基于LatLRR多层分解与显著性检测的抗晕光方法

Anti-halo Method Based on LatLRR Multilayer Decomposition and Significance Detection

针对夜间行车过程中的晕光干扰问题,提出一种基于潜在低秩表示(LatLRR)多层分解与显著性检测的抗晕光方法。该方法先对可见光图像做HSV色彩空间转换,提取出H、S、V分量;采用LatLRR与复合滤波结合的方式将可见光图像的亮度分量V和红外图像多层分解,得到对应的低频层和高频层;通过平均规则和基于显著性检测与权重映射的融合策略得到低频融合层和高频融合层,线性叠加得到新的亮度分量V′;对H、S、V′分量做HSV色彩空间逆变换得到消除晕光且清晰度较高的结果图像。实验结果表明,该方法可以有效消除图像中的晕光,并显著提升图像的细节信息和清晰度,使色彩和亮度更适于人眼视觉系统。

To solve the problem of halo interference during driving at night,an anti-halo method based on LatLRR multi-layer decomposition and significance detection is proposed.First,HSV color space transformation is performed on visible images,the H,S and V components are extracted.The luminance component V of visible image and infrared image is decomposed into multiple layers by combining LatLRR and composite filtering to obtain the corresponding low frequency layer and high frequency layer.The low frequency fusion layer and high frequency fusion layer are obtained by means of average rule and fusion strategy based on saliency detection and weight mapping,and the new luminance component V′is obtained by linear superposition.With HSV color space inverse transformation of H,S and V′components,the result image with high resolution and halo elimination can be achieved.Experimental results show that the proposed method can effectively eliminate the halo in the image,and significantly improve the detail information and clarity of the image,so that the color and brightness are more suitable for human vision system.