Study of Color Balance for Remote Sensing Imagery Mosaic

Abstract

Mosaicking is a process to stitch certain number of images together into a seamless complex one. The studies of Image Mosaicking usually about three aspects: Image Registration, Color Balancing and Seam-line Generation. This paper mainly talks about the color balancing between images to be stitched together. With deeply studies of two color balancing algorithms, histogram matching and gain compensation, and analysis of the influences of cloud and other special cultures in the images on color balancing, the paper improve the two algorithms. Improved algorithms do color balancing excluding the influencing factors, so that the complex image's color is more uniform. Using Environment Satellite 1 (HJ-1) data to do experiments to inspect and verify the advantage and practicability of the two algorithms. Experiments show that the color balancing algorithms excluding clouds and special cultures can make images well- proportioned mosaic, and keep original definition. In addition, this paper analyses the color balancing problems of large-area mosaics. According to nationwide image mosaicking, based on improved color balancing algorithms, a color balancing strategy which works from global to local areas is proposed. This strategy have basic effect to ensure the image classification, culture extraction, change detection etc processing carry on better, and benefit the environmental assessment and land monitoring over large areas.