Circle Detection System Using Image Moments

Abstract

The growth of system detection has significant development. The circle detection system is widely used to help people based on their needs, and it also could be used as learning media in educational fields. Especially for students with special needs, applying a circle detection system in Augmented Reality (AR) media would help them a lot. In order to make the study activity more effective and suit the learning material purpose, a circle detection system that only detects perfect full circles is needed to minimalize misconceptions in circle learning material. From the previous method such as Circle Hough Transform (CHT), circle detection faces the complex transition from cartesian coordinate into Hough coordinate. The use of image moments would give a coordinate of centroid that could use to find the radius by using the circle equation. Two groups of datasets would test the newly proposed method of detecting circles. Based on the experiment, the accuracy of the new method was 96.7%. The average time consumption is 0. 405 s which is faster than the CHT method with 1.024 s. Circle detection using image moments is also more robust towards noise than the previous CHT method.