The Research For The Super-Resolution Restoration Based On The Image Sequence With Displacement

Super-resolution(SR) reconstruction basing on the image sequence is atechnique for estimating a high-resolution image from an image sequencehaving slight displacement. It breaks through the resolution limit of imageacquisition equipment and can achieve data fusion on subpixel level. Image SRprocessing has proved to be useful in many practical applications, such asremote sensing, military detection, medical imaging and public security, etc. Thesubpixel technique is using a number of CCD to take photographs on the sametarget and acquire the image sequence with displacement on subpixel level. Ithas become the important embranchment of SR restoration because it cangreatly improve spatial resolution after reconstruction.The thesis introduces the background and development status of ourresearch firstly. The second step, many SR reconstruction algorithms areclassified and discussed. And then, the modulation transfer function of discretesampling process of CCD sensor is analyzed and the calculating formula ofaverage MTF of CCD sensor in subpixel imaging system is acquired. Comparedthe average MTF of general imaging and subpixel imaging system, the specialresolution of the image acquired by subpixel imaging system is higher than thegeneral one and it also shows that the imaging quality with multiple CCD isbetter than that of general imaging when the spatial frequency is lower than theNyquist frequency of the CCD sensor. It has theoretical meaning to the subpixelimaging technique with multiple CCD. The interpolation algorithms of imagesequence and a new kind of algorism which basing on wavelet and fractal theoryis presented. The image sequence with displacement is reconstructed by thedifferent interpolation algorithms. The results of experiment and computersimulations show that the new algorithm is feasible and efficient both onsubjective way and objective way. Finally, the summary and prospects of thethesis are given. The theoretical result has been acquired and it can providesome experience for future research of subpixel imaging technique in SRreconstruction.