Impulsive Noise Suppression from Images by Using Anfis Interpolant and Lillietest

  • PDF / 4,328,515 Bytes
  • 11 Pages / 600 x 792 pts Page_size
  • 34 Downloads / 198 Views

DOWNLOAD

REPORT


Impulsive Noise Suppression from Images by Using Anfis Interpolant and Lillietest Erkan Bes¸dok Computer Engineering Department, Institute of Science, Erciyes University, 38039 Kayseri, Turkey Email: [email protected] Received 14 July 2003; Revised 30 January 2004 A new impulsive noise (IN) elimination filter, entitled adaptive neuro-fuzzy inference system-based IN removal filter (Anfis-F), which shows high performance at the restoration of images distorted by IN, is proposed in this paper. The Anfis-F comprises three main steps: finding the pixels that are suspected to be corrupted, the Delaunay triangulation, and finally, making estimation for intensity values of corrupted pixels within each of the Delaunay triangles. Extensive simulation results show that the proposed filter achieves better performance than other filters mentioned in this paper in the cases of being effective in noise suppression and detail preservation, especially when the noise density is very high. Keywords and phrases: impulsive noise suppression, finite-element-based Anfis, Delaunay triangulation.

1.

INTRODUCTION

Image enhancement and noise filtering are the most common tasks in image processing. Images are affected by impulsive noise (IN) during image acquisition, transmission, or storage; therefore noise-free images are rare in the real world. Noise suppression of distorted images requires a balance between the gained improvement and the introduced degradation by a particular filter. Preservation of image details while eliminating IN is usually not possible during the noise suppression process, but both of them are crucial for the subsequent processing stages. It has been approved that the standard median filter (SMF) [1], as well as its modifications and generalizations, [2, 3, 4, 5, 6, 7, 8], offers satisfying performance in suppression of IN. However, these approaches are implemented invariantly across the image, thus they tend to alter the pixels undisturbed by IN and increase the edge jitters when the noise ratio is high. Consequently, achieving a good performance in the suppression of IN is usually at the expense of blurred and distorted image features. One way to avoid this problem is to include a decision-making component in the filtering structure based on very simple, but effective, impulse detection (ID) mechanism. The function of the ID mechanism is to check each pixel to detect whether it is distorted or not. Then, the nonlinear filtering scheme is achieved for the distorted pixels, while the noise-free pixels are left unaltered in order to avoid excessive distortion. Recently, such ID-based median filtering methods with thresholding operations have been realized by using differ-

ent modifications of impulse detectors where the output is switched between the identities or median-based filtering scheme [2, 4, 6, 7]. The Anfis-F, proposed in this paper, differs from the other median-based filters by performing the restoration of degraded images with high performance according to both subjective measures (e.g., visually pleasin