Based On Light And Color Temperature Automatically Adjusts Image Stitching Algorithm

Recently, image-based modeling and rendering (IBMR) becomes a kind of widely used rending technology of Photo realistic scene. Its specialty is that the rending complexity is independent of the scene complexity. The Panorama mosaic is a new and useful technology of IBMR.This thesis compares IBMR with traditional 3D model-based rendering technology in the first. Then discusses concepts, processes, methods related to panorama mosaics. Moreover it reviews related researches in domestic and abroad.The photos taken with an Automatic Camera usually have large illumination difference. The characteristics make both image alignment and panorama building more difficult than using photos taken by cameras calibrated by special equipment. Based on studying many image mosaic algorithms, we present a better one that is able to accurately stitch two similar images automatically with the limitation of the photos.Based on above discussion, this thesis researches and implements the following image mosaic method based on Illumination and Color Temperature adjustment.The firstly is the matching of image corners and the reducing of the outliers by RANSAC algorithm. Secondly, MSAC and regression analysis are used to fetch the linear transform parameters of the three-color channels. Then images are blended. Experiments show that our algorithm is effective.