| With the development of multimedia technology and computer technology, Image processing technology has been widely used in the domain of military, civil application and industry. Target detecting and tracking which is based on video is one of the significant research field of computer vision, and in this domain, the technology of detecting and tracking small target has become more and more ripe.In this thesis, some research has been done on the small moving target under the background of sea level. The video is under background of sea level where lots of noises exist, in this circumstance, in order to make perfect tracking to the target, we must reduce the influence of noise, and make some measurement to separate the target. The major task is to program under circumstance of Visual C++ on the aspect of detecting and tracking, and to do some research on bitmap correlative processing, and then program with emulation algorithm under condition of MATLAB, and compare the different algorithm and get a sight of the different effect, besides, this can also be used on video image.In the process of target detecting, this thesis proposes the moving targets detecting algorithm combining the method of Background Subtraction with the method of Symmetrical difference. The method of Background Subtraction can generate most of foreground image, but can't detect all the information of moving target, so the information that can't be detected can be got in the method of Symmetrical difference. As for the selection of threshold, this thesis proposes to use minimum error algorithm.In the process of target tracking, using a new algorithm which is named Mean Shift, it has many advantages, for example: real-time, robust for occlusion and target distortion, however, it has some defects, such as can't track fast moving target, so we combine this algorithm to Kalman filter. Kalman filter can predict the possible position on the next frame of the video image, and using the Mean Shift algorithm to search in this neighboring range, this method is effective on fast moving target, and it also makes sense to the target under the circumstance of low signal-noise rate.Simultaneously, through using the function of image processing in MATLAB, get one frame and then generate this bitmap, later, this bitmap is used in Visual C++ programming, such as get the result of this image after histogram equalizing or median filtering, and so on. There is function union of image processing in MATLAB, and it is good to make emulation to the image processing. In this thesis, also make research on bitmap processing, such as apart the line of sea level, apart the target which is near with the sea level, and morphologic processing. |
