The change of temperature can affect the electrical properties of liquid crystal (LC) materials. In order to further explore the electrical properties of LC materials at low temperature, firstly, the thicknesses of the LC cells and the polymide (PI) alignment layers were measured by UV spectrophotometer and surface profiler respectively. Then, a precision hot stage and a self-made refrigeration device based on the working principle of thermoelectric refrigeration were used to control the temperature of the LC cell, and the precision LCR meter which was calibrated was applied to measure the capacitance-voltage (C-U) curve of LC cell at low and normal temperatures. The C-U curve of LC layer could be obtained by using the capacitance model of LC cell, and based on this curve and dual-cell model, the effects of temperature on dielectric anisotropy and elastic constant of four kinds of LCs were obtained. The experimental results show that the dielectric anisotropy of the LC with positive dielectric anisotropy increases from 10.0 to 20.0, and the elastic constant k11 decreases from 17.0 pN to 7.0 pN, and k33 decreases from 31.0 pN to 15.0 pN. The dielectric anisotropy of the LC with negative dielectric anisotropy increases from -8.5 to -3.0, decreases from 21.0 pN to 11.0 pN for k11, and from 31.0 pN to 12.0 pN for k33. The dielectric anisotropy and elastic constant k11 and k33 of the LC with positive dielectric anisotropy decrease with the increase of the temperature, and increase and decrease respectively with increasing temperature for the LC with negative dielectric anisotropy. This conclusion is of great significance for exploring the methods to improve the performance of liquid crystal display at low temperature.

In order to better study the polymer stabilized blue phase liquid crystal devices, considering the difference of dielectric constant between polymer and blue phase liquid crystal and the change of electric field, we established a theoretical model suitable for polymer stabilized blue phase liquid crystal display. Firstly, we calculated the simulation result to fit the experiment results of blue phase liquid crystal displays with two group of polymer ratio, and obtained the saturation birefringence of blue phase liquid crystal close to the birefringence of liquid crystal. Then, we studied the influence of dielectric constant and content of polymer monomer on driving voltage. Finally, we calculated the influence of electrode width and gap on the operating voltage and transmittance of polymer stabilized blue phase liquid crystals devices driven by in-plane electric field, and obtained the device structure with high transmittance and low operating voltage. With the increasing dielectric constant of the non-liquid crystal material, the operating voltage is decreased, and the greater the content, the greater the operating voltage decreased. For the non-liquid crystal material content of the general blue liquid crystal being about 10%, the operating voltage can reduced from 97 V（εP=3）to 55 V（εP=30), the change of the maximum transmittance is very small, even can be ignored. In addition, the different dielectric constant of non-liquid crystal materials only affect the operating voltage, and has a little effect on the maximum transmittance and the shape of electro-optic curve. The results have important guiding significance for optimizing the material composition of polymer stabilized blue phase liquid crystal and improving polymer stabilized blue phase liquid crystal devices.

In order to improve the electrical performance of the thin film transistors, a kind of high dielectric constant material is used instead of the conventional SiO2 as the gate dielectric. Compared with the SiO2, Ta2O5 has a high-k value with semiconductor process compatibility, which exhibits the potential to overcome the problem of leakage current and thereby become a new generation of gate dielectric materials. Firstly, Ta2O5 thin films were prepared by magnetron sputtering. AZO-TFTs were fabricated with Ta2O5 as gate dielectric. Then, the effects of oxygen to argon ratio on the properties of Ta2O5 thin films and the electrical properties of AZO-TFT were studied. The influence of surface morphology and roughness of Ta2O5 thin films on the performance of thin film transistors was analyzed. Finally, the electrical properties of SiO2 gate dielectric and Ta2O5 gate dielectric thin film transistors were compared. Experimental results indicate that the performance of TFT is the best when the oxygen argon ratio of TFT is 10∶90. Compared with the TFT using SiO2 as the insulating layer, the on-off current radio of TFT with Ta2O5 as the insulating layer increases from 1.02×103 to 2×104. The subthreshold swing decreases from 5 V瘙
簚dec-1 to 1.5 V瘙
簚dec-1. The field effect mobility increases from 1.6 cm2/(V·s) to 12.2 cm2/(V·s). The electrical performance of the TFT has been improved to some extent.

