基于空域NSS的无参考图像质量评价

计算机与现代化

基于空域NSS的无参考图像质量评价

南京航空航天大学计算机科学与技术学院,江苏南京210016

收稿日期:2014-10-30 出版日期:2015-02-28 发布日期:2015-03-06
作者简介:夏裕建（1989），男，江苏南通人，南京航空航天大学计算机科学与技术学院硕士研究生，研究方向：图像处理，图像质量评价；孙涵（1978），男，副教授，博士，研究方向：数字图像处理，模式识别与计算机视觉。
基金资助:
国家自然科学基金资助项目（61203246,61375021）；江苏省自然科学基金资助项目（SBK201322136）

Noreference Image Quality Assessment Based on Spatial Natural Scene Statistics

College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China

Received:2014-10-30 Online:2015-02-28 Published:2015-03-06

摘要/Abstract

摘要：

为了有效地度量不同失真类型的图像质量，提出一种基于自然场景统计（NSS）模型的空域无参考图像质量评价算法。该算法利用自然图像归一化亮度系数的统计特征趋向于服从广义高斯概率分布

的特性，首先在空域计算自然图像的梯度，通过梯度密度选取自然图像的兴趣区域，提取兴趣区域图像统计特征，建立多元高斯分布（MVG）模型；然后对测试图像建立同样的MVG模型；最终通过计算测

试图像和自然图像在统计规律上的偏差来对测试图像的质量做出评价。实验证明该算法与主观评价具有较好的一致性。

关键词: 无参考图像质量评价, 自然场景统计, 兴趣区域, 梯度密度

Abstract:

In order to measure the image quality of different distortion types, a spatial natural scene statistics (NSS) based model of no reference image quality assessment

method is proposed. The normalized brightness coefficients of natural images obey the generalized Gauss distribution and the proposed method uses these features. The proposed

method can be described as follows. Firstly, the gradient values of natural images are computed in the spatial domain, the ROI (regions of interest) of natural images are

selected by gradient density, the statistical characteristics of natural images are extracted and the multivariate Gauss distribution (MVG) model is established. Secondly, the

same MVG model of test images is established. At last, the quality of test images is evaluated by calculating the distance of the MVG model between test images and natural

images. Experimental results show that the proposed method is more consistent with human subjective perception.

Key words: noreference image quality assessment, Natural Scene Statistics(NSS), region of interest, gradient density

中图分类号:

TP391.41

夏裕建,孙涵. 基于空域NSS的无参考图像质量评价[J]. 计算机与现代化.

XIA Yujian, SUN Han. Noreference Image Quality Assessment Based on Spatial Natural Scene Statistics[J]. Computer and Modernization.

［1］
高新波，路文．视觉信息质量评价方法［M］．西安：西安电子科技大学出版社， 2011:114115．
［2］Lee A B, Pedersen K S, Mumford D. The nonlinear statistics of highcontrast patches in natural images［J］. International Journal of Computer Vision, 2003,54(13):83

103.
［3］Wang Z, Bovik A C, Evan B L. Blind measurement of blocking artifacts in images［C］// Proceedings of IEEE 2000 International Conference on Image Processing. 2000,3:981

984.
［4］Narvekar N D, Karam L J. A noreference image blur metric based on the cumulative probability of blur detection(CPBD)［J］. IEEE Transactions on Image Processing, 2011,20

(9):26782683.
［5］Parvez Sazzad Z M, Kawayoke Y, Horita Y. No reference image quality assessment for JPEG2000 based on spatial features［J］. Signal Processing: Image Communication, 2008,23

(4):257268.
［6］Zhu X, Milanfar P. A noreference sharpness metric sensitive to blur and noise［C］// IEEE 2009 International Workshop on Quality of Multimedia Experience(QoMEx). 2009:64

69.
［7］Sheikh H R, Bovik A C, Cormack L. Noreference quality assessment using natural scene statistics: JPEG2000［J］. IEEE Transactions on Image Processing, 2005,14(11):1918

1927.
［8］Moorthy A K, Bovik A C. A twostep framework for constructing blind image quality indices［J］. Signal Processing Letters, IEEE, 2010,17(5):513516.
［9］Moorthy A K, Bovik A C. Blind image quality assessment: From natural scene statistics to perceptual quality［J］. IEEE Transactions on Image Processing, 2011,20(12):3350

3364.
［10］Saad M A, Bovik A C, Charrier C. Blind image quality assessment: A natural scene statistics approach in the DCT domain［J］. IEEE Transactions on Image Processing,

2012,21(8):33393352.
［11］Mittal A, Moorthy A K, Bovik A C. Noreference image quality assessment in the spatial domain［J］. IEEE Transactions on Image Processing, 2012,21(12):46954708.
［12］Mittal A, Soundararajan R, Bovik A C. Making a “completely blind” image quality analyzer［J］. Signal Processing Letters, IEEE, 2013,20(3):209212.
［13］Ruderman D L. The statistics of natural images［J］. Network: Computation in Neural Systems, 1994,5(4):517548.
［14］Moorthy A K, Bovik A C. Statistics of natural image distortions［C］// IEEE 2010 International Conference on Acoustics Speech and Signal Processing(ICASSP), 2010:962

965.
［15］Sharifi K, LeonGarcia A. Estimation of shape parameter for generalized Gaussian distributions in subband decompositions of video［J］. IEEE Transactions on Circuits

and Systems for Video Technology, 1995,5(1):5256.
［16］Lasmar N E, Stitou Y, Berthoumieu Y. Multiscale skewed heavy tailed model for texture analysis［C］// IEEE 2009 the 16th IEEE International Conference on Image

Processing(ICIP). 2009:22812284.
［17］Bishop C M. Pattern Recognition and Machine Learning［M］. New York: Springer, 2006:200202.
［18］Sheikh H R, Wang Z, Cormack L, et al. LIVEImage Quality Assessment Database Release 2［EB/OL］. http://live. ece.utexas.edu/research/quality, 20070717.

[1]	杨贞, 单孟姣, 殷志坚, 杨凡, 李翠梅. 细粒度图像分类的通道自适应判别性学习方法[J]. 计算机与现代化, 2022, 0(10): 68-74.
[2]	吴彪, 周庆华, 曾小为. 基于Mask-RCNN的纸质医药包装钢印字符识别[J]. 计算机与现代化, 2022, 0(02): 108-113.
[3]	杨璐,魏敏. 基于多任务无参考图像质量评价模型研究[J]. 计算机与现代化, 2019, 0(11): 7-.
[4]	林钦壮1，钟映春1，罗唯师2. 面神经序列图像感兴趣区域识别[J]. 计算机与现代化, 2017, 0(11): 51-54.
[5]	赵新峰;王恒. 一种基于感兴趣区域元数据的视频转码方案[J]. 计算机与现代化, 2012, 1(9): 115-117,.
[6]	范楷;莫建文;张彤;袁华. 基于限定区域和Hough变换的车道线的检测方法[J]. 计算机与现代化, 2011, 1(4): 37-41.