基于Contourlet变换的图像超分辨率重建算法研究-模式识别与智能系统专业论文.docxVIP

Abstract Abstract III Abstract With the development of modem technology,higher image quality is required in more aspects of life．In order to meet the requirement of application，people tend to work on certain algorithms wim the purpose of achieving high quality image instead of improving the device with 11igh costs．Research on image super-resolution recon- struction(SRR)，which is an image—enhancing approach aiming to reconstruct a high-resolution(HR)image from observed multiple low-resolution(LR)images，has been done recently．Nowadays，SRR techniques have been widely used in many aspects oflife． As an efficient set of techniques in signal processing，wavelet transform enjoys good local features in bom time domain and frequency domain of signal representation． Accordingly,the multi-scale analysis tools，which enjoy better quality,have been developed．Scholars have already put forth algorithms on wavelet—based SRR,and have achieved good performance in signal construction．Inspired by the ideas above，this paper introduces the multi—scale analysis tools into SRR technique，and more specifically describing，investigates the SRR algorithms based on Contourlet transform． ne main work related includes the contents as follows： (1)A SSR method for single-frame image is proposed，which is based on Contourlet transform combined、Ⅳim Soft—decision adaptive interpolation(shi) algorithm．The algorithm takes advantage of edge and detail capture features via Contourlet transform and details preserving ability in enhancement via SAI． Experimental simulation indicates that the proposed algorithm boosts quite good performance．And then，based on the analysis of the method above，an improved algorithm is proposed，谢tll feasibility proofvia experimental results． (2)A SSR method for sequential images is proposed in the paper，which is based on Contourlet transform combined with Projection Onto Convex Set(POCS)algorithm． POCS algorithm is classical in sequential image reconstruction techniques，whose ad