基于HSV颜色空间和局部纹理的阴影消除算法研究

Research on shadow elimination algorithm based on HSV color space and local texture

阴影消除目前在各类视频监控预处理阶段中阴影消除技术得到了极其广泛的应用,这一技术能够有效应对视觉背景提取运动目标检测算法的阴影问题,基于阴影消除技术提出了一种根据阴影颜色和纹理特征的阴影消除算法。本算法首先根据HSV颜色空间色度与明度信息来进行阴影判别,确定候选运动目标区域,并计算出以颜色信息为依据的运动目标阴影检测结果;再引入LBP算子对候选运动目标区域纹理信息进行提取,对比其与背景区域的纹理差异,得出以纹理特征为依据的运动目标阴影检测结果;最后合并上述两种结果,并对其进行逻辑运算融合,实现两者在阴影检测上的优势互补,得出最终准确结果。实验结果显示,基于融合颜色与纹理特征的阴影消除算法可实现对运动目标阴影区域的准确消除,无论是误检率还是漏检率都比平常方法更低。

At present, shadow elimination technology has been widely used in all kinds of video surveillance preprocessing stage. This technology can effectively deal with the shadow problem of visual background extraction moving target detection algorithm. Based on the shadow elimination technology, proposes a shadow elimination algorithm based on shadow color and texture features. This algorithm firstly makes shadow discrimination based on the chromaticity and brightness information of HSV color space, determines the candidate moving target region, and calculates the moving target shadow detection result based on the color information. Then, LBP operator is introduced to extract the texture information of the candidate moving target region, and the texture difference between the candidate moving target region and the background region is compared. Finally, the above two results are combined and combined with logical operation to realize their complementary advantages in shadow detection and get the final accurate result. The experimental results show that the shadow elimination algorithm based on the fusion of color and texture features can accurately eliminate the shadow region of the moving target, and the false detection rate or missing detection rate is lower than the usual method.