An Approach of Color Image Mosaicking Based on Color Vision Characteristics

Based on the characteristics of color space transform between RGB and HSI, an algorithm of color difference measurement for color image is presented. The degree of scatter about color space distribution of a given image region is determined by computing the sum of color difference between every pixel and the other pixels of the image region. Then an effective feature template, which is then used in color image matching, is chosen according to the degree value. Furthermore, template matching is applied to obtain relative displacement between two images to mosaicked. A function of brightness transform is built by Least Mean Square (LMS). According to the function built, brightness of each pixel in one color image is adjusted when necessary, and then the brightness difference between two images can be eliminated. Experimental results show that the proposed approach has good precision of image mosaicking and capacity of anti-jamming for brightness.