Two-stage hough transform track initiation algorithm based on subtractive clustering

When Randomized Hough Transform (RHT) is applied to detect lower observable targets under heavy clutters, too many samplings are required and many of them are useless for track initiation. A two-stage Hough Transform track initiation algorithm based on subtractive clustering was proposed. First, Standard Hough Transform was used to filter clutters. Second, RHT was applied to get the candidate track parameters by fewer sampling numbers. Then, the track parameters were filtered to reduce the probability of false track initiation by minimum distance decision. Finally the true tracks were obtained using subtractive clustering. Experimental results show that the proposed approach behaves well in reducing the total sampling numbers for RHT, and useless samplings are decreased as well. Besides, this method has stronger robustness to clutters, especially in complex environment.