Bisindo Alphabets Edge Detection Using Color Tracing of Object Boundary

Abstract

This paper proposes a method of edge detection for BISINDO alphabets formed using hand gestures. In image processing for edge detection process, it is very important section for using image analysis process for the next step. The goal of this study was to detect BISINDO letters edge using color tracing at object boundaries. BISINDO was an abbreviation of Indonesian sign language. There were 26 images of BISINDO letters tested. They represented letters A to Z. There was 3 steps in this study; the first was segmentation process using similarity measurement model based on color similarity and using thresholding technique, the second was morphology process using mathematical morphology with closing operator, and the last step was object boundaries tracing used in outside boundary tracing of an object and whole boundary in the object. The result of this study was the object boundaries detected with blue color whereas holes were detected with red color for 26 BISINDO letters (letter A to Z). Recognition accuracy rate for this study was 100%.