Real-time Robust Algorithm for Circle Object Detection

This paper presents a real-time robust algorithm to detect and accurately locate the circular objects in digital images. The algorithm consists of four steps.First the edge pixels are extracted using Canny edge detection algorithm followed by a noise removal process to remove the non-circle edge points.Afterwards, a direct least square fitting algorithm is developed to calculate radius and circle center information for each edge pixel cluster (a potential arc or a segment of a circle). In third step, a robust criterion is developed to distinguish the valid arcs from invalid arcs. Finally, those valid arcs belonging to the same circle are reassembled and fitting algorithm is run again to obtain the accurate information of that circle. The algorithm is implemented in Visual C++and tested on a laptop powered by an Intel Centrino Duo CPU at 1.66GHz. The experiment shows the algorithmpsilas three advantages. Its speed is fast, about 7 images/second for image size of 640X480. It is able toreliably detect full as well as partially-occluded circle objects even in a noisy environment, specifically 92%correct detection among 174 circles; the achievedaccuracy for radius and center location has reachedsub-pixel level on average.