人工智能
计算机视觉
计算机科学
颜色恒定性
图像复原
图像(数学)
颜色校正
图像增强
伽马校正
图像处理
出处
期刊:Journal of Flow Visualization and Image Processing
[Begell House Inc.]
日期:2021-01-01
卷期号:28 (3): 71-88
被引量:1
标识
DOI:10.1615/jflowvisimageproc.2021036393
摘要
Typically, the image enhancement of foggy images based on multiscale retinex color restoration (MSRCR) algorithms results in dim images with halo artifacts. This paper presents a foggy image enhancement algorithm that combines MSRCR with adaptive gamma correction. First of all, multiscale Gaussian functions are used to extract the illumination components as per the MSRCR processing of foggy images. Then, an adaptive gamma function is constructed and the distribution characteristics of the illumination components are used to adaptively adjust the gamma function's parameters. The adaptive gamma function can now be used to correct the brightness in the hue, saturation, value (HSV) color space, with the brightness value of over-illuminated areas being suppressed and the brightness value of areas that are too dim being increased. This results in a fused image that can be converted into an RGB color image where the fogginess has been minimized. Experimental results show that the proposed method eliminates the dimness and halo phenomena associated with the basic MSRCR algorithm. In comparison to other foggy image enhancement algorithms the proposed algorithm generally performs better in terms of dehazing and brightness enhancement and is especially effective at removing noise (between 53% and 89% better than the next best-performing algorithm).
科研通智能强力驱动
Strongly Powered by AbleSci AI