No reference image quality assessment using blocked-based and frequency domain statistical features: a machine learning approach
by Jayashri V. Bagade; Kulbir Singh; Yogesh H. Dandawate
International Journal of Communication Networks and Distributed Systems (IJCNDS), Vol. 12, No. 1, 2014

Abstract: Images are compressed using lossy compression for fast transmission and efficient storage. Due compression artefacts quality of images are degraded. In web application, unavailability of an original image is a major challenge to evaluate quality of images. Therefore there is an immense need to develop a quality metric that will automatically assess quality without referring the original image. In this paper, no reference image quality assessment scheme using the machine learning approach is proposed. The block-based features brightness, contrast, local amplitude, texture and other parameters of the degraded images are calculated along with first order and second order statistical features in frequency domain. These features are given as inputs to well-trained back propagation neural network whose output is a quality score. The mean opinion score is used as target. The result indicates that accuracy of quality assessment is better in comparison with traditional mathematical predictors.

Online publication date: Sat, 21-Jun-2014

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Communication Networks and Distributed Systems (IJCNDS):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com