Scholarly record

IMPULSIVE NOISE REMOVAL IN COLOR IMAGES BASED ON THE LOCAL NEIGHBORHOOD EXPLORATION BY GEODESIC DIGITAL PATHS

Bogdan Smolka

Abstract

In the paper a novel technique of impulsive noise suppression in color digital images is proposed. The new filtering design is based on the exploration of the local pixel neighborhood by geodesic digital paths, whose cost defined as a sum of squared Euclidean distances between adjacent pixels in the RGB color space, serves as a measure of pixel corruption. The digital paths link the border of a filtering window with its central pixel and the cost of a path with minimum value is used to discriminate between the noisy pixels and those not affected by the noise process. The filtering is performed by thresholding the minimum cost assigned to the processed pixel. If a predefined threshold value is exceeded, then the pixel is replaced by an average of pixels detected as not corrupted. Otherwise, the processed pixel is not changed by the filter. The proposed denoising technique has been compared with a set of competitive methods and the analysis of the obtained results exhibits its satisfying properties. The interesting feature of the new algorithm is the ability to enhance image edges, while retaining tiny image details. This feature of the image restoration algorithm, combined with its low computational complexity, makes it an interesting tool for the real-time enhancement of color images corrupted by impulsive noise.

Publication Impact Profile

• Citations
• Scopus - Citation Indexes: 2

• Captures
• Mendeley - Readers: 1

PLUMX see details

DOI resolver: doi.org/10.5593/sgem2017/21/s07.046

Publication details

Back to publication list