基于改进帧间差分的动态目标检测算法

Dynamic target detection algorithm based on improved inter-frame difference

为了解决动态目标检测中出现的空洞、边缘断裂等问题,提出了一种融合Sobel的四帧差分法。在帧间差分法的基础上叠加差分运算,采用四帧差分算法增强目标内部轮廓信息,结合动态目标二值数据与Sobel边缘数据进行逻辑或运算,提取动态目标二值图像。结果表明,该算法将目标信息完整度提升到80%,综合评估H值提升18.6%,有效提高了动态目标检测的准确率。

This paper proposes a four-frame difference method based on Sobel designed to solve the problems of voids and edge fractures in dynamic target detection.The study involves enhancing the internal contour information of the target by using the four-frame difference algorithm;proceeding logic or operation by combing dynamic binary data and Sobel edge data;and extracting binary image of dynamic object.The results indicated that the proposed algorithm can improve the target information integrity to 80%,and the H value increases by 18.6%,effectively improving the accuracy of dynamic target detection.