With the sound of energy saving horn, smart windows with dynamic adjustment of light transmittance are particularly important in today’s high-energy buildings. Due to the excellent external field response performance and good processing performance, liquid crystal/polymer composite systems exhibit various application prospects in smart window that dynamically modulates light transmittance. According to the regulation mechanism, smart windows fabricated by liquid crystal/polymer composite can be divided into electrochromic, thermoschromic, mechanochromic and photochromic. Among them, electrochromic and thermochromic windows are of great importance because of the feasibility of preparation and application. In recent years, the emerging technologies that combined smart window technology with solar cells, are expected to promote the green and intelligent innovation in the field of architecture. In this paper, five types of liquid crystal/polymer composite for smart window application were introduced, that is, polymer dispersed liquid crystal, polymer stabilized liquid crystal, a coexistent system of polymer dispersed and stabilized liquid crystal, polymer network liquid crystal, and polymer wall stabilized liquid crystal. Their microstructures and response mechanisms were summarized. The research progress and combined applications with other emerging technologies were reviewed. Finally, the future development trend of smart windows is briefly prospected.

Cellulose is an abundant renewable polymer resource. Cellulose nanocrystals (CNCs) with high crystallinity can be obtained by hydrolysis with strong acid. After evaporation at a suitable concentration, the suspension of CNCs is dried and self-assembled to form cholesteric liquid crystal film of CNCs. CNCs chiral liquid crystal film exhibits excellent circularly polarized light properties, and is of grant importance in the field of optoelectronic technology. In this review, the liquid crystal behaviors and preparation methods of CNCs are summarized. The structural color properties as well as the pitch regulation of CNCs film by physical and chemical methods are emphatically discussed. At the same time, the circularly polarized light properties of CNCs chiral liquid crystal film are comprehensively introduced. The research progress and application technology of circularly polarized light reflectance properties and circularly polarized luminescence emission performance are summarized and prospected. The difference between the left-handed and right-handed circularly polarized light properties of CNCs chiral liquid crystal film makes it have broad application prospects in liquid crystal displays, optical anti-counterfeiting, sensor devices and other fields. In addition, CNCs film materials with circularly polarized luminescent performance that could be prepared by a variety of methods (such as adding luminescent molecules or metals) can be used as chiral probes to detect excited states, which has potential application values and development space in a series of optical applications.

In order to research the generation and evolution of topological defects in nematic liquid crystal microfluidics within different interfacial constraints, we prepared microchannels with different surface anchoring conditions. Three microchannels with varied surface anchoring conditions were created by bonding a substrate with different surface properties to a structural layer made of polydimethylsiloxane, and utilizing oxygen plasma to vary the chemical properties of the channel wall in a reasonable manner. The flow morphology of nematic liquid crystal in microchannels was examined, and the emergence regular of various forms of topological defect structures was summarized. Experimental results indicate that the liquid crystals in the planar alignment microchannel are arranged continuously without obvious defect structure. The hybrid alignment microchannel forms stable defect lines near the upper surface under the competition between interfacial anchors, and the number of defect lines is related to the breadth depth ratio of the microchannel. The director field in a homeotropic alignment microchannel shows excellent continuity under normal conditions. The backflow generated by controlled flow rate mutation can effectively disrupt the continuity of director field within the microchannel, thus creating dynamic topological defects. When the flow rate is constant, the defective structure maintains a brief dynamic equilibrium, followed by regular fluctuations in the channel.

The polyvinyl alcohol (PVA) polarizing films were prepared by the flow-casting method. The crosslinking or uncrosslinking iodine-PVA polarizing films with different polymerization degree were obtained after the determination of iodization and crosslinking time by orthogonal experiments. The transmittance was measured and the polarization degree of PVA polarizing film was calculated by using UV-Vis before and after hygrothermal aging tests. The chemical morphology and crystal structure of the polarizing films with different polymerization degree were characterized by using Raman and XRD. The molecular structure of the polarizing films with different polymerization degree before and after hygrothermal aging was analyzed by using FT-IR, and the effects of polymerization degree and crosslinking on the optical properties and aging resistance of the polarizing films were systematically investigated. The results show that as the polymerization degree increases, the adsorption of PVA films to iodide ions improves, the proportion of I5- in films increases, and the hygrothermal resistance aging performance of polarizing films becomes better as well. From the perspective of the change of polarization degree of polarizing films, the higher degree of polymerization is, the better optical performance and the lower failure rate of aging are. Under the same polymerization degree, the hydrolysis resistance and the hygrothermal resistance aging performance of the polarizing films are significantly improved after crosslinking. High humidity promotes the disorientation of polarizing films and the hydrolysis of borate ester bond in crosslinking structure, resulting in the destruction of films structure and iodide ions orientation. The iodide ions loss is the main reason for the failure of polarizing films.

