Detecting Edges in Noisy Multimedia Environments

This paper examines the process of detecting edges using partial derivatives of Boolean functions in noisy multimedia environments. Applicable to binary and multi-bit (grayscale) data, this novel approach examines the binary representation of the original data to reconstruct edge locations within the scene. Competitive with traditional detectors in noise-free situations, the strength is its ability to detect edges in corrupted signals with no need for pre-filtering. Evaluation is performed on several synthetic and natural 2D images corrupted with impulse noise