Accelerated Randomized Hough Transform for Circle Detection Using Effective Accumulation Strategy

randomized Hough transform is an efficient method for the circle detection in images.A new randomized Hough transform based on effective accumulation strategy is proposed to accelerate the circle detection and enhance the detection robustness.The proposed algorithm keeps the accumulation distribution of the parameter space after detecting a circle, which inherits effective samplings from the previous detections.The value corresponding to the invalid parameter is set to a negative value.With the generalized Bernoulli process in statistical theory, the principle of the acceleration algorithm is described and it is deduced that keeping the accumulation distribution of the parameter space will minimize both the sampling times and the parameter space clearance time.Compared to other three state-of-art algorithms, our algorithm shows a better noise immunity and computational efficiency.