基于多源信息的阴影检测  

Shadow Detection Based on Multi-source Information

作　　者：李文举[1,2] 高连军[1] 崔晓松[1] 尉秀芹[1] 

机构地区：[1]辽宁师范大学计算机与信息技术学院,辽宁大连116081 [2]上海应用技术学院计算机科学与信息工程学院,上海201418

出　　处：《辽宁工业大学学报（自然科学版）》2013年第1期15-18,共4页Journal of Liaoning University of Technology(Natural Science Edition)

基　　金：上海市科委重大项目(12510503800);上海应用技术学院引进人才项目(YJ2013-36)

摘　　要：阴影检测在车辆检测中具有关键作用。针对目前车辆检测算法中阴影常被误检为前景目标的问题,提出了一种基于多源信息的阴影检测算法。首先进行前景目标的提取;然后分别进行基于边缘、HSV颜色空间和光学不变性信息的阴影检测;最后根据三种信息检测结果的综合分析确定阴影点。实验结果表明,该算法具有很好的阴影检测效果。Shadow detection plays a key rule in the vehicle detection. Aiming at the problem that shadows are detected incorrectly as foreground objects in the vehicle detection, a novel shadow detection algorithm based on multi-source information was proposed hereby. Firstly, foreground targets are gained. Secondly, the shadow detection algorithms based on edge, HSV color space and optical invariant information are executed, respectively. Finally, the shadow points are determined by comprehensive analysis of the three information detection results. The experiment results show that the proposed algorithm is highly effective for shadow detection.

关 键 词：阴影检测 多源信息 边缘 HSV 光学不变性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]