基于HSV非均匀量化的CAMshift目标跟踪算法  

Modified CAMshift algorithm for object tracking based on HSV non-uniform quantization

作　　者：张辰[1,2] 彭玉旭[1,2] 

机构地区：[1]长沙理工大学计算机与通信工程学院,长沙410114 [2]综合交通运输大数据智能处理湖南省重点实验室(长沙理工大学),长沙410114

出　　处：《智能计算机与应用》2017年第3期21-25,共5页Intelligent Computer and Applications

摘　　要：针对经典CAMshift(Continuously Adaptive Meanshift,连续自适应均值偏移算法)算法易受色度相似背景像素干扰的问题,提出了基于HSV非均匀量化的CAMshift目标跟踪算法,有效地解决了经典算法存在的缺陷。通过在经典CAMshift算法颜色直方图中引入亮度和饱和度分量,并对颜色空间进行非均匀量化,提高目标与背景的区分度,抑制背景像素对目标的干扰。在多个视频数据上的仿真实验结果表明,该算法有效地克服了经典CAMshift算法对背景像素敏感的问题,提高了与背景色调相近场景下目标跟踪的准确性。Because the classic CAMshift algorithm is vulnerable to the pixel interference of a similar background color,the CAMshift object tracking algorithm based on non-uniform quantization of HSV is proposed to effectively solve the flaws of the classic CAMshift algorithm. By introducing the luminance and saturation components to the color histogram of the classic CAMshift algorithm and non-uniformly quantizing the color space,the designed algorithm improves the discrimination of the target and background,thus suppresses the interference of the background pixel to the target. Simulation results based on the several video data show that the problems of the classical CAMshift algorithm sensitive to background pixels is effectively solved and the target tracking accuracy of scenes with the similar hue in the background is improved.

关 键 词：CAMSHIFT 目标跟踪 非均匀量化 颜色直方图 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]