光学 精密工程, 2015, 23 (12): 3465, 网络出版: 2016-01-22
彩色融合图像的质量主观评价
Subjective evaluation of quality for color fusion images
彩色融合图像 主观评价 主观评价指标 综合指标预测模型 分值归整及相关性分析 color fused image subjective evaluation subjective evaluation index comprehensive prediction model score round and correlation analysis
摘要
考虑人眼主观评价在彩色融合图像质量评价中的重要作用,设计了52人的彩色融合图像质量主观评价实验。 采用“目标背景的感知对比度”,“清晰度”,“颜色协调性”,“颜色自然感”4个单一评价指标以及“基于目标探测的图像感知质量”,“基于场景理解的图像感知质量”2个综合评价指标对绿地、海天和城镇3类典型场景、8种融合算法获得的240幅彩色融合图像进行了主观评价,并依此进行指标分值归整及相关性分析,建立了2个综合指标的预测模型。实验结果表明:彩色融合图像的“颜色协调性”和“颜色自然感”有较高的相关性;基于目标探测的图像感知质量可用“目标背景的感知对比度”和“清晰度”来描述,基于场景理解的图像感知质量可用“颜色协调性”和“清晰度”来描述。但是对于不同的场景,每个指标的影响效果不同,所以预测模型中的权重系数不同。
Abstract
In consideration of the important role of human eye subjective evaluation in quality evaluation of color fusion images,a human eye subjective evaluation experiment by 52 people was designed. 240 color fused images gotten by eight kinds of fusion algorithms in three typical scenarios of greenbelts, sea to sky and cities and towns were evaluated subjectively by designed single-evaluation indexes(the perceptual contrast between target and background, sharpness, color harmony and color naturalness) and two comprehensive-evaluation indexes(Perceptual Quality based on Scene Understanding, PQSU and Perceptual Quality based on Target Detectability, PQTD). Then the subjective evaluation data gotten by 52 people for each index score and correlation analysis were carried out and two evaluation models were established. The results show that there is a high correlation between color harmony and color naturalness. The PCTD can be described by the perceptual contrast between target and background and the sharpness and the PQSU can be described by color harmony and sharpness. But,for different scenarios and visual tasks, the weight of each evaluation index is different.
金伟其, 贾晓婷, 高绍姝, 马国利, 潘定平, 刘佳妮. 彩色融合图像的质量主观评价[J]. 光学 精密工程, 2015, 23(12): 3465. JIN Wei-qi, JIA Xiao-ting, GAO Shao-shu, MA Guo-li, PAN Ding-ping, LIU Jia-ni. Subjective evaluation of quality for color fusion images[J]. Optics and Precision Engineering, 2015, 23(12): 3465.