Research Of The Image Super-resolution Reconstruction Algorithm

Array detector is receiving device in Modern optoelectronic imaging systems. The limited spatial resolution of array detector has become one of the main factors for improving of current photoelectric system performance index. In the conditions of optical system parameters and detectors invariability, one way of improving the optoelectronics imaging systems distinguishing ability and identify probability is improving sampling frequency of the photoelectric detecting system and then realize the super distinguish reconstruction of images.The basic idea is that in the premise of existing imaging system, we take certain software technology approach and overcome the ultimate of imaging, meanwhile obtain higher spatial resolution of the image.This paper center on the algorithm about super-resolution reconstruction of images, and introduced the theoretical studies about super-resolution reconstruction of images chiefly (including theoretical models of super-resolution reconstruction of images, relations between the characteristics of image sequences and the quality of reconstruction image). On the basis of this, the paper analyses the theoretical maximum conditions for reconstruction super distinguish image. We propose a method which used to obtaining complementary information of sub-pixel which based on micro-scanning device of double wedge. The experiments verify that the device has realized the sub-pixel displacement. Compare all kinds of motion estimation algorithms, and focus on the actual images, the paper proposes a multi-resolution motion estimation algorithm based on the optical flow. Simulation experiment using images from laboratory verifies the method reaches the precision of sub-pixel. Further research on sequential super-resolution images, the paper improves the projection onto convex sets (POCS). Reconstructed the sequential experimental images show the new algorithm is effective, and highlight the details of the edge, improve the resolution of single image. At last, the paper analyses the evaluation methodology for super-resolution images from subjective and objective, propose a combine evaluation methodology. The experiment results show that the method evaluates the reconstructed super-resolution images accurately.