Fuzzified Gabor Filter for Text Recognition in Images

Abstract

This paper presents a fuzzified Gabor filter in applications of a text recognition in images. The filter orientation and wavelengths are fuzzified to optimize the filter's response. It was shown that Bell's function and triangular membership function are the most appropriate for fuzzification. The algorithm was tested on various images that contain text, and has provided satisfactory results. The components of interest were efficiently extracted. The use of fuzzified Gabor filter produced satisfactory results with efficient text detection and extraction, and good noise robustness.