【摘要】 提出了一种采用主动轮廓模型来融合多种图像信息的运动目标提取算法.通过对含有运动信息的差分图像进行统计分析,可以获得运动目标的粗略位置,从而解决了主动轮廓的启动问题.利用图像中的色彩信息,构建了一种新的色彩力.由于这种色彩力在作用范围、定位精度上与反映目标边缘特征的图像力具有互补性,因此在这两种力的共同作用下,主动轮廓可以有效地解决其初始位置远离目标边缘的难题,并精确地提取目标的真实轮廓.实验表明,这种算法在长达数百帧的图像序列中,始终能准确地分割出运动目标,并稳定跟踪目标的运动.更多还原

【Abstract】 An automatic segmentation algorithm by means of active contour model to fuse multiinformation is proposed. The coarse position of moving targets could be obtained by statistically analyzing the difference image that contains motion information, thus the startup problem of active contour model is solved, and a new color force is constructed according to the image color information. This color force is complementary with image force corresponding to edge features of targets on the respects of action range and location accuracy. The common role of above two forces facilitates solving the problems caused by an intitial position of the active contour distant from targets’ edges, and then real contours of targets are detrected perfectly. The experiments confirm that this algorithm can segment the moving targets accurately, and track them robustly among the image sequence of hundreds of frames.更多还原