Fast straight-line extraction algorithm based on improved Hough transform

In order to effectively solve the real-time problem of extracting straight-line features in high speed image processing field, a new fast lines extraction method is presented based on multi-constraint Hough transform (HT). Firstly, three structural characteristics of straight lines were analyzed in digital discrete image model, and then the elementary line segments structure (ELSS) and its angle were defined to represent the candidate line contour points and its candidate angle in the same row, column or 45 (or 135) degrees scanning lines. Secondly, the standard Hough transform of angle constraints was only applied to the center point of ELSS rather than the total straight edge points, which reduces the computational complexity and increases the computational speed. Straight line parameters were recovered from the peaks in HT parameter space. The efficacy and efficiency of our method are proved by experiments on both synthetic image and real world image. Compared with gradient Hough transform (GHT), our method increases computational speed by 190 and 22 times respectively.