In order to make up for the shortcomings of the traditional liquid crystal orientation, a low-cost, simple and better orientation technology was developed. In this study, a simple experimental method was used to self-assemble ZnO nanoparticles onto indium tin oxide (ITO) glass substrate, and the nanostructure of ZnO was regulated by changing the calcination temperature. Then combined with the friction method, the preinclination of liquid crystal molecules can be adjusted. The experimental results show that the higher the calcination temperature of ZnO is, the more the preinclination of the liquid crystal molecules is close to 0°. On the contrary, the lower the calcination temperature is, the closer the liquid crystal preinclination angle is to 90°. In this study, by adjusting the self-assembled ZnO nanostructure, the preinclination of liquid crystal molecules in the range of 0°~90° was controlled.

Polymer dispersed liquid crystal (PDLC) is widely used in projection displays, smart windows and other fields. Patterned flexible smart window based on PDLC would further expand its application. In this paper, a patterned flexible PDLC smart window is demonstrated using one-step UV exposure. The PDLC precursor is sandwiched between two flexible PET-ITO substrates, and a layer of patterned mask with specific UV attenuation is attached on the cell for UV curing. The linewidth resolution of 120 μm is obtained for the proposed patterned flexible smart window. In this paper, the different electro-optic characteristics of the different regions of patterned device are studied. The experimental results show that the PDLC with high UV curing region has high threshold voltage and saturation voltage. The rising times of high UV curing region and low UV curing region are 4.1 ms and 15.4 ms, and the decay times are 14.1 ms and 33.7 ms, meeting the application requirements of smart windows. By applying different voltages, the patterned smart window can present three modes: total scattering state, total transparent state and patterned transparent display state. Finally, the pattered smart window working in plane state and bending state are exhibited in the paper. This study shows that the patterned PDLC flexible device prepared by the above one-step UV exposure has the advantages of simple fabrication, low cost and high precision, which is conducive to its potential applications in the field of customized flexible smart windows.

The eigenmode analysis of liquid crystal optical waveguides (LCOWs) was the basis and theoretical support for the design of its devices. The accuracy of mode field analysis directly affects the performance of LCOW-based devices. We used the following procedure to accurately analyze the relationship of the eigenmode field components in the LCOW. Firstly, the magnetic field coupling equations of the nematic liquid crystals were obtained from Maxwells equations. Then, the quantitative relationship between the rotation angle of the liquid crystal director and the ratio of the magnetic field components of the eigenmodes excited by the extraordinary light wave was obtained. Additionally, the numerical simulation is carried out using OptiBPM13.1. The simulation results verify the quantitative relationship that the ratio between field components and the tangent function of director rotation angle are opposite to each other. According to this relationship, the relative order of magnitude between the transverse magnetic field components of the eigenmode field in the LCOW can be controlled by changing the applied voltage. Then, the conversion of polarization mode excited by an extraordinary light wave can be realized. This law is helpful to promote the research of polarization rotator and mode converter based on LCOW.

The deformable mirror is one of the core units of the Adaptive Optics (AO) System. This article proposes a new voice coil actuator structure based on the principle of minimum magnetoresistance. The soft magnetic material-Permalloy is added to the traditional voice coil structure design to reduce magnetic leakage and thus improve output force and efficiency. The model is established by ANSYS Maxwell for simulation and the structural dimensions of each component are optimized. The results show that the actuator structure can generate a force of 3.4 N, an efficiency of 9.05N×W-1/2, and a response time of 0.03 ms. At the same time, it is found that the working influence between adjacent actuators is very small, so it is established that the coupling of deformable mirror is 11.7% when the actuator spacing is 15 mm, which meets the system requirements. This work lays a solid foundation for the development of deformable mirrors with large travel, fast response, low power consumption and high actuator density, which have good application prospects.

The three-dimensional medical electronic endoscope system is widely used for medical diagnosis, minimally invasive surgery and medical training. The geometric distortions during the visualization of the three-dimensional endoscope will affect the relative size and distance estimation accuracy of the nidus for the medical staff, and also may be one of the important factors inducing the visual discomfort. Existing researches only focused on the distortions of the two-dimensional image and its correction. The three-dimensional space geometric distortions during the visualization due to the mismatching of the three-dimensional display system and the characteristics of human eye three-dimensional perception are usually ignored. Based on the analysis of three-dimensional medical electronic endoscope image acquisition, display and perception process, the model of three-dimensional medical electronic endoscope system is established. The ideal three-dimensional endoscope image visualization is the equally magnified virtual space based on the real-world space in all dimensions, and the matching of parameters between three-dimensional endoscope image capture and visualization can achieve this goal, and finally realize the precise visual perception and well visual comfort. The results may provide some theoretical guidance for more precise and comfortable utilization of the three-dimensional image capture and visualization in medical domain.

