A template-covering based algorithm to count the bundled steel bars

An algorithm to count the number of bundled steel bars by using image processing is introduced. The image of bundled steel bars is seriously connected. To count the connected steel bars, a new counting algorithm based on template covering is presented. First of all, it preprocesses the bar section image to get a binary image. In the binary image, it firstly probes the target region which of large area with the horizontal template, then searches the probable center of the steel bar by the cross template, finally tentatively covers the image with the octagon template at the probable center, and calculates the area of the foreground region under the octagon template. To find the best location of the steel bar, it detects the connectivity of the covered region with the 8-connected image labeling algorithm based on the index table of the label, and detects the circularity with the boundary tracking algorithm based on Freeman chain codes, so that every steel bar can be recognized accurately and the number of bundled steel bars can come out. A set of experimental images is given in the conclusion of this paper. The experimental result shows that this algorithm can accurately count the number of bundled steel bars and display the counting results directly. The algorithm realizes a higher counting precision and covering accuracy than other algorithms.