摘要
基于图像差分算法的运动目标检测中,固定好的摄像头在受到重型车辆驶过、强风等因素影响下,所拍摄到的图像仍会出现抖动,而图像抖动对提取图像中的运动目标干扰非常大。本文设计了一种基于预先划定一个或多个目标匹配区域的颜色直方图匹配算法,通过分析和比较抖动前后两幅图像目标区域和候选区域颜色直方图的相似度,找到最优匹配块并得到对应块运动位移,从而求出两幅图像的绝对运动位移,最后消除干扰位移。实验结果表明,该算法鲁棒性好,能够准确消除抖动位移给运动目标检测带来的干扰。
By the effects of heavy vehicles passing, the strong wind or other factors, the image captured by fixed camera will pro- duce dithering. But in moving target detection algorithm based on image difference, the image dithering will bring a lot of interference to extract moving targets from image. In this paper, a new method of image dithering elimination based on histogram matching in one or more pre-defined areas is proposed. By analyzing and comparing color histogram similarity in target area and candidate area between dithered image and normal image, the optimal matching block will be found and absolute motion displacement of two images will be cal- culated, finally, to eliminate interference displacement. Experimental results show that the method have a outstanding performance in robustness which can accurately eliminate the interference in moving target detection caused by dithered displacement.
出处
《山西大同大学学报（自然科学版）》
2017年第2期17-20,27,共5页
Journal of Shanxi Datong University(Natural Science Edition)
基金
太原工业学院青年科学基金[2016LQ04]
关键词
抖动干扰
运动目标检测
图像差分
颜色直方图匹配
dithered interference
moving target detection
image difference
color histogram matching
