Canny准则小波边缘检测在图像融合中的应用  

Application of wavelet edge detection based on Canny criteria in image fusion

作　　者：狄红卫[1] 张文琴[1] 

机构地区：[1]暨南大学光电工程研究所,广东广州510632

出　　处：《光电工程》2005年第6期79-82,92,共5页Opto-Electronic Engineering

基　　金：广东省自然科学基金资助项目(04010465)

摘　　要：提出了一种新的图像融合方法。该方法选择Canny准则作为边缘检测的基础,并结合小波变换算法,确定图像边缘位置。在小波域中,对高频信息依据其是否为边缘点采用不同的融合策略,对低频信息利用加权法进行融合,再进行小波逆变换重构融合图像。实验结果表明,提出的算法在抑制噪声的同时,能有效地突出边缘细节,更好地保持图像的空间分辨力。对于多聚焦图像的融合,偏差度为0.0520,熵为7.6609,相似度达到0.9985。A new image fusion method is proposed. This approach firstly determines the image edge positions based on Canny criteria and in combination with wavelet transform algorithm. In wavelet domain, different strategies are adopted for fusing high frequency coefficient according to whether it is a pixel of edge. Average-weighted mean is used for low frequency coefficient fusion and then the fusion image is reconstructed by wavelet inverse transform. Experimental results show that the method can not only inhibit noises but also effectively enhance the edge details and improve the spatial resolution of the image. For fusing multi-focusing image, the deviation index is 0.0520, the entropy is 7.6609 and similarity is 0.9985.

关 键 词：小波变换 边缘检测 图像融合 空间分辨力 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]