Remote Sensing Image Enhancement Via Edge-Preserving Multiscale Retinex
Remote Sensing Image Enhancement Via Edge-Preserving Multiscale Retinex
复制标题
DOI：
10.1109/jphot.2019.2902959
复制
发表时间：
2019-03
影响因子：
2.4
通讯作者：
Lu Teng;Feng Xue;Qiudi Bai
中科院分区：
文献类型：
--
作者：
Lu Teng;Feng Xue;Qiudi Bai
近年来，图像增强由于其广泛的应用而受到相当多的研究关注。在本文中，我们提出了一种新的算法来增强遥感图像的对比度，同时避免颜色失真和细节丢失。首先，基于Sobel算子强度，得到局部自适应权重参数，对原引导滤波的固定正则化参数进行惩罚。然后，在不牺牲去噪性能的情况下，同时放大由固定正则化参数引起的平坦块和边缘块之间的灰度差异。最后，将具有颜色恢复的原始多尺度视网膜的高斯滤波替换为基于Sobel边缘检测算子的引导滤波，以估计图像的入射分量。与之前的算法相比，所提出的算法可以获得更多的边缘细节和更好的颜色保真度，但计算复杂度更低，效率更高。
Recently, an image enhancement receives a considerable research attention due to its wide application. In this paper, we propose a novel algorithm to enhance the contrast of remote sensing image while avoiding color distortion and detail loss. First, based on the Sobel operator strength, the local adaptive weighting parameter is obtained to penalize the fixed regularization parameter of original guided filtering. Then, the grayscale difference between flat patch and edge patch, which is caused by the fixed regularization parameter is amplified meanwhile without sacrificing the denoising performance. Finally, the Gaussian filtering of original multiscale retinex with color restoration is replaced with the guided filtering based on the Sobel edge detection operator to estimate the incident component of an image. Comparing with the previous algorithms, the proposed one can obtain more edge details and better color fidelity but with lower computational complexity and higher efficiency.