关键词: 多尺度Retinex算法, 图像增强, 亮度分量, 对比度

Abstract: Aiming at the problem that the traditional single color image enhancement algorithm based on multiscale Retinex did not consider the brightness component and local contrast when image enhancement was performed, which led to the image enhancement not reaching the optimal scale value, and the obtained enhanced image had the distortion and fog film, we proposed a single color image enhancement algorithm based on the improved multiscale Retinex. Firstly, the Gaussian environment function was used to convolve the three grayscale images and perform grayscale correction. The three corrected grayscale images were combined to obtain a preliminary single color image enhancement result. Secondly, we improved multiscale Retinex single color image enhancement algorithm to enhance the average brightness of single color image pixels. Finally, local contrast enhancement processing of color image after brightness enhanced was performed by nonlinear function, and image color after local contrast enhancement was restored by linear adjustment method to obtain   the best image enhancement results of single color image. The experimental results show that the proposed algorithm can obtain single color
 image enhancement results with good brightness and contrast.

Key words: multiscale Retinex algorithm, image enhancement, brightness component, contrast