With the improvement of the quality of life, large-size, high-refresh frequency, and high-resolution display devices are becoming more and more popular. However, high-spec products are also accompanied by more display problems, and vertical crosstalk is one of them. One of the cause of vertical crosstalk is mainly due to the coupling capacitor Cpd between the data line and the pixel electrode, another cause is the leakage current when the TFT is turned off (Ioff). The great reduction in storage capacitance (Cst) and increase in line density of high-resolution products (8 K) result in serious vertical crosstalk. This paper simulates the influence factor of Cpd through software, and also determine the influence of Cpd based on the phenomenon of the samples with different pixel electrodes 2ITO overlap. At the same time, the best storage capacitance and leakage current conditions are determined by changing various process parameters; with the optimum storage capacitor and leakage current process parameters, the matching 2ITO overlap is also discussed. With all the optimal process parameters, the defect ratio drops from the initial 55.6% to 4.2%, and the image quality is greatly improved.

At present, there are only some qualitative design rules for via hole connected by ITO in TFT-LCD panel, and there is lack of quantitative calculation and design method of via hole. In this paper, the quantitative calculation method for the resistance of this kind of via hole is studied, and the relationship between the resistance and the breakdown current of via hole is explored, which provides guidance for us to design via hole reasonably to prevent the hole from burning. Firstly, by analyzing the structure of via hole, the equivalent resistance circuit diagram is abstracted and the decomposition formulas for the resistances of via holes with various structures are obtained. The resistance of via hole is mainly composed of the contact resistances of deep and shallow holes, and the resistance of ITO between the deep and shallow holes. Then, by designing the contact holes between metal and ITO in different sizes, Kelvin Four-terminal sensing method is used to measure the relationship between the contact resistances of deep and shallow holes and the hole sizes. Meanwhile, the resistance of ITO between the deep and shallow holes can be calculated by the square resistance and size of ITO between the deep and shallow holes. Therefore, we can calculate the resistance of via hole with various structures only according to the hole size and the square resistance of ITO. By analyzing the correlation between the measured and calculated resistance values of a large number of via holes, the linear correlation coefficient is 0.96, which proves the reliability of the calculation method. Finally, by measuring the resistance values and breakdown current values of a large number of via holes, a significant power function relationship between the resistance value and breakdown current is found, the power index is around -1.3, and the correlation coefficient is 0.98. Thus, we have established a set of guiding methods about resistance calculation and breakdown current evaluation of via hole quantitatively.

Image quality assessment (IQA) is one of the basic research issues in the field of computer vision. At present, most image quality models are constructed based on grayscale images, and color image quality assessment is still an open issue in the field of IQA. The key of color image quality assessment research is to construct a quantitative description of color information consistent with human color cognition. This paper constructs a color image quality assessment model based on colornames(CN). It maps each pixel value of the image to a CN probability vector, uses the Wasserstein distance to calculate the perceived color difference of two images, uses the lightness and gradient features, and uses the saliency weighting in the pooling stage to obtain the objective image quality scores. The experimental results on the public test databases show that the proposed model performs best on TID2008, TID2013 and the latest KADID-10k databases, with SROCC values of 0.900 9, 0.890 1 and 0.863 7, respectively. The overall assessment effect is comparable to the current best traditional method (Non-deep learning method). But for color distortion, it has obvious advantages.

In order to evaluate the image quality better and solve the problem of the differences between the block images that are obviously ignored in the image quality evaluation model based on the convolutional neural network (CNN-IQA), a CNN model of multi-feature fusion is proposed. Firstly, the whole image is divided into non-overlapping blocks, the information entropy and texture feature of each divided image are extracted. Then, the two features are combined to calculate the importance weight of each block image to measure the influence of the block image on the quality of the distorted image. Finally, the loss function is modified according to the calculated importance weight to highlight the role of the block image with high importance in the training process. Validation and comparison experiments on the LIVE data set found that the SROCC and LCC indicators of the algorithm are 0962 and 0.960, which are higher than the original algorithm at least 0.9%. The validation and comparison experiments on the TID2008 data set show that the SROCC and LCC indicators obtained by the algorithm are 0.922 and 0.924, which are higher than the original algorithm at least 0.6%. And the results on the two data sets are better than other comparison algorithms. The experimental results prove that it has good performance and generalization in predicting image quality.

In view of the problem that the quality of high dynamic range image appearance needs to be improved, a tone mapping algorithm based on collaborative filtering is proposed in the chromaticity brightness color space. Firstly, the brightness and chromaticity information are extracted from the input high dynamic range image using chromaticity brightness color space. Then, they are reconstructed and decomposed, respectively. Bilateral filtering technology is utilized to decompose the brightness information and obtain the brightness basic layer and brightness detail layer. According to the consistency of the large-scale edge of brightness and chromaticity information, a chromaticity brightness collaborative filtering algorithm is constructed to break down the chromaticity information, and obtain the chromaticity base layer and chromaticity texture layer. Finally, the brightness and chromaticity information are reconstructed and converted to the RGB color space to output the final tone mapping results. The results of objective index show that our proposed algorithm is improved by 25.24%, 18.89% and 45.89% respectively in image quality score, structural fidelity and image naturality. Therefore, the boundary details and color information can be better maintained.

