In recent years, organic light-emitting diode (OLED) displays have gradually become the mainstream display technology due to their self-luminescence, fast response speed, lightness and thinness, and ease of flexible preparation. Flexible OLED and printed OLED are two key development directions of OLED display technology. Flexible OLED has been initially used in high-end mobile phones, but there are still many problems to be solved. Printed OLED has attracted the attention and development of many companies and scientific researchers due to its cost-saving features. Although the printed OLED technology is not yet mature, it is developing very rapidly and industrialization is just around the corner. Aiming at the technical problems of flexible OLED and printed OLED, this article reviews the research progress of flexible OLED and printed OLED, and looks forward to its future development trend.

In order to improve the security and reduce the leakage of information, and further broaden the application fields of information privacy, this work reported a novel information cheating technology based on a light-driven bimorph soft actuator. Firstly, the as-prepared gold nanorods with different aspect ratios were uniformly dispersed into the polypropylene/polyimide composite membrane by means of mask depositing and coating, and then a photothermal biomorph soft actuator was fabricated, after that, the obtained bimorph soft actuator was selectively cut. Herein, we defined the state presented by the bimorph soft actuator without external light stimulation as the first information; only when triggered by the corresponding laser, this soft actuator could display second type of information, which we defined as the second false information; since the highest photothermal conversion of gold nanorods only occurred under the laser with corresponding wavelength, the temperature of different areas observed by the infrared camera showed the new information, which played an important role in information cheating. The experimental results showed that the soft actuator kept in flat state without illumination, and the cutting area of bimorph soft actuators reached a bending angle of at least 50° under the stimulation of corresponding laser, thus the displayed information can be clearly observed by the naked eye; while the difference of photothermal area and non-photothermal area can only be observed in the infrared camera, where the gap of temperature was at least 10 ℃, and the displayed true information could be clearly seen. The combination of the soft actuators and the information cheating is novel and effective, which provides a new strategy toward development and future of information security issues.

In this paper, a flexible curved holographic grating is fabricated by wrapping a polymer dispersed liquid crystal film in a polyethylene terephthalate (PET) film with a fixed radius of curvature, and using plane wave interference holographic technology.The article studies the diffraction and reproduction characteristics of flexible curved holographic gratings.The article use the microelement method and coupled wave theory to derive the diffraction characteristic formula of the curved flexible grating when the curvature changes.A flexible holographic grating with a cylindrical wave to reproduce the curvature change is proposed to achieve the best reproduction of the diffraction effect. The paper deduces the relevant technical parameters, such as the focal position of the cylindrical wave, the focal length of the cylindrical lens and the relationship between the lateral magnification of diffraction and so on. A flexible grating with a central diffraction efficiency of 88% was prepared under the conditions of an exposure time of 70 s and an exposure intensity of 30 mW/ cm2. The experiment achieved the theoretically calculated lateral magnification characteristics by changing the radius of curvature of the flexible grating and using cylindrical wave reproduction.

Based on the ablation mechanism of femtosecond laser, the theoretical model of ITO area blackening method for repairing bright spots was established and verified through experiments, and in order to improve the success ratio of repair, the influence of various process parameters were also analyzed on the remediation effect. In this work, the repair mode was firstly established by a series of experiments with light detection test, microscope images and Focused Ion Beam(FIB) images. Then, process parameters that effect success ratio were analyzed by contrast experiments, and these parameters included the energy density per unit area of single-pulse laser, scanning speed, the laser frequency and sport size. Finally, in order to deeply analyze the ablation mechanism of ITO area blackening method, the variable “w” containing all process parameters was proposed and verified, and it successfully provides a theoretical basis for improving the success rate. Experimental results indicated that, when the spot size was 4.0~10.0 μm, energy density per unit area of single-pulse laser was 1.0 mJ/mm2 or more, and the “w” was 4 250 ~ 12 500 mJ/mm2, the blackening effect of ITO after repairing was up to the dark point standard and other undesirable phenomenon was not produced. According to the experimental result, the theoretical model and process parameter analysis was applied to mass production, the success rate of bright spot defective repair reached to 95.5%.

