一种背景自适应的运动目标检测算法

鉴于传统混合高斯模型背景更新的不足,融合边缘检测、帧间差分,提出一种背景自适应的运动目标检测算法。该算法利用Sobel算子提取图像的边缘信息,采用了三帧差分法把每帧图像分为背景区域、背景暴露区域以及目标运动区域,对背景暴露区域、背景区域以及运动区域采用不同的背景更新策略。实验表明,算法对缓慢运动物体、光线突变及背景融入等条件有较好的适应性,能够有效地检测运动目标。

Study on the background self-adaptative algorithm aiming at moving object detection

In light of the inadequate updating of the background of Gaussian mixture model,this essay puts forward the background self-adaptative algorithm aiming at moving object detection on the basis of the edge detection and inter-frame difference. This algorithm extracts edge information of images by using the So-bel operator with three-frame-difference method. At the same time the algorithm divides each image into back-ground area,background exposure area and target moving one,to all of which different background updating strategies are applied. Experimental results show that this algorithm has good adaptability to the slowly moving object,light mutation and background into conditions,and can effectively detect moving targets.