雾天图像增强计算模型及算法

摘要

The degradation of foggy images is non-linear with the depth of field, a new adaptive Retinex algorithm in foggy image enhancement is proposed. The algorithm adjusts the parameters of Retinex automatically according to the strong degree of fog reflected by the features of cumulative distribution function. A new calculation model of image enhancement is built based on this. The model combines the results enhanced by adaptive Retinex algorithm and CLAHE algorithm with different weights adaptively, makes enhanced images be able to keep both color constancy and luminance constancy, and make foggy images clear. Subjective observation and objective evaluation show that, the proposed method is more effective than HE and MSR algorithms both in details enhancement and color preserving.%针对雾天图像退化程度与景深呈非线性关系的特点,提出一种自适应Retinex雾天图像增强算法,根据图像像素点所处的人类视觉区域反映出的雾的厚薄程度自动调整算法的参数取值.在此基础上,构建一种新的图像增强计算模型,将自适应Retinex增强算法与CLAHE增强算法的增强结果自适应地加权融合,使得增强结果能够同时保持色彩恒常性和亮度恒常性,实现对雾天图像的清晰化.主观观测和客观评价表明,本文方法比HE和MSR算法在雾天图像细节增强及色彩保持方面具有更好的效果.