In order to obtain the radiation spectrum of different light source color temperatures, a method based on the combination of colored glass and coating for color temperature conversion was proposed. The halogen lamp with 3 000 K color temperature was used as the standard light source and transformed it into 4 800 K standard blackbody radiation color temperature. According to the target color temperature transmittance spectrum curve, QB11 was selected as the substrate. In order to further modified the radiation spectrum, the optical film is coated on QB11 substrate. A 5-layer film system composed of TiO2 and SiO2 was designed by using Essential Macleod film design software. After optimization, the film thickness was 353 nm. The experimental results show that, the 3 000 K color temperature of standard halogen lamp can be effectively transformed into 4 800 K color temperature of standard blackbody radiation by the combination of color glass and optical film. The spectral deviation quality index was 3.67%. From the final spectral deviation quality index, the color temperature transformation method is effective through above method.

In order to improve the stereo feeling of display system, and generate a screen-free aerial image, a floating display system is established and its applied basic theory and key optical devices are investigated. Firstly, the two main factors of floating display are studied based on the binocular stereo vision of human: the first is the display system can generate a real image, and the second is the real image can be observed by both eyes and the binocular parallax is achieved. Then, based on the analysis of floating display, a display system based on lens is established to verify the feasibility of the theory. Finally, according to the simulation analysis, a floating display system using micro lens array is designed and established. Experimental results indicate that the images are displayed in mid-air without any auxiliary screen, and the aerial images are at least 6.5 cm away from the display system. It verified the analysis of floating display and the design of floating display system.

Ambient light has always been one of the essential obstacles to improving display devices image quality. In this study, a tone mapping algorithm is proposed, which can adjust the luminance and color of the image to achieve reliable luminance mapping and color correction of the image in different luminance ranges. Firstly, the influence of ambient light on the display device is analyzed, and the display device model is established. Secondly, the contrast perception model was used as a bridge to solve the luminance mapping curve to minimize the contrast perception difference. Then, using the color appearance model CAM16, the original image is mapped into the environment where the display device is located, while the perception of color and brightness remains unchanged. Finally, luminance mapping and color correction are combined, and a soft clip function is designed to correct the overexposure problem. Experimental and simulation results show that under different environments, the algorithm can repair the color deviation caused by luminance compression, reduce the color error to less than 60% of the traditional method, reduce the contrast perception error, and obtain a stable image performance. The requirements of display devices in a complex environment can be met.

To ensure display stability of LCD in a certain process fluctuating range, LC Margin is usually estimated for new products developing. Especially, the property of photo spacer (PS) mattered significantly for LC Margin. This work focuses on influence factors of PS characters. First of all, the relationship between ambient temperature and PS Young’s modulus is analyzed to detect their variety trend. Then, the influence of process temperature on PS elastic recovery ratio(PS ER) is studied. Finally, the elastic recovery rate of PS with same specific characters at different cell gap is discnssed. Related experiment results indicates that PS Young’s modulus decreases with the increasing of ambient temperature, and the PS ER increases slightly with the in creasing of process temperature. On the other hand, the PS in higher cell gap recovers a little more quickly refer to lower cell gap.

Event-based vision sensor is a new type of bionic vision sensor, which is more similar to the working mechanism of human eye, and makes it widely concerned. Different from the working mechanism and output mode of traditional frame-based cameras, the pixels of event-based vision sensors can individually detect logarithmic change of light intensity, and output event information including location, time, and polarity when the amount of change exceeds a certain threshold. The event-based vision sensor has the advantages of low latency, high dynamic range, and low power consumption. Its unique output mode and working characteristics make it especially suitable for occasions with high-speed movement, large changes in lighting conditions, or low energy consumption. This article introduces the development history, classification, working principle, advantages and disadvantages of event-based cameras, as well as their applications in the fields of rapid motion tracking and monitoring, target recognition, simultaneous localization and mapping(SLAM) in recent years. Finally, it summarizes the challenges that the event-based camera still exists in different application fields, and looks forward to its future development. The wide application of event-based cameras can provide new solutions for high-speed motion and high dynamic range occasions where traditional cameras are still tricky. With continuous updates and development in the future, it will be able to play a role in more complex application scenarios.

Camera calibration is the most basic part of the vision system. It uses the camera imaging principle to solve the internal and external parameters of the camera model, so as to better realize the application of the vision system in practice. Based on the analysis of the characteristics of camera calibration and the development of camera calibration at home and abroad, the calibration objects are divided into three-dimensional calibration method, two-dimensional calibration method, one-dimensional calibration method and zero-dimensional self-calibration method. This paper introduces firstly the origin and development of each camera calibration method, then analyzes the basic principle of each camera calibration method in detail, and summarizes the advantages and disadvantages of each calibration method. The development trend of calibration technology is predicted. In the case of complex scene and large amount of computation, the main research contents should be high precision, high efficiency and high robustness, so as to further expand the scope of application and enhance the practicability. At the same time, the cost of camera calibration should also be taken into account, to find a more suitable calibration method.

