Comparison of Random Circle Detection and Hough Transform Method in Detecting Obstructed Circle Object

Detection of circles in digital imagery is important in object recognition. In real life, there are many circular objects are imperfect and are blocked by other objects. In this study, the randomized circle detection method is used to detect unobstructed circle objects in a digital image. If the Hough transform method uses an accumulator to store parameter information, then randomized circle detection does not need to use accumulator. An initial stage of preprocessing is applied to reduce noise and obtain the edge of object in the image. Each object's pixel edge will be stored which will then be recognized by the randomized circle detection method and the Hough transform method. The results of these two methods are compared each other to according to three parameters: accuracy, processing time and memory storage. In this research, 110 sample images order to analyze the performance of both methods. The results showed that randomized circle detection method was more effective and efficient than Hough transform method. The randomized circle detection method can detect the circle until 20% degree of visibility. Hough transform method can detect a circle is only 40% degree of visibility. Computation time of randomized circle detection method is faster than Hough transform method and 2, 18659 seconds. The memory used randomized circle detection method smaller than HT method 128,494, 2 Kb