Research And Implementation Of Fusion Calibration And Matching Images Of Algorithm

Today, automation precision equipment and mobile handsets continue to develop and become widely popular. In order to further satisfy human perception at subjective and objective video quality, some techniques and algorithms associated with image processing become a hot research and focus. In particular, with respect to which the image chromatic aberration correction, image fusion, features matching and other technologies in the development process of many devices have been widely involved. In this thesis,an algorithm of image correction and feature matching is proposed in the development process of wide format scanner, an algorithm of scene change is proposed before video coding and a improved algorithm of video stabilization and related applications are proposed and researched. Specifically, the following three components.(1) The algorithm of chromatic aberration correction based wide format scanner,solving chromatic aberration inconsistency due to images captured by independent sensors. The algorithm of image fusion based wide format scanner, computing the mean of the edge pixels for fusioning, through the overlapping area at the edge and the method of window mathching. Further, the engineering drawings need to eliminate the bluebackground to retain more details.(2) An algorithm of scene change based on hierarchical threshold of scene feature is proposed. In order to reduce the time for detecting scene change position based on the original algorithm, calculating different threshold based on different scene features to select the scene change position, before coding video.(3) An improved algorithm of video stabilization based on corner character and adaptive cropping the edge of window is proposed. Fist, the strong corner feature should be extract in a shaking video sequence. Second, in order to calculate the affine transform parameters, the above strong corners should be filter by the optical flow vector. Finally,in order to play video smoothly, the frames should use affine transform by the affine transform parameters which should be smooth based on frames per second. In addition,because the edge of frames which have been affine transform maybe is blank. According to the affine transform parameters and the center of frame is unchanged, calculating the max rectangle region in the transformed frame. At last, selecting the minimum region at the above all max rectangle region to decide the size of global video frames.Research on fusion, image correction and feature matching the above, from the test results, chromatic aberration correction and matching fusion algorithm to solve image problems and eliminate blue image of engineering drawings based on wide format scanner; scene change algorithm to detect a scene change position, and the time complexity is reduced; adaptive cutting edge of stabilization algorithm plays smoother video, and the information as much as possible, have a good performance. Work of this paper is desirable to provide some useful and practical reference image and video processing applications.