Accelerating CORDIC for hough transform

CORDIC, a hardware efficient algorithm, is commonly employed to implement the linear Hough transform (HT) for detecting straight lines in images. In this paper, we exploit the properties of the CORDIC algorithm to accelerate the HT computations. Unlike the conventional approaches, where CORDIC is used for computing HT for a given set of angles, the proposed method uses CORDIC to define the angle set based on the required resolution. It is shown that with minimum number of iterations, the desired accuracy can be obtained and adjusted by just varying the register lengths. The proposed method is shown to provide 6× speed up without any additional hardware when compared with existing CORDIC based HT computation methods with 9× lesser number of computations.