静态背景下运动目标边缘提取

Edge extraction of moving object in static background

针对运动目标在单帧图像中所占的比例较小,传统的边缘提取方法对整幅图像计算时产生大量冗余,对噪声敏感,提取出的运动目标轮廓不明显,提出一种基于帧差法和图像分块相结合的运动目标边缘检测方法。该方法首先对序列图像进行差分,按照特定的阈值对图像进行分块,完成对运动目标的细分割,分离出运动区域和非运动区域,然后对分割出来的运动区域进行边缘检测,将边缘检测结果和差分结果进行"与"运算,从而提取出运动目标轮廓。实验结果证明,分块边缘检测方法能较为准确地提取出运动目标且能提取出清晰的运动目标边缘轮廓,能满足实时性。

Since the moving target occupies a small proportion in single-frame image,the traditional edge extraction method has the obvious drawback of a large number of redundant generated in calculation process of a whole image,is sensitive to noise,and can＇t extract a clear contour of the moving target. Therefore a moving object edge detection method combining the frame difference and image block is proposed. This method is used to perform the difference for sequence image,block the image according to a certain threshold,and achieve the moving object fine segmentation to separate the moving area and non-moving area. The edge detection is carried out for the segmented moving area. The＂and＂operation is proceeded for the edge detection result and frame difference result to extract the contour of moving object. The experimental results show that the blocking edge detection method can extract the moving object and its clear edge contour accurately,and satisfy the real-time performance.