摘要
针对运用单一颜色特征描述运动目标时抗干扰性较差的问题,提出一种融合灰度共生矩阵和颜色特征的Mean-Shift目标跟踪算法.采用灰度共生矩阵推导的6个纹理特征参数和颜色特征分别表征跟踪目标,引入马氏距离计算纹理特征的相似度,并结合Bhattacharyya系数计算颜色特征的相似度,同时利用Mean Shift算法进行目标定位.实验表明,改进算法能在复杂背景下,有效、准确地实现目标跟踪.
Aimed at the problem that there is poor anti-jamming ability when the single color feature is used to describe the moving target, a target tracking algorithm was proposed based on Mean Shift which mixed gray level co-occurrence matrix （GLCM） and color feature together. Six parameters of texture de- duced with GLCM and color feature parameters were used to characterize the tracking target. The similari- ty of texture feature was calculated by introducing Mahalanobis distance, and the Bhattacharyya coefficient was applied to calculate color feature similarity. Meantime, Mean Shift algorithm was used for target posi- tioning. Experimental result showed that the improved algorithm could be used to track moving target ef- fectively and accurately on a complex background.
出处
《兰州理工大学学报》
CAS
北大核心
2013年第4期89-93,共5页
Journal of Lanzhou University of Technology