A motion image deblurring algorithm based on a deep residual generative adversarial network is proposed for the motion image blurring problem arising from motion, jitter and electronic interference during image capture. Firstly, this paper investigates the image blurring model and the blind deblurring process. Secondly, the generative adversarial network is introduced, and the structure of the residual block is improved. The improved residual block contains three convolutional layers, two ReLU activation functions, a Dropout layer, and a skip connection block, which improves the quality of the recovered image. Thirdly, the structure of PatchGAN is improved, and the receptive field of the lowest layer is more than twice of the original one with only a few additional paramters and network complexity. The tests are conducted using the GOPRO dataset and Lai dataset. The test results show that the deblurring algorithm based on deep residual generation adversarial network proposed in this paper can achieve high objective evaluation indexes and can recover clear images of high quality. On the GOPRO dataset, compared with other similar methods, the algorithm proposed in this paper has better recovery ability and can achieve higher peak signal-to-noise ratio (28.31 dB) and higher structural similarity (0.831 7). On the Lai dataset, the higher quality images can be recovered.

Aiming at the problems of loss of detailed feature information and blurred contours of blood vessels in the segmentation process of retinal vessels, an improved cycle segmentation confrontation network algorithm is proposed. The algorithm improves the network model of the segmenter, adds dense connection structure in the up and down sampling processes of the U-Net network, fully retains the image feature information, which improves the generalization ability and robustness of the model, and alleviates the over segmentation phenomenon. In order to prevent network degradation, the loss function is replaced by the least square function, which improves the quality of image segmentation and the stability of network model training. The experimental results show that the segmentation accuracy and sensitivity of the network are 96.93%, 84.30% and 96.94%, 79.92% in the DRIVE and CHASE datasets respectively. The algorithm has good network generalization ability and segmentation accuracy, which can provide an important basis for disease diagnosis.

Traditional high dynamic range (HDR) imaging methods need to capture multiple images, which is not only easy to introduce field of view errors, but also difficult to apply to the emerging image acquisition system. To this end, this paper proposes a simple and effective HDR imaging method based on remapping and exposure fusion. This method only needs to capture a pair of short and long exposure images, and a series of multi-exposure sequence can then be generated through remapping. Next, according to the exposure evaluation, the weight map of the sequence can be obtained. The weight map is decomposed by Gaussian pyramid, and the multi-exposure sequence is decomposed by Laplacian pyramid. Finally, a high-quality HDR image with good details is generated based on multi-scale fusion. Experimental results show that the proposed method performs better in multiple scene tests, and the image quality is greatly improved in the MEF-SSIM score compared with some mainstream algorithms. The image qulity is improved by 13.33% compared with GF-MEF, and improved by 2.62% compared with CNN-MEF. The proposed method provides a new reference for the field of HDR imaging, and can be potentially applied to CCD/CMOS-based imaging systems for high-quality image acquisition.

SRGAN is a typical method of image super-resolution based on deep learning, the reconstruction effect is good, but the algorithm still has some shortcomings, and there is still more room for improving the image quality and operation speed. An optimization model is proposed based on the SRGAN network model. Because the batch normalization (BN) layer often ignores some image details in super-resolution image reconstruction and increases the complexity of the network at the same time, the BN layer is removed from the generator of SRGAN and the ECA channel attention is introduced so that each residual block generating feature map gets a corresponding weight in order to process more image details. After training and comparison experiments on public datasets, the results show that the proposed improved model has richer image details recovery, better visual effects, better peak signal-to-noise ratio and structural similarity performance, and fewer total number of model parameters compared to the comparison model.

To address the problems of poor detection flexibility, low yield rate and long detection time of smartphone glass cover defect detection methods, an improved YOLOv3 defect detection method for smartphone glass cover is proposed. A channel attention mechanism is added to the feature extraction network to solve the problem of inconspicuous defect features, a feature map of 104×104 dimensional size is added to the feature detection network to solve the problem of multi-scale defects, and finally the model is pruned to reduce the model parameters to improve the defect detection speed. The defect dataset is constructed by obtaining images covering four types of defects, such as chipped edge, pit, dirty and scratches, from the production site of smartphone glass cover. The proposed method and algorithms such as Faster R-CNN, YOLOv3 and YOLOv4 are compared for experiments and analysis. The experimental results show that the detection mAP (mean average precision) of the proposed method is 81.0% and the detection speed is 43.1 fps. Compared with the original YOLOv3 algorithm, the detection mAP is improved by 3% and the detection speed is increased by 6.7 fps. Compared with other deep learning algorithms, the detection speed and detection precision are improved. The proposed method meets the need for high-precision and efficient detection of defects in the industrial production site of smartphone glass covers.