Image Segmentation of Adhering Bars Based on Improved Concavity Points Searching Method

Abstract

It is difficult to track, count and separate the bars moving at a high speed on production line for their overlap under occlusion. Therefore, it is necessary to establish a reliable, practical splitting mechanism for the adhered bars. This paper proposed a new solution to the problem of bars adhesion: the plane array camera was utilized to acquire the images of moving bars so as to recognize the centroid coordinates of the bars ends and compute their area with a Blob algorithm, two geometric parameters were utilized to detect adhered bars, and the presence of adhered bars was analyzed according to the convex hull. For the adhered bars, the segmentation points were searched using scanning method by a series of the rules to determine the optimal segmentation line. The proposed method can segment the adhered bars effectively with matched concavity points. The experimental results show that the method can well segment and count bars moving at a high speed on production line, with the counting accuracy near to 100% and the recognizing time in millisecond.