基于改进Hough变换的井盖目标检测与定位

Improved classical Hough transform applied to the manhole cover's detection and location

市政井盖的实时告失对城市管理具有重大意义。针对图像的复杂背景,提出了改进的Hough变换算法。在经边缘提取的二值图上,利用轮廓跟踪找到所有的轮廓,接下来设计轮廓过滤器排除虚假目标,进而在轮廓链表上进行Hough变换从而确定目标可能出现的区域,最终获得准确的井盖区域定位。该方案有效地提高了传统Hough变换的处理速度,减少了无效采样和累积。实验结果表明,该算法在多种复杂背景、不同视角、不同拍摄时间和条件、目标被部分遮挡等情况下都有很好的定位效果。

The lost alarm of sewer manhole cover is very important in municipal administration.An improved classical HT(Hough transform)is proposed for the complex background of image.On binary edge image,all the contours are found using contour tracking.Afterwards,contour filter is designed to eliminate inveracious targets and HT is used on every contour neighborhood in contour chain.This scheme has the advantages of high speed and solves the problem of useless sampling and accumulation.The experimental results show that this approach can get acceptable results in image with complex background,different angles of view,various conditions of sunlight and partly covered phenomenon.