Multi-focus Image Fusion Technique Based On Wavelet Transform

Image fusion is a new research field in image processing,which analyzes the characteristics of each fusing images,and then fuses them by appropriate strategies. Finally it can get a new image which accords with human beings' and computers' vision and is adapt to further image processing.Multi-focus image fusion is fusing several images which are got by each target focused separately.The purpose is getting a clear image,in which each target is distinct.This technology can effectively improve utilization ratio of image information and the identification of the target detection system reliability.So it is widely used in machine vision,digital camera and target recognition.A comprehensive research on multi-focus image fusion based on wavelet transform is made in this dissertation,which mainly includes the following contents:First of all,we introduce the development of image fusion and wavelet transform and the application of wavelet transform in image processing.Then several kinds of present conventional fusion algorithms and evaluation methods are listed and compared.On the basis of these analyses,two algorithms are proposed,that is,the algorithm of multi-focus image fusion based on wavelet packed and round-neighbor characteristics and the adaptive algorithm of multi-focus image fusion.The former adopts wavelet packed instead of orthogonal wavelet to decompose images.And considering the neighbor dependency,it fuses the coefficients by the strategy based on region eigenvalue.In this paper,we replace the traditional square-neighbor by the round-neighbor,which can reduce time complexity obviously.It is proved by the experiments that the algorithm can get better results,especially when fusing images which contain lots of details.The latter emphasize particularly on better vision impression.It fuses different kinds of coefficients by different methods.It is proved that this algorithm can get better vision impression.At last,the methods for mult9-focus image fusion proposed in this paper are tested respectively.Experimental results show that these methods often obtain better fusion performance than the existing typical methods.