Video inpainting, which aims at filling in missing regions of a video, remains challenging due to the difficulty of preserving the precise spatial and temporal coherence of video contents. In order to solve the problems of discontinuous semantic information, video blurriness and temporal artifact, and more and more complex network design, the overall speed of the network becoming slow, this paper proposes a residual convolution attention network (RCAN) for video inpainting. By introducing the self-attention mechanism and the global attention mechanism into the residual network, the ability of the network to learn the spatio-temporal features of all input frames is enhanced. This method proposes a spatial-temporal adversarial loss function to optimize RCAN, which improves the quality of video inpainting. At the same time, the network can define the number of layers and parameters with a high degree of freedom to improve the practical application ability of the network. Experimental results show that the network can achieve an average inpainting result in that the PSNR is 3068 dB, the SSIM is 0.961, and the FID is 0.113 on DAVIS and YouTube-VOS data sets. This method meets the inpainting quality requirements of the actual scene on the model and provides a new idea for video inpainting.

Aiming at the problem of trapezoidal distortion of the displayed picture due to the fact that the camera is not perpendicular to the monitoring picture in the monitoring system, a hardware structure design of an adaptive real-time video keystone correction system is proposed. Firstly, a series of preprocessing such as gray scale transformation, edge detection, and morphological opening operation are carried out in sequence for each frame. Then, based on the Hough transform, the outline of the trapezoid distortion in the picture is extracted. After that, four pairs of mapping point coordinates are determined according to the trapezoidal distortion contour and the connection point method is used to calculate the correction parameters. Finally, the gray-scale interpolation method is used to obtain the corrected image. The whole process uses pipeline and ping-pong buffer structure to optimize the algorithm, and uses the vertical blanking time in the video sequence to calculate, which greatly reduces the hardware resource consumption and improves the algorithm efficiency. It is implemented on the ALINX AX7103 FPGA hardware development platform. The experimental results show that when the video resolution is 640×480 and the refresh rate is 60 Hz, the system can correct the monitoring picture with the trapezoidal distortion range within ±30°, and the correction accuracy is 1°. Basically, the system can meet the self-adaptability and real-time requirements of the monitoring system.

Aiming at the problems of the current popular image dehazing algorithms, such as excessive dehazing which causes image color distortion or insufficient dehazing, an image dehazing algorithm with adaptive gamma correction estimation is proposed. Firstly, according to the brightness of the image, the different gamma correction functions are used to fit the relationship between the hazy image and the fogless image under different scene depths, and the minimum channel of fogless image is adaptively estimated. Then, the minimum channel of the non-fog image is corrected by the guided filtering algorithm to maintain the linear relationship between the hazy image and the non-fog image in the local area with the same depth of the scene. The initial transmittance according to the atmospheric scattering model is obtained. Then, the initial transmittance through Gaussian relativity is optimized to obtain the precise transmittance with structure retention and local smoothness. By increasing the search field, the upper half of the blue channel of the hazy image is used as input to improve the quadtree algorithm, and stably obtain the hazy image pixel value corresponding to the deepest part of the scene as the atmospheric light value. Finally, the brightness of the restored image is enhanced by the gamma correction function. The experimental results show that the contrast and average gradient of the restored image are increased respectively by 40.66% and 20.98% on average, and have higher information entropy. The above algorithm dehazes significantly, and the restored image has higher definition

Aiming at the problem that the accuracy and real-time performance of current aircraft remote sensing image target detection algorithms can not be balanced, an aircraft remote sensing image target detection algorithm based on single shot multibox detector(SSD) is proposed. Firstly, the improved deep residual network is used to replace the skeleton network of SSD. For the problems of lack of feature information association between feature maps, and different channels from feature maps have no weight values, this paper designs a new feature pyramid network with feature receptive field enhancement module and attention mechanism module. The network is used to fuse the feature information of different levels and the weight coefficients between the training feature channels, so that both the deep network and the shallow network can obtain fusion features with rich structural levels, which provides a good prerequisite for the classification and positioning of the subsequent network. In addition, the focus classification loss function is also used in the improved SSD algorithm to solve the problem of imbalance between positive and negative samples. Related experiments are carried out on the aircraft remote sensing data set, and the average accuracy reaches 92.45%, and the frame per second is 35.6. The results show that the improved SSD algorithm can balance high detection accuracy and real-time performance at the same time.