Research on Stainless Steel Pipes Auto-Count Algorithm Based on Image Processing

For solving complex stainless steel pipes auto-count problem, an algorithm is developed based on image processing. The algorithms include three sections: edge enhancement, objects extraction and pipe number count. In edge enhancement algorithm, a Canny operator combine with morphological for picking up interest area and removing noise. A novel convolution template is prompted for enhancing the thin steel pipe edge. For obtaining connected domains, an algorithm is designed by using the morphological theory. Divide each connected domain and the amount of steel pipes can be counted. Using this algorithm counting the pipe number in a real image, the 93.2% correct rate is got.