激光与光电子学进展, 2018, 55 (2): 021007, 网络出版: 2018-09-10
彩色图像质量评价的广义平均池化策略 下载: 947次
General Mean Pooling Strategy for Color Image Quality Assessment
图像处理 彩色图像质量评价 全参考 广义平均 色度信息 池化策略 image processing color image quality assessment full reference general mean chromatic information pooling strategy
摘要
彩色图像质量评价(CIQA)是目前图像质量评价(IQA)研究的一个热点。色度信息对人类视觉系统(HVS)会产生一定的影响,将RGB图像转换到另一色彩空间YIQ,将灰度图像的结构相似性指标(SSIM)、基于梯度的SSIM(GSSIM)扩展得到彩色图像的SSIM和GSSIM(即C-SSIM和C-GSSIM)。HVS是一个复杂的非线性系统,采用2种广义平均池化策略来刻画HVS特性,以提高C-SSIM、C-GSSIM及彩色图像的特征相似性(C-FSIM)的评价效果。在TID2013图像数据库中进行数值计算,结果表明:基于广义平均池化策略的C-SSIM、C-GSSIM及C-FSIM能有效提高IQA的准确率。
Abstract
Color image quality assessment (CIQA) is a hot spot in researching image quality assessment (IQA). Chromatic information has a certain effect on the human visual system (HVS). Based on the conversion of RGB images to another color space YIQ, we obtain SSIM and GSSIM of the color image (C-SSIM and C-GSSIM) by extending the structural similarity index (SSIM) and gradient-based SSIM (GSSIM) of the grayscale image. In addition, considering HVS as a complex nonlinear system, two general pooling strategies are used to describe HVS characteristics to improve the evaluation effect of C-SSIM, C-GSSIM and feature similarity of the color image (C-FSIM). The numerical results, performed in TID2013 image database, demonstrate that C-SSIM, C-GSSIM and C-FSIM based on the general mean pooling strategy can effectively improve the accuracy of IQA.
马月梅, 陈海英, 刘国军. 彩色图像质量评价的广义平均池化策略[J]. 激光与光电子学进展, 2018, 55(2): 021007. Yuemei Ma, Haiying Chen, Guojun Liu. General Mean Pooling Strategy for Color Image Quality Assessment[J]. Laser & Optoelectronics Progress, 2018, 55(2): 021007.