A Novel No-Reference Stereoscopic Image Quality Assessment Method

Author

Shao, Feng ; Gu, Shanbo ; Gangyi Jang ; Yu, Mei

Author_Institution

Fac. of Inf. Sci. & Eng., Ningbo Univ., Ningbo, China

fYear

2012

fDate

21-23 May 2012

Firstpage

1

Lastpage

4

Abstract

No-reference stereoscopic image quality assessment (NR-SIQA) is an important research issue in three- dimensional researches. In this paper, a novel NR- SIQA method is proposed by analyzing the detail characteristic of the specific distortion type, and distortion-specific features are extracted to describe the distorted stereoscopic image. The relationship between stereoscopic features and subjective score is established by using support vector regression (SVR). Experimental results show that compared with the full-reference structural similarity image quality (SSIM) assessment method, the proposed method is more effective in quantifying image quality.

Keywords

feature extraction; regression analysis; stereo image processing; support vector machines; visual perception; NR-SIQA method; SSIM; SVR; feature extraction; image quality quantification; no-reference stereoscopic image quality assessment method; stereoscopic image distortion; structural similarity image quality; support vector regression; Databases; Image coding; Image quality; Stereo image processing; Training; Transform coding; White noise;

fLanguage

English

Publisher

ieee

Conference_Titel

Photonics and Optoelectronics (SOPO), 2012 Symposium on

Conference_Location

Shanghai

ISSN

2156-8464

Print_ISBN

978-1-4577-0909-8

Type

conf

DOI

10.1109/SOPO.2012.6271061

Filename

6271061