Efficient Technique to Detect Edge in Images with Fuzzy Rules

Abstract

There exists an increasing demand to detect edge of an image for many real time applications. In this paper an innovative technique is proposed for detecting text using fuzy rules. The projected system primarily divides the image into fragment of 3 x 3 matrix. The proposed system uses fuzzy rules using input size of eight pixels and one output pixel. The output pixels will either be one among black, white or edge pixel. The fuzzy sytem is applied with sixteen rules for categorizing the pixel as target pixel. Fuzzification is performed which converts the input pixel into the fuzzy interval between zero and one. It is followed by calculating a degree of Hesitation, which is also called as the intuitionstic fuzzy indicator. The last step is the Defuzzification process where the pixel identified as the pixel is converted to its original image pixel with the interval between 1 and 255. The proposed system is weighed against existing edge detecting methods like Canny, Sobel, and ACO algorithm. The proposed algorithm works fine even for exigent scenarios of the image.