Title: A multi-resolution transform for deblurring of images in the presence of impulse noise for real-time images

Authors: J. Amudha; R. Sudhakar

Addresses: EEE Department, Dr. Mahalingam College of Engineering and Technology, Pollachi 642003, Tamil Nadu, India ' ECE Department, Dr. Mahalingam College of Engineering and Technology, Pollachi 642003, Tamil Nadu, India

Abstract: Noise in an image tends to reduce the quality of the image by modifying the contrast and resolution and thereby making the process of extraction of information from the image is a challenging task. Deblurring of images in the presence of noise as it may be impulsive or multiplicative is a challenging task. It is essentially an issue of the trade-off between deblurring and denoising. A non-subsampled Contourlet transform has been used in a hybrid approach to address the trade-off factor in this paper. Removal of blur using Point Spread Function (PSF) or other methods introduces an amplification of noise in high-frequency regions of the image. The proposed work exploits the directionality features of the Contourlet transform to provide a balance in the optimisation problem. The experimentations have been conducted on standard test images and performance measured in terms of Peak Signal to Noise Ratio and Mean Squared Error.

Keywords: image restoration; contourlet transform; point spread function; Laplacian pyramid; image deblurring; denoising; non-subsampled; peak SNR; PSNR; signal to noise ratio; mean square error; multi-resolution transform; impulse noise; real-time images; optimisation; image processing.

DOI: 10.1504/IJBET.2017.082648

International Journal of Biomedical Engineering and Technology, 2017 Vol.23 No.2/3/4, pp.97 - 108

Received: 26 Apr 2016
Accepted: 07 Jul 2016
Published online: 04 Mar 2017 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article