Aiming at the prospect of applying liquid crystal (LC) lens to zoom glasses, an electrically tunable-focusing and polarizer-free LC lens with an alignment conversion layer is developed. Through simulation, the feasibility analysis of double-layer modal LC lens was conducted. After optimizing the synthesis parameters of alignment conversion film, a polarizer-free LC lens sample was prepared. The imaging and optical interference of LC lens was studied to discuss the polarization independence and image quality. The test results show that the 8 mm aperture LC lens realizes polarization independence and possesses diopter adjustment from 0 D to 1.590 D.

When the moire pattern on the metal grid touch screen is simulated and analyzed by the sub-pixel transmittance calculating matrix of LCD, if the sub-pixel shape of LCD is non-standard non rectangular shape, the accuracy of algorithm prediction results will be low because of the large shape error. In order to solve this problem and provide the reliability of algorithm prediction results, this paper aims at a non rectangular sub-pixel LCD touch screen with specific shape, by improving the generation algorithm of sub-pixel shape, the computer simulation is used to generate sub-pixel shape graphics closer to the real shape, scan and calculate its transmittance matrix. Then, the obtained moire simulation pattern is processed by neighborhood filtering and low-pass filtering in the frequency domain to obtain the characteristic frequency reflecting the moire visibility. A moire visibility prediction table with metal mesh parameters as independent variables is generated, and the prediction results of moire visibility are given in combination with the judgment frequency obtained from the experiment. The experimental results show that when 2 lp·mm-1 is used as the judgment frequency, the overall matching rate of the improved algorithm is increased from 76.4% to 79.5%, and the available matching rate is increased from 67.57% to 84.51%. The accuracy and reliability of the algorithm are significantly improved, which can provide more reliable prediction results for the manufacturers of metal grid touch screen.

In order to effectively improve the electrical properties of flexible thin film transistors, HfO2 and Ta2O5 were used on polyimide (PI) substrates at room temperature. The stack structure combining two high dielectric constant materials replaces the single layer Ta2O5 as the gate dielectric to explore its effect on the device electrical properties. The thin films were prepared by magnetron sputtering, and the effects of Ta2O5 layers in the stacking gate dielectric structure on the device electrical properties under different sputtering time and different oxygen-argon ratio were investigated. The HfO2/Ta2O5 stacking gate dielectric devices were compared to Ta2O5 single-layer gate dielectric devices. The results show that when the sputtering time of Ta2O5 gate dielectric layer is 1 h and the ratio of oxygen to argon is 10∶90, the electrical performance of the device is the best. Moreover, the introduction of the stacking gate dielectric structure significantly improves the electrical performance of the device with the Ion/Ioff ratio of 1.27×106, the threshold voltage of 9.1 V, the subthreshold slope of 0.54 V/decade and the carrier mobility of 7.03 cm2/(V·s).

In the common ITO manufacturing process of TFT(Thin Film Transistor),the leakage current(Ioff) of some samples is abnormally high, and these abnormal samples are subjected to the same photo-resist strip equipment. This anomaly causes the strip equipment to suspend production, resulting in a loss of mass production. It is urgent to identify the reason and solve this problem for productivity and quality assurance. Firstly, the stripper of abnormal and normal equipment are collected and analyzed, it is found that the Al ion content in the stripper corresponding to the abnormal equipment was high. Secondly, it is found that Ioff will increase with the increase of stripping time in abnormal equipment. Then, it is proposed that Al ions in stripper are converted into Al2O3 particles adhered to the passivation(PVX) layer of TFT device during the stripping process, forming a parasitic gate effect which causes Ioff to be high. Finally, a series of solutions are proposed based on TRIZ, and then verified. Experimental results show that the Al ion concentration in the stripper increases from 1×10-8 to 2.189×10-6, and the Ioff increases from 3.56 pA to 7.56 pA. Once the stripper contains Al ions, Ioff increases as the number of stripping increases. The increase of Ioff can be suppressed by two measures: enhancement of the plasma treatment power before the PVX depositon and the increase of PVX thickness. In summary, it can be determined that the abnormally high Ioff is caused by the parasitic gate effect formed by Al ion in stripper solution, which can be suppressed by strengthening plasma treatment before PVX deposition and the increase PVX thickness.

In order to realize the automatic detection of LCD characters efficiently and accurately, a character defect detection algorithm based on multi-feature moment fusion was proposed for the disconnection and brightness defects of LCD character display. Firstly, the ROI region of the original LCD image was extracted, Hu invariant moments were used to describe the structural characteristics of characters, and Zernike moments were used to compensate for the information of higher moments that Hu moments could not describe. The color moment was used to describe color features of the characters,and the gray moment was used to make up the gray information that the color moment can not describe. The matrices were fused using 2DPCA technology. The similarity between the standard image fusion characteristic matrix and the image fusion characteristic matrix to be detected were measured by Euclidean distance. By setting the similarity threshold, the goal of defect detection was achieved. Experiment results show that the algorithm is stable and practical subjectively. Objectively, compared with the similar algorithms, the algorithm has a lower false judgment rate of 1%, a lower misjudgment rate of 0% and a higher efficiency of 0.6 s, basically meeting the actual detection needs.

