基于帧差与背景差分的改进目标识别算法

ModifiedTarget Recognition Algorithm based on Frame Difference and Background Difference

为改进静态背景下三帧差法检测容易出现虚假边缘、空洞等现象以及背景差分法中背景更新缓慢、适应能力差的问题,提出一种融合边缘检测的改进三帧差法与背景差分法相结合的算法。该算法加入了边缘检测算子,对传统的三帧差分法进行改进,然后与基于Surendra算法的背景差分法进行结合,运用多种逻辑运算,经图像后处理综合得到运动区域图像。实验结果证明,此算法有效避免了鬼影、空洞、双影等的出现,对环境适应性好,与传统方法相比,算法准确率高,可满足实时性的要求。

In order to improve the phenomenon of false edges and voids in the three-frame difference detection under static background and the problem of slow background update and poor adaptability in the background difference method, a modified three-frame difference method in combination with background The algorithm, with the addition of an edge detection operator, could improve the traditional three-frame difference method. Then, combined with the background difference method based on the Surendm algorithm, and by applying multiple logical operations, a motion region image is obtained through image post-processing. The experimental results indicate that this algorithm could effectively avoid the appearance of ghosts, voids, double shadows, etc., and has better adaptability to the environment. Compared with the traditional method, this algorithm has high accuracy and can meet the requirement of real-time.