摘要
提出一种结合人眼视觉敏感特性的无参考图像清晰度客观评价方法,利用高通滤波划分出人眼视觉敏感(图像细节)和非敏感区域,分别计算对应区域的清晰度,最后加权求和得出整幅图像的清晰度。利用公开的4个图像库与现有的6个无参考清晰度评价模型进行测试和比较,所研究的评价模型计算结果与主观评价分数有较好的一致性,且运算速度较快。
A non-reference objective image sharpness evaluation method considering human visual sensitivity is proposed.A high-pass filtering is used to divide the human visually sensitive area that mainly contains image details and the non-human visually sensitive area.Then the sharpness of the two areas are calculated respectively.and its weighted sum is the final sharpness of the whole image.The published 4 image libraries to test and compare is used.with 6 existing no-reference sharpness evaluation models.The results show that the model is in good agreement with the subjective evaluation score and possesses faster calculation speed.
作者
刘玉玺
郑秋梅
高绍姝
张晓东
张威
LIU Yuxi;ZHENG Qiumei;GAO Shaoshu;ZHANG Xiaodong;ZHANG Wei(College of Computer Science and Technology,China University of Petroleum,Qingdao 266580,China)
出处
《光学技术》
CAS
CSCD
北大核心
2020年第3期354-360,共7页
Optical Technique
基金
国家自然基金项目(61801517)
中央高校基本科研业务费专项资金(19CX02027A、19CX02029A)。
关键词
图像质量评价
无参考清晰度评价
人眼视觉敏感特性
高通滤波
Image quality evaluation
no reference sharpness evaluation
human visual sensitivity
high-pass filtering
