基于Retinex的自适应非均匀低照度图像增强算法

Retinex-based adaptive non-uniform low light image enhancement algorithm

摘要:

非均匀低照度图像会影响目标识别跟踪的效果,为了增强非均匀低照度图像,提出一种基于Retinex理论的自适应亮度层图像增强算法。首先将图像HSI模型中I层图像分离出来；然后通过Retinex理论和多尺度引导滤波器获得I层的入射分量,根据输入图像的均值自动获取调整参数k,对入射分量的亮度进行自适应调整；最后通过主成分分析法(PCA)提取两个图像中的细节特征并进行融合。对比实验表明,该算法能有效提升非均匀低照度图像的整体亮度和细节信息,同时减少非均匀照度的影响。

Abstract:

The effect of target recognition and tracking will be affected by non-uniform low light image.In order to enhance the non-uniform low light image, This paper proposes an adaptive brightness layer image enhancement algorithm based on Retinex theory. First, isolate the I-layer image in HSI model are extracted and separated; Then use the Retinex theory and the multi-scale guided filter to obtain the illumination component of the I-layer. Automatically obtain the adjustment parameter k according to the average value of the input image, and adaptively adjust the brightness of the illumination component; Finally, according to the principal features analysis (PCA), the detail features of the two images are extracted and fused. Comparative experiments show that the algorithm can effectively improve the overall brightness and detail information of non-uniform low-illuminance images, while reducing the effects of uneven illuminance.