基于边缘信息和时空马尔可夫模型的运动目标检测方法

摘要

In order to overcome the affect of shadows and noise, a moving object detection algorithm based on spatiotemporal markov model and edge information is proposed.Firstly, three continuous edge images are obtained by edge extracted from three continuous images, then two initial labels are derived from the three successive edge images by frame-differencing.Secondly, the AND-label is obtained with the AND-operation on the two initial labels.and optimized labels are got by using the iterated conditional mode algorithm.Finally, the object is extracted with threshold segmentation and morphology.Comparison results show that the proposed method can detect object more accurately and robustnessly than other methods.%针对传统方法易受阴影和噪声的影响,不能精确分割出运动目标的情况,提出了一种基于边缘信息和时空马尔可夫模型的运动目标检测方法.首先对3帧连续的图像进行边缘提取,然后通过差分法运算获得两帧初始标记场,随后对两帧初始标记场进行"与"操作获得共同标记场,利用迭代条件模型求解共同标记场的全局最小值,进而实现近似求解最大后验概率的估算,获得优化的标记场,最后通过阈值分割和形态学处理完成对目标的检测.与多种方法进行比较表明,该方法能对运动目标进行准确检测,且具有很好的鲁棒性.