基于Snake活动轮廓模型的视频跟踪分割方法

A video object segmentation algorithm based on Snake models

基于Snake活动轮廓模型,采用时空融合的方式,根据短时间内相邻帧的运动趋势差异相似的前提,首先将视频序列分成若干个小段,每段有k帧视频,选取段内的前两帧为关键帧,通过运动检测的方式自动得到这两帧中运动对象的大致区域;然后进行帧内Snake演变,搜索精确轮廓;最后以关键帧间运动对象形心的运动矢量预测勾勒后续帧的初始轮廓,进行帧内Snake精确轮廓定位,从而实现所有帧的视频对象分割。相比于传统方法,本文方法克服了手动绘制初始轮廓的缺点,在空域对Snake贪婪方法进行了改进而且精确度高,速度快。实验表明,本文方法成功地实现了前后帧图像之间运动对象的对应匹配关系,并通过改进后的Snake贪婪方法得到了精确的分割结果。

A new video object segmentation algorithm based on the improved greedy Snake model is proposed to solve the problem of object tracking. This algorithm combines temporal and spatial information together. Firstly,the video sequence can be divided into segments due to the fact that the movement trend of adjacent frames remains similar in a short period of time, and each segment has k frames;Secondly,the first two frames of each segment are recognized as key frames,and the rough contours of the moving object in the first two frames are acquired automatically by using the motion detection; Thirdly, the improved inter-frame greedy Snake iteration is applied to get the precise contour;Fourthly,the intraframe moving vectors of the moving object centers in the key frames are used to predicate the initial contours of the moving object in the subsequent frames; Fifthly, the improved inter-frame greedy Snake iteration is applied for the non-key frames to get the precise contours on the basis of the initial contours,and then the video object segmentation can be realized for all the frames. Compared with the traditional methods, the proposed algorithm overcomes the disadvantages of drawing the initial contour manually. Furthermore,the greedy Snake method in the spatial domain has been improved with high accuracy, speed and many other obvious advantages. Experimental results indicate that the new metihod realizes the corresponding match of adjacent moving objects and gains accurate segmentation results through the improved greedy Snake method.