In four mask process, it is extremely critical to control the photoresist remain of halftone zone for the morphology and electrical properties of TFT. Therefore, this study focused on the key factors of photoresist remain in four mask process based on oxide-TFT technology. Firstly, the optimal conditions of soft-bake temperature, vacuum drying time and developing time are obtained via full factorial experiment. Furthermore, the ideal quantity of exposure is evaluated through single factor test. The results show that the experiment design could effectively optimize the remain and uniformity of photoresist, so that the ideal experiment conditions aere obtained. When the soft bake temperature, vacuum drying time, developing time and exposure are 115 ℃, 10 s, 52 s and 67 mJ/cm2, respectively, the PR remain and CD are 0.51 μm and 11.17 μm, which exhibites good uniformity (<5%).

After the sensor photo process is completed for SLOC products, macro observation shows that there is a large area of Mura on the color filter chamfered side of the substrate. The critical dimension (CD) value of the Mura area and the normal area is tested. The CD value of the Mura area is significantly larger, and some points exceed spec. In addition, the average incidence of Short in the module section of the SLOC product in the production process is 2.82%, and the location of the Short is basically consistent with the larger areas of Mura and CD, and has a strong correlation. Through tests such as spin coating and actual tape-out observation, the device that caused Mura to be locked is the developer. As the substrate from the DEV#1 to the DEV#2 is dried at the end of the substrate, the liquid solution is dried, which leads to poor development, which in turn causes Mura. This article refers to this as dry Mura. In order to solve this problem, the speed of substrate transferd from DEV#1 to DEV#2 is modified by software. The drying Mura is slightly improved but not eliminated. Later, by adding a secondary aqua knife method in the DEV#1 chamber, the dry Mura of SLOC sensor photo is eliminated. The CD uniformity of SLOC products increases from 3.3% to 1.9%, and the short defect rate decreases from an average of 2.82% before improvement to an average of 0.27%.

A simple design and small size liquid crystal drive circuit is proposed in this paper. The circuit includes an RC Schmitt oscillation circuit and an inverter. Through a specific connection method, the function of driving a liquid crystal device using only a DC signal within a certain voltage range is realized. Through theoretical analysis and experimental test, the influence of component values on the output waveform characteristics of the circuit is discussed. The test results show that the inverter circuit has a response speed of less than 0.4 μs. The power consumption is less than 1 mW when the input voltage is below 3 V. The input and output voltage have a good linear relationship, which can realize precise control of the transmittance of the liquid crystal device. This research provides new technical ideas for driving liquid crystal devices in miniaturized smart wearable devices.

In order to solve the signal attenuation through channel at high data rate, a 1.5 GHz adaptive equalizer using error feedback at Rx(Receiver) is introduced. An equalizer based on discrete signal process is applied to reduce convergence time. Firstly, PRBS used as the training data will be transported to the equalizer filter for high frequency compensation. Then it is sampled and sliced by the S/H circuit and slicer to obtain steep rising and falling characteristics. Finally, it is sent to a comparator which is different with conventional comparator. There is an internal register in proposed comparator to store input signal. Error between input and desired value will be fed back through the closed loop to increase equalizer boosting gain. After several times iteration, useable gains are acquired and the median is selected as the optimal gain. If no suitable gain is found, Tx(Transmitter) will increase pre-emphasis setting and adaptive iteration progress at Rx repeats. Finally, the setting gain will be sent to inform Tx. The compensation range is from 1 to 16 dB. An experiment at 165.1 cm(65 in) 8 K module using CSPI protocol is carried out, where the validity of the proposed design is verified by feedback setting at Tx. The requirement that CSPI protocol data could be received and recovered without distortion by Rx at high data rate is met.

The LED smart glass display design is proposed, which combines Android embedded technology and Internet technology, realizes the real-time interaction between people and screen. This system adopts CPU+FPGA dual-core architecture equipped with Android operating system. Through the development of Android system application software, the wireless network is used to realize the communication between the LED smart glass display and the mobile terminal. Based on the Android system source code, this system adds the LCD display driver and mapping data transmission driver using the “bus-device-driver” model. Device tree is used to describe device information, which reduces a lot of code redundancy. JNI method is written in Android system for Android application program to call. Java language is used for Android system application software development. Socket network programming is used to realize the network communication between LED smart glass display and mobile terminal. This system has good ease of use and portability, which brings convenience to product application and improves construction efficiency.

