ORIENTED MORPHOLOGY AND APPLICATION TO COUNT THE NUMBER OF STEELBARS IN THE BUNCHES BY IMAGE PROCESSING

Abstract

Counting objects in the image is a problem in the practice such as counting cells, counting red blood cells through images from electron microscopes, counting industrial products via industrial cameras... The biggest difficulty with counting by image processing is the images of the objects are often touch together. That is causing erroneous results compared with the reality result. So far, many methods have been used to increase the accuracy of results. However, the results of these methods are not yet accurate to 100%. In this paper, we propose a new method of applying oriented morphology to separate objects touch together in the image to bring the absolutely accuracy of the counting results in some  applications in industry. The counting experiment was performed on photos of steel bundles. The result of counting the number of rebar in a bundle can reach absolute accuracy. This method can be applied instead of manual counting in existing steel mills.