With the application of generative adversarial networks in the field of image super-resolution (SR)?, some perception-driven SR methods can recover SR images with richer texture details, effectively alleviating the over-smoothing problem of the PSNR dominated SR methods. Gradient information is an important representation of image texture. However, few SR methods can make use of this information accurately and efficiently. In this paper, a gradient-aware single image super-resolution(GASR) is proposed, using gradient information better from two aspects. On the one hand, the feature map of gradient domain is used as the convolution kernel imposing on the feature map of image domain, which can effectively avoid the domain conflict caused by the concatenation of feature map of different domains. On the other hand, by elaborating the network details such as convolution kernel size, etc., the image fields output at the corresponding positions of the two branches are consistent with the feel fields of the feature maps in the gradient domain. In addition, the proposed GASR algorithm also effectively reduces the number of parameters and the amount of computation due to the increased demand for network lightweight in practical applications. Compared to SPSR, GASR can achieve the same performance at the cost of about 1/6 of the parameters and 1/10 of the computation of SPSR. On Set14 dataset, LPIPS and PSNR increase by 0.002 2 and 0.217, respectively. The experimental results show that GASR can achieve a good trade-off between texture details and image smoothness. In addition, GASR can not only reconstruct high fidelity SR image, but also alleviate the generation of messy textures.

The densely sampled light field can provide immersive visual experience and have extensive applications in areas such as 3D reconstruction and virtual reality. Light field reconstruction technology can obtain densely sampled light fields from sparsely sampled light fields, but when the baseline of the light field image increases, it will cause occlusion on edges and textured duplicate areas blur in the reconstructed view. In this paper, we propose a light field reconstruction network based on spatial-anglular decouple and fuse, it can make full use of the depth information and spatial information contained in the light field image and solve the reconstruction problems of the complex regions. The multi-branch input is used to supply the multi-directional information of the light field to solve the occlusion problem. In each branch, multi-scale features are extracted by atrous spatial pyramid pooling module to obtain disparity information and context information for the large receptive fields. The spatial-angular decouple and fuse module is designed to explore the spatial information, guiding the light field angle super-resolution reconstruction so as to obtain the reconstruction view with clearer texture duplicate region. Experimental results show that the average PSNR and SSIM reached 39.73 dB and 0.965 on the HCI, HCI old and 30 scenes datasets,showing high accuracy and outperforming the compared method.

The 3D object detection algorithm is constrained by the large amount of point cloud data, and can not achieve the balance of real-time speed and accuracy. This paper presents an improved 3D object detection algorithm—Pillar RCNN. The algorithm firstly divides the target point cloud space into voxels, presents a 3D object detection backbone network based on sparse convolution which gradually converts voxels into column voxels, quantifies 3D information into dense 2D information, and then processes the dense 2D information through the 2D backbone network. At the same time, the voxel features of different scales in the 3D backbone network and the 2D backbone network detection results are cascaded through the multiscale voxel feature aggregation module, and the result is further refined by the loss function. The algorithm is tested on KITTI public datasets and has a recognition speed of 2.48 ms on RTX 2080Ti hardware platform. Compared with the PointPillars benchmark algorithm, the performance indicators of the three categories of car, pedestrian and bicycle are improved. The detection results of the mode of bicycle and the hard of car are improved by 13.34% and 8.85%, and the detectation accuracies of other categories are improved also, achieving a balance between speed and accuracy.

In view of the significant intra-class differences between different cameras, and the change of the clothes of some pedestrians, which leads to the degradation of recognition, an end-to-end combination of generative adversarial network (GAN) data generation and person re-identification learning is proposed, which makes up the federated learning framework. Firstly, on the basis of the DGNet network, the residual block (ResBlock) in the generator and discriminator is replaced with DenseBlock, which strengthens feature propagation and avoids the problem of gradient disappearance. Then, the generative module generates high-quality synthetic images by switching appearance and structure encodings. Finally, a new normalization-based attention module (NAM) is added to the discriminative module to suppress less salient weights and pay more attention to the desired target regions. The images generated by the generation module are fed back online to the identification module to identify the true and fake images, and the results are fed back to the person re-identification module for classification and identification (the identification module and the pedestrian re-identification module are shared). The rank-1/mAP on Market-1501 and DukeMTMC-reID datasets reach 95.7%/88.6% and 87.1%/75.7%, respectively.