Skin color detection is widely used in face detection, gesture recognition, sensitive image filtering, medical diagnosis, image enhancement and so on. It is an important technology for computer vision and image processing fields. Because human skin color is easily affected by lighting conditions, races, etc., it is difficult to achieve ideal detection results under unconstrained environments. The emergence of adaptive methods makes the above problems improved. In this paper, adaptive skin detection methods are comprehensively reviewed. The paper focuses on the analysis of adaptive skin detection methods based on parameter dynamic adjustment and high-level semantic features. Meanwhile, it discusses their respective advantages and disadvantages. On this basis, the future development trend is prospected, which aims at providing reference for related researchers.

In order to solve the current problem about low-light image enhancement,an algorithm of image enhancement based on dual-residual convolutional network is proposed. First,according to Retinex theory, the normal-light image is synthesized into low-light image, the synthetic is decomposed onto the three components of R、G and B,and learning the mapping relations between low-light image and normal-light image on all components through the module of feature extraction as well as dual-residual. Then the enhanced image on all components can be obtained, and finally the enhanced RGB image is synthesized. Subsequently, the bilateral filtering is used to optimize the enhanced RGB image so that the obtained image is analogical to the reference image.The experiment results show that the algorithm proposed in this paper, whose Peak Signal to Noise Ratio can reach up to 25.931 1 dB and Structural Similarity Index can reach up to 0.945 2 in terms of processing synthesized low-light image,and whose novel blind image quality assessment can exceed other compared algorithms and the algorithm in this paper goes faster in terms of processing real low-light image.Therefore,the proposed algorithm is superior to the contrast algorithms.

The traditional CNN-based method can not distinguish between low-frequency structural information and high-frequency detailed information in the process of reconstructing low-resolution images. And there is a lack of information communication between layers of the network, which leads to the problem of missing information in the high-resolution reconstruction image. In order to obtain more information about the structure and details of each level of image features a residual dense network is constructed based on the wavelet domain (WRDSR). In the wavelet domain formed by the two-dimensional discrete wavelet transform, the network uses dense connections and residual connections to fully extract the information of different frequencies of the image, and then generates high-resolution image wavelets by inputting the fused features into sub-pixel convolution. Finally, the final high-resolution image is generated by two-dimensional discrete wavelet inversion. Compared with Bicubic, A+, SRCNN, VDSR, LapSRN, DWSR,SDSR etc., WRDSR improves 2.824 dB/0.059 5, 0.896 dB/0.018 2, 0.747 dB/0.016 8, 0.016 dB/0.002 4, 0.025 dB/0.004 3, 0.21 dB/0.004 7 and 0.20 dB/0.0057 on average on PSNR/SSIM, respectively. While making more efficient use of the original image information, WRDSR solves the drawback of missing information, making the reconstructed image texture clearer, richer in details and better in visual effect.

In order to restrain the effect of CMOS image sensor fixed pattern noise(FPN) caused by the gain mismatch of readout amplifier circuit, to improve the quality of imaging, a method of fixed pattern noise correction based on the dark current is proposed in this paper. Compared with the traditional correction methods, this method has simple correction system, and the algorithm is easy to be implemented. The images before and after correction were evaluated via the grayscale variance, and the evaluation results show that in the background image formed by dark current, the grayscale variance of after corrected image reduced by 4 orders of magnitude compared with that before correction. As for the CMOS image under uniform illumination condition, the grayscale variance of after corrected image increased by 2 orders of magnitude compared with that before correction. The evaluation results and related conclusions are useful for the future study of dark current background images and correlation correction methods.

Due to the obvious intra-class differences caused by camera perspective, many researchers begin to use GAN to expand data to maintain intra-class invariance. Nevertheless, the images generated by existing GAN have some defects such as blurred image and unreal background. To solve the above existing problems, a person re-identification algorithm that uses mixed coding of posture and appearance features is proposed in this paper. In the training phase, the image of the person is decomposed into posture features and appearance features, and the generated network can generate high-quality images by switching the appearance feature and posture feature and then mixing the features in the two images. The discriminant network feeds back the appearance characteristics of the generated image to the appearance coder of the generated network to achieve joint optimization, and uses multi-loss joint to further improve the quality of the generated image. In the testing phase, the network model is tested using the original data set. The rank-1/mAP on the Market-1501 and DukeMTMC-reID data sets can reach 93.4%/82.2% and 84.3%/70.5% respectively.