Project: Impulse Noise Removal

PEOPLE

Mike Moore,  E. Abreu, M. Lightstone, S.K.Mitra

OBJECTIVE

One of the most common signal processing tasks involves the removal of impulses from signals. One group of nonlinear filters, called decision-based filters or state-conditioned filters, estimates the state of each sample prior to replacement. If the sample is determined to be uncorrupted, it is passed through the filter unaltered. If the sample is corrupted, it is replaced with an appropriate estimate. The signal-dependent rank-ordered mean (SD-ROM) filter [6] is a decision-based filter that has been shown to be very effective at removing impulses from 2-D scalar-valued signals (grayscale images). Excellent results were obtained for both two-state and multi-state versions of the filter. The filter has also been used to remove both oriented and random scratches from grayscale images [5]. A 1-D version of the filter has been used to remove impulse noise from speech and audio sig­nals [4]. A vector version of the filter has also been developed for removing impulse noise from vector-valued signals (color images) [3]. The simplest version of the SD-ROM algorithm is the two-state implementation. The two-state image-processing algorithm relies upon four thresholds. Initially, the thresholds were selected by trial-and-error. A statistical model for the performance of the algorithm with respect to its thresholds has been developed [2]. This model can be used to predict the thresholds assuming input image and noise distributions [1].

PUBLICATIONS

These materials are presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each authors copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.

1. M. S. Moore and S. K. Mitra, "Statistical Threshold Design for the Two-State Signal-Dependent Rank Order Mean Filter," in Proc. IEEE International Conference on Image Processing (ICIP 2000) , Vancouver, Canada, vol. 1, p. 904-907, September 2000.

2. Michael S. Moore and Sanjit K. Mitra, "Performance Analysis of the Two-State Signal-Dependent Rank Order Mean Filter," Proceedings of the SPIE, Nonlinear Image Processing X, San Jose, CA, vol. 3646, p. 56-66, January 1999.

3. Michael S. Moore, Moncef Gabbouj, and Sanjit K. Mitra, "Vector SD-ROM Filter for Removal of Impulse Noise from Color Images," Proc. EURASIP, DSP for Multimedia Communications and Services (ECMCS), Krakow, Poland, June 1999.

4. C. Chandra, M.S. Moore, and S. K. Mitra, "An efficient method for the removal of im­pulse noise from speech and audio signals," Proc. IEEE International Symposium on Circuits & Systems,Monterey, CA, June 1998, pp. 206-209.

5. E. Abreu and S. K. Mitra, "A simple method for the restoration of images corrupted by streaks," Proc. IEEE International Symposium on Circuits & Systems, Atlanta, GA, May 1996, pp. 730-733.

6. E. Abreu, M. Lightstone, S. K. Mitra, and K. Arakawa, "A new efficient approach for the removal of impulse noise from highly corrupted images," IEEE Transactions on Image Processing, Special Issue on Non-linear Image Processing, vol. 5, June 1996, pp. 1012-1025