基于图像块和边缘增强的运动目标检测

针对复杂背景下的噪声、动态背景和阴影等影响，提出一种基于图像块和边缘增强的运动目标检测算法。用图像块建模，按掩膜结构特点修补断桥使掩膜连续，采用高层图像金字塔建模得到目标质心，用于提取目标，用Sobel平移算子得到细化的边缘，按边缘信息对掩膜进行增强处理从而得到完整的目标。实验结果表明，该算法可有效提取目标，消除单高斯建模法去除不了的复杂背景和阴影，克服传统帧差法常见的边缘不完整的缺点。

Moving Object Detection Based on Image Block and Edge Enhancement

Aiming at the noise, moving background interruption and shadow in the complex background, a motion detection approach based on image block and edge enhancement is presented. Image block is used to model background. The obtained mask is repaired to become connected depending on its structural features. High level image-pyramid model is used to get object centroid and extract target. The Shift-Sobel operator is used to obtain thin edges, which are used to enhance the edges of mask twice so that the mask edges become compact. Experimental results show that the proposed approach can deal with shadow and complex background disturbance better than using the single Gaussian model and thus extract moving object effectively and accurately. Some drawbacks of frame-difference approach such as incomplete edge are overcome in the suggestion.