Indian Institute of Information Technology, Design and Manufacturing-Kancheepuram, Chennai 600127, India.

---

Abstract—One of the most common distortions in images is blur which makes performing lot of image processing and computer vision tasks like detection and recognition a very difficult task. Assessment of quality images corrupted blur is very important and active area of research. Nonsubsampled contourlet transform is one of the transformation which is shift invariant and used in lot of tasks of image processing like denoising, image enhancement etc. In this paper we propose an image quality measure for blurred images which uses non subsampled contourlet transform (NSCT) features and support vector regression for the task of image quality assessment of images corrupted by blur. Comparative studies with other well known image sharpness and quality measures on standard databases and correlations with Human visual system (HVS) is also presented in the paper.

Index Terms—Blur, human visual system(HVS), image sharpness, image quality assessment(IQA), no-reference image quality assessment (NR-IQA).

[PDF]

Cite: Kanjar De, Masilamani V., "Image Quality Assessment for Blurred Images Using Nonsubsampled Contourlet Transform Features," Journal of Computers vol. 12, no. 2, pp. 156-164, 2017.