Functional image quality and performance evaluation for JPEG compression and de-noising filters

  • Authors:

  • Yun Fah Chang;Omar Mohd Rijal;Syed Abdul Rahman Abu Bakar

  • Affiliations:

  • Department of Mathematical and Actuarial Science, Universiti Tunku Abdul Rahman, Malaysia;Institute of Mathematical Sciences, University of Malaya, Malaysia;Faculty of Electrical Engineering, Universiti Teknologi Malaysia, Malaysia

  • Venue:
  • MATH'08 Proceedings of the 13th WSEAS international conference on Applied mathematics
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

As a performance indicator of algorithms, the quality of the reference and processed images is compared and its similarity value is calculated. This performance evaluation process can be done visually. However, the visual inspection is a highly subjective process and depends on the requirements of a given application. Consequently, many full reference image quality metrics were introduced in which the reference image being compared is regarded as 'perfect'. In this paper, the reference and processed images are considered as imperfect and we propose a functional metric using the coefficient of determination for unreplicated linear functional relationship model as a measure of the similarity, which in turn may be used as a definition for image quality and performance indicator of compression and de-noising filters. The sensitivity of the proposed metric is also studied and it provides a consistent interpretation for the distortion area of a processed image. Our experimental results show that the proposed functional metric is a good quality measure for JPEG compressed image and it is sensitive in differentiating the performance or low-pass and high-pass filtered image.