基于Retinex理论的低照度图像对比度增强算法

Low Illumination Images Based on Retinex Theory Contrast Enhancement Algorithm

低照度图像的对比度增强是图像处理领域中的热点问题,但图像对比度增强的同时难以有效保持图像的纹理及细节,图像的质量得不到保证。为了解决这个问题,将图像分解、亮度转化函数及Retinex理论相结合,提出了一种简化的基于Retinex理论的低照度图像对比度增强算法,可对单张低照度图像进行处理。首先,基于低秩纹理先验的结构-纹理图像分解模型对图像进行分解,提取原始图像的纹理图,再对原始图像进行初始照射图估计、照射图纹理去除、边缘清晰及亮度增强等处理;然后利用增强后的照射图和Retinex理论,得到原始图像的反射图像;最终,将反射图与纹理图相结合得到最终融合图像。实验结果表明,所提出算法的自然统计特性(NIQE)评价值更低,可同时降为1.82和1.89,具有更好的图像质量,较其他方法表现出更好的增强效果,证明了该算法在图像对比度增强的同时保证了图像的纹理以及细节。

Low illumination image contrast enhancement is a hot issue in the field of image processing,but it is difficult to effectively maintain the texture and details of the image during contrast enhancement,and the image quality can not be guaranteed.In order to solve the problem,combining image decomposition,brightness conversion function and Retinex theory,a simplified low illumination image contrast enhancement algorithm based on Retinex theory was proposed,which can process a single low illumination image.Based on the low rank texture transcendental structure-texture image decomposition model of image decomposition,the texture map of the original image was extracted,and then the original image was processed by initial irradiation image estimation,irradiation image texture removal,edge clarity and brightness enhancement.Then the reflection image of the original image was obtained by using the enhanced irradiation image and Retinex theory.Finally,the final fusion image was obtained by combining reflection image with texture image.Experimental results show that the NIQE value of the proposed algorithm is lower,which can be reduced to 1.82 and 1.89 simultaneously.The proposed algorithm has better image quality and better enhancement effect than other methods,which proves that the proposed algorithm can enhance the image contrast while ensuring the texture and detail of the image.