基于颜色和区域梯度方向特征的阴影检测算法

来源期刊:中南大学学报(自然科学版)2013年第12期

论文作者:解文华 易本顺 肖进胜 甘良才

文章页码:4874 - 4881

关键词:目标检测;阴影检测;背景建模;混合高斯模型

Key words:target detection; shadow detection; background modeling; Gaussian mixture model

摘    要:针对利用颜色或者纹理等单一特征无法有效检测运动阴影的问题,提出一种结合颜色与区域纹理信息进行阴影检测的算法。在融合策略上,根据视频图像不同区域的纹理复杂程度,选择使用颜色信息或者区域梯度方向直方图对运动阴影进行检测。通过分析,证明区域梯度方向的光照不变性,提出一种利用该特征进行基于纹理的阴影检测新方法,可有效克服传统算法以像素为最小检测单位容易受到噪声干扰的不足。实验结果表明:该算法在不同场景下表现出较好的阴影检测性能,同时具有计算复杂度低、实时性好的优点。

Abstract: As single features such as color or texture can not be used to detect moving shadow effectively, a method of shadow detection based on color and regional texture was proposed. Through analyzing and proving the illumination invariance of regional gradient direction was proved, and a new texture-based method of shadow detection using this feature was proposed, by this way, the noise effect when using one pixel as the smallest detecting unit in traditional methods can be significantly reduced. Then, in the combination of color and regional texture, the color information or the regional gradient direction histogram was used respectively to detect moving shadow according to the complexity of texture in different image areas. The results show that the proposed algorithm is effective in detecting shadow under different scenes, meanwhile, it has characteristics of low computational complexity and excellent real-time ability.

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号