视频图像活动轮廓目标检测跟踪研究

研究视频图像目标跟踪定位精确度问题。由于在图像中通常会发生缩放,造成图像目标模糊不清。传统的目标跟踪算法该类算法仅以目标发生平移运动为假设前提,图像质量差。为解决上述问题,提出了一种活动轮廓目标跟踪定位检测算法。首先选择合适的滑窗,采用减背景法来确定视频对象的运动区域,采用卡尔曼形态滤波来消除残余的噪声,然后针对目标在活动轮廓局部内具有较高灰度值的特征,通过自适应阈值来判别滑窗中心位置是否存在目标。当滑窗遍历整幅图像后,就可以得到目标的定位结果。仿真结果表明,改进算法不仅能够消除差分图像中的显露背景,从而得到运动视频对象精确的轮廓,并且可进行多目标的分割与跟踪,具有一定的实际应用价值。

Improved Active Contour Algorithm for Target Tracking and Positioning

Study the problem of target tracking and positioning accuracy.The traditional algorithms are difficult to effectively meet the target,which usually occur in the image scaling,rotation and shearing,etc.The paper proposed a target tracking method based on active contour orientation detection algorithm.In this method,we selected the appropriate sliding window,used inter-frame difference to determine the video object motion regions,and used morphological filter to eliminate residual noise.Then,according to the features that target in activity profile has a high gray value,an adaptive threshold was used to determine the center of the sliding window target.When the sliding window traverse through the whole image,we can get the target location results.The simulation results show that the improved algorithm can not only eliminate the revealed background in difference image to obtain precise contours of moving video objects,but also carried out multi-target segmenting and tracking,and is of some practical value.