Image Super-resolution Reconstruction System Design Based On Subpixel Displacement

In the field of improving the limit of image detector, the super resolution imaging technology is a hot spot. The image detector is widely used in various fields of production and life, and technology improvement has an increasing demand of image quality in various areas. However, the internal structure and external conditions of the imaging device(such as pixel size, optical system aberration, atmospheric disturbance, motion blur and defocus blur) will degrade the image quality therefore it can not meet the actual demand. The resolution of the imaging devices can not be improved by ways of hardware improvement because of the high economic costs and the unbreakable technical difficulties. The super-resolution technology, which has caused widespread concern the first time it was proposed, is designed to make up the hardware deficiencies through software. The super-resolution imaging technique of the sequence image is the focus of the current research. With several interlocking original low-resolution sequence images which can complement each other can be obtained an image with more information breaking the information limitation of the original image. The photoelectric imaging super-resolution technique has been studied systematically in this article.In this paper, the theoretical module of the super-resolution image reconstruction has been studied and the relationship between redundant information sub-pixel displacement sequence image and high-resolution image has been analyzed. Based on the study of the structure of the CCD image sensor, It can be analyzed the acquisition method of the sub-pixel displacement sequence image with redundant information. A micro-displacement device based on piezoelectric ceramic is constructed and the feasibility of the device is analyzed. Under the existing laboratory conditions, redundant information is acquainted from many displacement standard resolution board images. The super-resolution reconstruction algorithm of the sequence images has been studied. According to the characteristics of many micro-displacement images with redundant information, the solving linear equation method has been optimized. The sequence images collected in the laboratory are reconstructed. Results show that compared with the original algorithm the images reconstructed by the optimized algorithm are fine and prominent in edge detail, which means the reconstructed images are much better. The subjective and objective quality evaluation method of the super-resolution based reconstruction images has been researched. Various objective evaluation function and Fourier spectrum figures are adopted to evaluate the images, provide more specific information. Results showing that the method described in this paper is feasible.