融合背景差分的二次重构和内外标记 分水岭的行人检测方法

应用视频处理技术对行人交通进行研究受到广泛的重视，已成为智能交通领 域的研究热点.为了精准地提取交通场景语义信息，提出融合背景差分的二次重构和内外 标记分水岭的行人检测方法.首先对图像进行灰度级形态学开闭重构和背景差分运算，凸 显出前景区域，锐化背景区域；然后根据灰度图像局部极大值和邻域综合信息提取内部 标记，进行“准欧式”距离变换提取外部标记；最后对梯度图像进行修正和分水岭变换，提 取出图像中运动的行人.结果表明，该方法能显著地去除运动噪声的影响，检测到相对完 整的目标，很好地抑制了过分割问题, 在动态场景的行人检测中取得了较好的效果.

Pedestrian Detection of Integrating BS Based on Quadratic Reconstruction and IE Marker Watershed

Pedestrian detecting is an important part of Intelligent Transportation Systems (ITS) application. To accurately refine traffic information, the paper proposes pedestrian detection method of Integrating background subtraction (BS) based on quadratic reconstruction and internal and external (IE) marker watershed. First, the morphological quadratic reconstruction is used to denoise and filter, and the BS is carried to highlight prospect and sharpen background. Then, the internal mark is extracted according to the integrated information of gray image local maximum value and neighborhood value, the“Quasi-Euclidean”transformation is conducted to extract external marks; Finally, the gradient image is modified and watershed transform is employed to segment gradient image corrected for which extracts movement pedestrians. The results shows that this method significantly eliminates motion noise, targets are completely detected and the over-segmentation problem is effectively remitted. Furthermore, the pedestrian detection in dynamic traffic scene has achieved better results.