BIQI和JND相结合的图像增强客观评价方法

发布时间：2018-05-09 11:32

本文选题：(BIQI) + 临界可见偏差(JND)　；参考：《光电子·激光》2016年06期

【摘要】：提出一种结合盲图像质量指标(BIQI)和临界可见偏差(JND)的图像增强客观评价方法。方法基于Retinex理论,将测试图像分成反射图像和照度图像,分别对反射图像和照度图像进行BIQI和JND评分,得到的评分之和作为图像增强客观评价得分。对原图像失真和背景亮度的变量控制系列的实验表明,Retinex分解得到的反射图像只受失真的种类和失真程度影响,而照度图像代表原图像环境光照信息具有亮度阈值效应;应用本文算法对增强后的图像评分得到Score的最高分为0.609 3,与主观评价结果一致;而PSNR、SSIM的评分都是和原图像本身比较,不能表明图像增强效果。实验结果表明,本文算法的评价结果与主观视觉一致,能用于定量对图像增强进行客观评价。
[Abstract]:An objective evaluation method for image enhancement based on blind image quality index (BIQI) and critical visual deviation (JND) is proposed. Methods based on the Retinex theory, the test images were divided into reflected images and illuminance images. The BIQI and JND scores of the reflected images and illuminance images were evaluated, respectively. The sum of the scores obtained was used as the objective evaluation score of image enhancement. The experiments on the original image distortion and background brightness variable control series show that the reflected image obtained by Retinex decomposition is only affected by the kind of distortion and the distortion degree, while the illuminance image represents the luminance threshold effect of the illumination information in the original image environment. The maximum score of Score after image enhancement is 0.609 3, which is consistent with the subjective evaluation results, but the score of PSNRN SSIM is compared with the original image itself, which can not show the image enhancement effect. Experimental results show that the proposed algorithm is consistent with subjective vision and can be used to evaluate image enhancement objectively.
【作者单位】：上海理工大学出版印刷与艺术设计学院;
【基金】：上海市教委科研创新重点(13ZZ111)资助项目
【分类号】：TP391.41

【参考文献】