静态背景下运动目标的边缘提取

Edge Detection of Moving Objects in Static Background

摘要:

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在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的缺点,提出了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,根据分块大小自适应选取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,能满足实时性需要。?更多还原

Abstract:

For moving targets only taking a small proportion in a single frame image, while traditional edge detection method is too complicated to have a redundant calculation and is sensitive to the noise, it can not extract contour of the moving target obviously, so in this paper, an edge detection method combining the frame difference with image blocking is proposed. This method makes image difference firstly, and blocks the result with a certain threshold secondly, so as to extract the area of moving target. And thirdly, it detects the edge of this area, and finally does the “and” operation to the result of edge detection and frame difference to extract the target contour information. Experimental results show that the method can make edge detection with image blocking exactly. Also this method can detect the moving object accurately and effectively.