基于视觉特性的多聚焦图像融合

本文提出了一种考虑人眼视觉系统特性的多聚焦图像融合算法.融合过程是首先将配准的源图像分割成若干个块,计算出每个块的对比度方差,作为图像均匀度参数,通过选取两幅图像中清晰的图像块形成融合图像.文中讨论了分解图像块的大小和阈值参数对融合性能的影响.实验结果表明本文提出的算法实时性好,对于严格配准的多聚焦图像能够达到甚至超过基于小波分解的融合算法.

Feature of Human Vision System Based Multi-Focus Image Fusion

In this paper,a multi focus image fusion algorithm is proposed which takes features of human vision system (HVS) into account.First,registered original images are divided into several blocks.Then contrast error of every block is calculated considering contrast masking sensitivity.Finally,the contrast errors of corresponding blocks are compared to decode which should be used to construct the fused image.The influences of two parameters,size of decomposed block and threshold,on fusion performance are discussed.Experimental results show that the proposed method can operate real time.To stringent registered multifocus images,the proposed algorithm can achieve equivalent or superior performance to the algorithm based on wavelet decomposition.