背景帧差与分块帧差相融合的运动目标检测

提出一种融合使用背景帧差和分块帧差的运动目标检测方法。该方法通过对图像的每个像素点进行学习，然后建立初始背景，通过不完全覆盖分块法对图像进行分块，对各子块进行帧间差分实现对前景图像的粗提取，采用otsu算法获取阈值，运用背景差分对前景图像进行细提取。背景采用分段学习的更新方法，能够消除光照变化、背景物体摇动等噪声。实验结果表明，该方法快速、准确，抗干扰能力强，能较好地满足实时检测运动目标的要求。

Moving Object Detection Based on Background Subtraction and Consecutive Blocks Frame Difference

An approach for moving objection detection with background frame difference is proposed and block frame difference. After studying every single pixel of the image for a while, an initial background model is set up. With the method of incomplete cover , the image is divided into several blocks. By performing inter-frame difference on these blocks, the coarse extrat of the foreground image can be realized. With otsu, the threshold can be got. Finally the thin extrat of the foreground image can be realized with background difference. The Segment learning was used to overcome the light changing and background interference. Experimental results indicate that the method has the charac- teristics of fast operation, accuracy, a good anti-interfence ability and it can detect the moving object effectively.