本图像的分割精度，为新类对象识别提供了新思路。”摘要：Few-shot image semantic segmentation is a very challenging task that attempts to segment objects of new classes using only a few labeled samples. The mainstream methods often have problems of low discriminative feature and prototype deviation. To alleviate these problems， a new few-shot image semantic segmentation method based on a bi-aggregation and self-merging network is proposed， which can fully mine the similarity of features and reduce prototype bias. Firstly， we propose a feature-mask bi-aggregation module to provide global semantic information for the feature aggregation and mask aggregation by constructing a dense similarity relation between the support features and the query features covering all spatial locations. Specifically， an enhanced feature and an initial mask with guiding information can be obtained for the query image by performing feature and mask bi-aggregation on the similarity matrices. Then， a self-merging decoder is proposed， which reduces the prototype bias by adding the initial mask-based self-prototype with the known support prototypes， and conveys rich category semantic information to the decoder by fusing the merged prototype with the enhancement feature. Finally， the prediction results obtained by the decoder are further optimized by the prediction results of the base classes. The mIoU values of our method on the dataset PASCAL-5i achieve 68.3% and 71.5% in the 1-shot and 5-shot cases， respectively， and on the dataset COCO-20i achieve 46.5% and 51.4% in the 1-shot and 5-shot cases， respectively， which is superior to the segmentation performance of the mainstream methods， and can segment the target region of the new class more accurately.关键词：Few-shot semantic segmentation;Similarity of features;Bi-aggregation;Intra-class diversity;Self-merging
- 70
- |
- 18
- |
- 0
发布时间：2024-10-09
Image Processing
- Address：No.3888 Dong Nanhu Road, Changchun, Jilin, China 130033 Postal code：130033
- Tel：0431-86176059 Email：yjxs@ciomp.ac.cn
- Technical support is provided by Beijing Founder electronics co., LTD 吉ICP备11002662号-17
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0