基于Log-gabor小波的运动目标阴影检测  

Moving target shadow detection based on Log- gabor wavelet

作　　者：杨尚斌[1] 刘秉瀚[1] 

机构地区：[1]福州大学数学与计算机科学学院,福建福州350116

出　　处：《福州大学学报（自然科学版）》2016年第1期26-32,共7页Journal of Fuzhou University(Natural Science Edition)

基　　金：福建省自然科学基金资助项目(2013J01186;2012J01263)

摘　　要：针对阴影覆盖前后的场景纹理相似的特点和Gabor小波在纹理特征提取上的不足,利用Log-gabor在纹理特征提取上的较优性能及对光照的不敏感性,提出一种基于Log-gabor的运动目标阴影检测方法.首先根据阴影在颜色空间的特点,利用像素点亮度变化规律,对前景中包含阴影的运动目标进行检测,提取疑似阴影区域.然后对疑似阴影区域分块,获取块Log-gabor纹理特征.最后,针对运动目标和道路纹理区别,提出合理有效的阴影判定准则,使用判定准则识别阴影.实验结果表明,所提方法在不需场景先验知识情况下,能较快较准确检测出阴影.Before and after shadow covering, the textures of scenes are similar. Compared with Gabor- wavelet, Log - gabor wavelet have better performance on the texture feature extraction, and not sensi- tive to light, so we put forward a method of moving shadow detection based on Log - gabor. Firstly, according to the feature of the shadow in color space, we judge the probable shadow via analyzing the brightness of moving targets including shadow. Secondly, we put the probable shadow area divided into small regions, extract texture feature with Log - gabor wavelet. At last, due to the difference betweenmoving targets textures and road textures, we put forward a reasonable and effective criterion for judgment, then we judge shadow area with the criterion. The experimental results show that the proposed algorithm is efficient and robust in shadow detection without priori information.

关 键 词：阴影检测 运动目标 Log-gabor纹理特征 判定准则 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]