Abstract: According to the way how human eyes obtain information, this paper proposes a new image enhancement method. It can ensure that the gray differences between adjacent regions are maximally perceived by human eyes under the premise of keeping information. In this method, based on the adjacency relation of image regions, a gray consolidation strategy is proposed to represent image using the least gray. Then according to the Just Noticeable Difference(JND) curve, it signs a gray mapping relation for maximum perception of human eyes to enhance image. Experimental results show that this algorithm is better than current image enhancement methods in evidence.

Key words: image enhancement, Retinex algorithm, histogram equalization, image information, Human Visual System(HVS), Just Noticeable Difference(JND)

摘要： 结合人眼获取信息的方式，提出一种面向人眼视觉的图像增强方法。在保持图像原有信息的前提下，使图像中相邻区域间的灰度差异最大限度地被人眼感知。根据图像区域间的邻接关系，设计一种灰度合并策略，用最少的灰度表示一幅图像，基于人眼临界可见偏差(JND)特性建立一种灰度映射关系，通过灰度映射方式实现图像增强。实验结果表明，该方法的图像增强效果优于目前常用的图像增强 方法。

关键词: 图像增强, Retinex算法, 直方图均衡化, 图像信息, 人眼视觉系统, 临界可见偏差

CLC Number: 

• TP391