基于色度分析与L1加权正则的镜面高光抑制算法

Specular Highlight Suppression Algorithm Based on Chromaticity Analysis and L1-weighted Regularization

针对成像过程中镜面高光掩盖目标表面特征,从而影响诸多视觉应用的问题,提出一种基于色度分析与L1加权正则约束的镜面高光抑制算法。该算法不依赖于聚类、分割等前期处理;结合目标图像强度分布,提出了中值无高光机制(MSF),有效分离高光分量,抑制噪声干扰;同时联合色度分析与局部色彩相关性进行L1加权正则约束,融合变量分裂法进行快速求解,高效恢复目标原有的纹理、色彩和边缘等细节信息。实验结果表明,对于不同目标场景,该算法能实现镜面高光分量的准确抑制分离,目标原有的纹理结构、色彩等特征得到极大保留,算法可靠性强,计算时效性高。

To solve the problem that the specular highlight covers the surface features during the imaging process and thus hinders many visual applications,this paper proposes a novel specular highlight suppression algorithm based on chromaticity analysis and L1-weighted norm regularization constraints.This method doesn’t rely on such preprocessing procedures as clustering and segmentation.By using the distribution of image intensity,the Median Specular-Free（ MSF） mechanism is put forward to effectively separate highlight components and suppress noise interference.Meanwhile,by combining the chromaticity analysis with local chromaticity correlation,the L1-weighted norm regularization is carried out.Then,the variables splitting strategy is introduced to realize fast optimization,and the inherent details of the targets were effectively restored,such as their texture,color and edges.Experimental results demonstrate that our method can realize accurate suppression and separation of highlight components,and preserve the original characteristics of the targets to a great extent.This algorithm is highly reliable with increased computational efficiency.