Effective accumulation method for Hough transform-based track initiation

To solve the problem of peak clustering in Hough transform-based target detection, an effective accumulation method was proposed. Its main idea is that it is more likely for an estimated parameter corresponding to the cell with narrower sample dispersion and more vote-points to be the actual parameter. In the proposed method, the confidence factor of the cell, which represents the probability of the corresponding parameter to be the actual one, is defined first using the vote number of the cell and the transformed values of its votes and its neighbors' simultaneously; then the contributions of each cell's votes are determined based on the confidence factor of that cell and its neighbors. Simulations of multitarget track initiation with varied measurement noise and varied clutter densities are performed. The results show that compared with binary accumulation the method alleviates the problem of "peak clustering" and makes significant improvement in reducing false tracks. Besides, the proposed method has stronger robustness to measurement noise and clutters, especially in complex environment.