Multiscale Fuzzy Reasoning (MFR) for Object Extraction

A. Sheikh Akbari and J.J. Soraghan (UK)

Keywords

Object extraction, Edge detection,Wavelet transform, Fuzzy reasoning.

Abstract

This paper describes a new Multiscale Fuzzy Reasoning (MFR) technique that is applied to automatic object extraction. The MFR utilizes skewed versions of an input frame. Each skewed frame is processed independently using fuzzy reasoning. The MFR achieves optimal edge detection using the wavelet decomposition of the signal followed by a novel fuzzy based decision technique that is applied across the scales. The resulting information is de skewed and combined using a fuzzy union operation. Results indicate a significant improvement in object extraction in undefined noisy environments using the MFR compared to object extraction using Mallat's multi scale technique

Important Links:



Go Back