This paper presents an adaptive enhancement framework for low contrast images to improve the deteriorated details and color information. The proposed algorithm is composed of two adaptive steps: Contrast enhancement and Color restoration. Firstly, according to the luminance map separated from the RGB image, a new spatial contextual similarity histogram (SCSH) is developed. The new SCSH considers the gray-level similarities among the adjacent pixels, following the standard deviation value of the luminance image. The contrast-enhanced image is obtained by applying the intensity transformation function derived from the cumulative distribution function obtained from the normalized values of SCSH. An adaptive weighted fusion is carried out between the luminance enhanced channel and original luminance channel with the weights computed from the enhanced luminance channel. Secondly, an adaptive color restoration technique is designed to reconstruct the more consistent color image using original chromatic information and modified luminance. The results are analyzed using the statistical paired t-test. Both objective and subjective experimentation prove the supremacy of the proposed algorithm with conventional and state-of-the-art algorithms.

本文提出了一种针对低对比度图像的自适应增强框架，以改善劣化的细节和色彩信息。该算法由两个自适应步骤组成：对比度增强和色彩还原。首先，根据与RGB图像分离的亮度图，开发了一种新的空间上下文相似度直方图（SCSH）。新的SCSH遵循亮度图像的标准偏差值，考虑相邻像素之间的灰度相似度。通过应用强度转换函数获得对比度增强的图像，该强度转换函数源自从SCSH的标准化值获得的累积分布函数。在亮度增强通道和原始亮度通道之间执行自适应加权融合，并根据增强亮度通道计算出的权重。其次，设计了一种自适应色彩恢复技术，以利用原始色度信息和修改后的亮度来重建更一致的彩色图像。使用统计配对分析结果t检验 客观和主观实验都证明了所提出的算法与传统算法和最先进算法的优越性。