A Method of Workpiece Coherent Line Detection Based on Progressive Probabilistic Hough Transform

Abstract

The detection and extraction of workpiece lines is the basis and key in the industrial production practice. In order to solve the discontinuity and disconnection problem during the line detection and extraction of a workpiece, we propose a coherent line detection method, which is Improved Progressive Probabilistic Hough Transform line detection(PPHT). Improved PPHT first performs edge detection combine with original PPHT algorithm to find lines of workpiece object. After discarding noise lines, this method divide the detected lines into several groups by finding collinear candidates. Then we calculate the supporting pixels of every group with foreground image, and apply Least Square Regression to achieve final line results. In this paper, we performed an experiment on thirty images of three different rectangular workpieces.The results indicate that, comparing to the PPHT, Improved PPHT decreased the relative error rate of the linear detection accuracy p by 62.06% on average, and the relative error rate of the recall rate R has decreased by 43.6% on average, thereby significantly mitigate the discontinuity existing in PPHT.