与主观感知相一致的颜色校正评估数据集建立

发布时间：2018-05-26 17:31

本文选题：颜色校正 + 主观平均得分　；参考：《电子学报》2017年07期

【摘要】：为了获得与用户主观感知相一致的颜色校正算法和对校正结果进行客观评估,本文首先创建了一个针对颜色校正的数据集ICCD(Image Color Correction Database).ICCD数据集中的颜色差异涵盖了多种类型和粒度,其中颜色差异类型包括亮度、色相、饱和度、曝光度、对比度以及RGB中的R和G通道,每类颜色差异包括3个修改粒度.本文挑选了6种具有代表性的颜色校正算法对目标图像进行校正,并通过用户调查获得校正结果图像的主观平均得分值.基于ICCD数据集,本文对6种颜色校正算法的性能进行评估,得出在大多数颜色差异和粒度上,Pitie提出的迭代颜色分布转换算法的校正性能最好,同时具有较好的稳定性.最后,本文对14种图像质量评估方法进行评估,挑选出与已有的评估方法相比与主观感知一致性更好的评估方法.
[Abstract]:In order to obtain a color correction algorithm consistent with the user's subjective perception and to evaluate the correction results objectively, this paper first creates a color correction dataset for color correction. The color differences in the ICCD(Image Color Correction Database).ICCD dataset cover a variety of types and granularity. The color difference types include brightness, hue, saturation, exposure, contrast and R and G channels in RGB. Each color difference includes three modified granularity. In this paper, six typical color correction algorithms are selected to correct the target image, and the subjective average score of the corrected image is obtained by user survey. Based on the ICCD data set, this paper evaluates the performance of the six color correction algorithms. It is concluded that the iterative color distribution conversion algorithm proposed by Pitie has the best correction performance and good stability in most color differences and granularity. Finally, this paper evaluates 14 image quality assessment methods, and selects a better evaluation method compared with the existing evaluation methods and subjective perception consistency.
【作者单位】：福州大学数学与计算机科学学院;福建省网络计算与智能信息处理重点实验室;
【基金】：国家自然科学基金(No.61300102) 福建省自然科学基金(No.2014J01233);福建省自然科学基金杰出青年科学基金(No.2015J06014)
【分类号】：TP391.41

【相似文献】