Edge detection is an important operation in digital image processing and also very important in field of computer vision, image segmentation and object recognition. Edge is line between two corners or surface which also a significant colour transition in an image. It also can be defined as an abrupt change in intensity of pixels and discontinuity in image brightness. The primary goal of edge detection methods is to extract the important feature or information in an image. In this study seven different techniques are used to extract the edge points for two different images. The seven techniques are involved the classical edge detectors as well the hybrid of the filters such as Sobel, Prewitt, Freichen, Robert, Sobel-Prewitt, Sobel-Freichen and Robert-Freichen. Performance factors are analysed in term of qualitative and quantitative aspect. Frequency distribution is used to measure the number active pixels in edge detected images. Frequency distribution is a measurement of quantitative based on the edge maps to each other relatively through statistical evaluation. The evaluation process is all added with qualitative aspect by visual analysis in term of good localization using fuzzy logic. A set of rules based on intensity of edges such as rate of ‘missing edges’,’thick edges’ and ‘broken edges’ defined. The conventional method required the human interpretation to decide upon the detection. Finally, performance evaluation is compared using Edge detection index. The indices used in Edge Detection Index are the sum of frequency distribution and visual perception scale of an image which will be obtained from fuzzy logic. The higher value of edge detection index indicates the better the filter. Overall findings indicated hybrid of Robert Freichen outperformed other combination of gradient filters with value of 2.73 in edge detection index for image 1(Lena) and 2.65 for image 2(Mechanical parts).
