摘要
针对增强处理后图像的客观评价值和主观评价值不一致的问题,提出了一种基于人眼视觉特性的图像质量评价方法。通过分析人眼视觉系统的感应特性,给出了位置影响因子、细节影响因子和灰度影响因子的定义,综合这些影响因子后得到了基于人眼视觉的质量影响权值。采用该权值对图像失真敏感度、失真度、信息熵增量和结构相似度等客观评价指标进行重构,并对各重构后的客观评价指标进行综合,得到了图像评价的综合指标。实验结果表明,该方法的线性相关系数比峰值信噪比和结构相似度分别提升了80.02%和78.85%,绝对误差均值比峰值信噪比和结构相似度分别提升了3.96%和7.96倍,均方根比峰值信噪比和结构相似度提升了6.45%和8.37倍,Spearman等级相关系数比峰值信噪比和结构相似度分别提高了1.01倍和1.69倍,表明该方法的评价正确性高、单调性最优,因此更适合于对射线检测数字图像的质量评价。
Focusing measurement for subjective assessment of enhanced radiographic image,an image assessment method based on human visual system is proposed.The response characteristic of human visual system is analysed,and the weight coefficient based on human visual system is suggested.The appropriate objective indexes are then chosen to construct the weighted objective indexes,which are synthesized as image assessment index.The proposed strategy promotes the linear correlation coefficient 80.02% and 78.85% compared with peak signal noise ratio（PSNR） and structural similarity index measurement（SSIM） to indicate higher correctness.The mean absolute error in this strategy is 3.96% higher and 7.96 times,and the root mean square is 6.45% higher and 8.37 times than PSNR and SSIM,showing a better consistency.And the approach improves by 1.01 and 1.69 times in Spearman rank of correlation coefficients,showing its fine monotonicity.
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2013年第7期91-95,共5页
Journal of Xi'an Jiaotong University
基金
国家自然科学基金重点资助项目(51175402)
陕西省重大科技创新项目(2011ZKC06-17)
关键词
人眼视觉特性
射线图像
图像评价
human visual system
radiography image
image assessment
