Space-mapping based image panorama stitching algorithm is proposed to solve the problems that the results of the traditional image panorama stitching algorithms have low accuracy and obvious seams. At first, the image distortion center and lens distortion parameters from each camera are measured, and the ideal image with specified size of field of view is obtained by image distortion correction processing. And then, the actual installation location of each camera is calculated by using the similarity of overlapping field region of view of the adjacent cameras. The accuracy of image stitching is improved and image stitching seams is removed using brightness uniformity processing and boundary fusion methods. At last, according to actual installation location of each camera, the field of view and image resolution, panorama image is obtained by calculating image stitching table. Experimental results show that the algorithm has high stitching accuracy and the stitching images are seamless and smooth.