Performance Comparison of Image Edge Detection Operators for Lontara Sanskrit Scripts

Abstract

Processing of image digital is a technology that can be used to enhance images and information about objects in images. Edge detection in digital image enhancement is a process that produces parts of image objects for segmentation and creativity of objects in the image. This research aims to get the best operators to detect the character of the word Lontara in Sanskrit manuscripts. This research also uses morphological operations in binary images to identify the many forms of the character of the word Lontara. Edge detection operators used are Sobel, Canny, Prewitt, and Roberts. There are 2 types of images used in this study are good quality images and poor-quality images. The parameter used to measure operator performance is Mean Square Error (MSE). The results we obtained from Roberts operator are the best operators to detect the location of the script with an MSE value of 0.8370 in images of good quality, MSE value of 0.8688 in images of poor quality.