提出了一种单张低动态范围图像曝光校正的方法, 可增强图像过曝光和欠曝光区域的细节.分别对图像过曝光区域与过欠光区域进行校正.在欠曝光区域, 使用WLS滤波将图像分解为照度图和反射图, 以提升区域亮度并增强对比度; 在过曝光区域, 扩展Retinex模型, 提出暗亮度用于表征过曝光程度, 压缩区域亮度并增强细节.然后利用基于显著性的融合算法将校正后的图像进行融合, 得到多区域增强图像.使用两种通用图像质量评价指标比较了多种方法的结果.实验证明, 本文结果的指标均优于其他方法的结果.另外, 在视觉效果的比较中, 本文方法可以得到细节丰富、色彩还原度高的校正图像.

This paper presents a method of single low dynamic range image exposure correction, which enhances details in over-/ under-exposed areas. Firstly, over-exposed areas and under-exposed areas are corrected respectively. In the under-exposed area, WLS filter is used to decompose the images into illumination map and reflection map, which enhances the brightness and contrast; In the over-exposed area, dark brightness is proposed to characterize the degree of over exposure and the Retinex model is extend to compress the brightness and to enhance details. After the corrected images are fused using the proposed fusion algorithm based on saliency, the multi-region enhancement image is obtained. The results of several methods are compared by using two common image quality evaluation indexes. Experiments show that the indexes of our results are superior to other results. In addition, in the comparison of visual effects, the proposed method can obtain the correction image with rich details and high color reproduction.