基于改进Hough变换的线性目标检测  

Detection of linear target based on improved Hough transform

作　　者：张国英[1] 程益钰 朱红[1] 

机构地区：[1]中国矿业大学(北京)计算机系,北京100083

出　　处：《计算机工程与设计》2014年第2期536-540,共5页Computer Engineering and Design

基　　金：国家863高技术研究发展计划基金项目(2011AA060203)

摘　　要：为提高检测高分辨率遥感影像中线性目标的精度,通过分析线性目标的几何特征,提出了一种用于检测线性目标的改进Hough变换算法。将Freeman链码和Hough变换相结合,采用对线性目标的边缘链码进行Hough变换的策略,有效地解决了重复检测、效率低、精度差的问题。对资源三号卫星影像进行实验,实验结果表明,该算法能有效地检测图像中的平行线性目标,实时处理性好。To improve the precision of the detection of linear targets in high-resolution remote sensing image, the geometric fea- tures of linear target are analyzed, and an improved Hough transform algorithm for linear target detection is proposed. The al- gorithm combines the Freeman chain code and Hough transform, and the method of Hough transform for the edge of the linear object is also employed. A problem of duplicate detection, low efficiency, poor accuracy is solved effectively. The algorithm is tested on No. 3 resources satellite images, the experimental results prove that the proposed algorithm can detect parallel linear targets effectively in the images, and can process well in real-time.

关 键 词：卫星影像 线性目标检测 平行线对 HOUGH变换 FREEMAN链码 

分 类 号：TN911.73[电子电信—通信与信息系统]