Research Of Counting Method For Steel Bars Based On Digital Image Processing

As one of the largest productions and the most widely used functional materials, steel is irreplaceable. Bar steel is one of the highest consumption, and steel enterprises are increasing production by researching advanced technology. In the process of packaging bar steel, traditional method which counting bar steel manually is inefficient and uneconomical. Automatic counting devices are imported to finish the job, but most of the devices can not reach the requirements of actual production. This research focuses on developing the automatic counting equipment.This subject analyzes the number of the bar steel using digital image processing technology, then receive the simulation results in Matlab. The contents as follow:in the first bar steel photos are pre-researched and the pretreatment plan is designed. Meanwhile image gray is selected to simplify the image information to improve the processing speed. Then in the purpose to make photo more identifiable for computer system, some useful information in the photo are given out while unwanted information are weakened or removed. This process is named Image Enhancement. After choosing a suitable binarization methods, processing with mathematical morphology can simplify the image data. It can Remove irrelevant structure, without damage the basic shape of the image features, to get a recognizable digital image.This paper uses Area Method which is effective to the adhesion effect photo. After calculating the average area of the bar end, this method fixes the number of bar steel in each connected domain by counting the area of domain, and then sums the result. The experimental and simulation results of photos which contains different bar steel are also mentioned. The experimental results reveal that the special colors of the bar end, in particular yellow and blue, impacts the accuracy of the counting. This research develops the experimentally proved improvement algorithm based on Fisher, in which commonly appeared colors, white, blue and yellow, are separated from black and gray in the hackground At last inadequate and research direction are also discussed...