Abstract:

        This paper presents a new image mosaic algorithm for the ghosting phenomenon in image merging of smart car security system. We initially extract characteristic points from an image with SURF algorithm. We then match them and transform the image. We eventually merge the overlap area with a nonlinear weighted fusion algorithm. We set less weights for the border of the overlap area, and greater weights for the center of the overlap area to achieve the smooth transition between two adjacent images. On the other hand, the overall brightness of a merged image will not be smooth if the brightness of the image is different. Brightness transform is achieved in color space lαβ by introduction of brightness unification process and space transform for an image. Simulation results show that the algorithm can effectively eliminate patchwork and ghosting and guarantee consistency of the brightness of an image.

Key words: image mosaic, brightness unification, SURF, weighted fusion