A No-Reference Quality Metric for Blur Image

Author

Zhang Taojia ; Zhang Hong ; Yuan Ding

Author_Institution

Image Process. Center, BeiHang Univ., Beijing, China

fYear

2013

fDate

20-23 Aug. 2013

Firstpage

1813

Lastpage

1816

Abstract

In this paper, we address the problem of no-reference quality assessment for blurred picture from digital camera. To evaluate the digital picture corrupted by defocus and motion artifact, the inherent information of blur kernel is estimated from the blurred image. We consider blur priori based on specific gradient domain, which is inferred in Alpha channel image. This metric is validated in LIVE Blur database and CSIQ Blur database, the results shows the high prediction accuracy with human judgment.

Keywords

gradient methods; image restoration; Alpha channel image; CSIQ Blur database; LIVE Blur database; blur image; blur kernel information; blurred picture; defocus artifact; gradient domain; motion artifact; no-reference quality assessment; no-reference quality metric; Channel estimation; Databases; Image edge detection; Kernel; Measurement; Principal component analysis; Vectors; Alpha channel image; Gradient; No-Reference; blur assessment;

fLanguage

English

Publisher

ieee

Conference_Titel

Green Computing and Communications (GreenCom), 2013 IEEE and Internet of Things (iThings/CPSCom), IEEE International Conference on and IEEE Cyber, Physical and Social Computing

Conference_Location

Beijing

Type

conf

DOI

10.1109/GreenCom-iThings-CPSCom.2013.335

Filename

6682346