The Fusion Effectiveness Assessment Of Color Images Based On Structural Similarity And Contrast

The quality assessment of fused images plays an important role in image fusion. Via thequality assessment of fused images, the comparisons on different image fusion algorithms orthe same image fusion algorithms with different parameters can be made, which helps theselection of the optimal algorithms or parameters and facilitates the further researches andanalyses of fused images. The quality assessment of color fused images is a relatively newdevelopment direction, which is derived on the basis of the quality assessment of the originalgray-scale fused images. As the color images have better identifying characteristics for humanvisual system, a better assessment result which is consistent with human subjective perceptioncan be easily got. The research on the quality assessment of color fused images is animportant research direction. The color fused images taking the color fused source images asthe reference images tend to retaining as much useful information of the color source imagesas possible. In this paper, a systematic research is made on the color fused images.In this paper, based on human visual characteristics, a quality assessment algorithmbased on contrast sensitivity weighted structural similarity is proposed for color fused images.And then taking into account of the noise in the color image fusion process, a four-parameterquality assessment algorithm for color fused images is proposed. The main research contentsin this paper are as follows:(1) The research status of fused images at home and abroad is described. On the basis ofthe analysis on the quality assessment of the fused images at home and abroad, a detailedintroduction on the currently using quality assessment of fused images is made. A summaryon the development of image fusion method at home and abroad is also made.(2) In allusion to the performance of color fused images on human visual characteristics,a quality assessment algorithm for color fused images based on color contrast sensitivitycharacteristics weighted structural similarity is proposed, after the analysis of the features ofthe existing image fusion quality assessment method. In this method, the luminancecomponent and chrominance component of the color images are separated; the structuralsimilarity of the luminance component is calculated based on contrast sensitivity characteristics weighting; then the structural similarity of the color image is obtained with theweighting with the structural similarity of the chrominance component. Taking into fullaccount of the human visual characteristics and the color feature, the assessment results of thedifferent image fusion assessment method show that the quality assessment of the color fusedimages can be better completed by this method.(3) Then noise information in the process of color image fusion is considered. Afour-parameter based method is proposed for the quality assessment of color fusion images,which takes the average mutual information, the average contrast entropy, grayscale/chromaticity standard deviation ratio and information preservation of color fused images asthe four-parameter.