Study On Steel Slab's Serial Number Recognize Based On BP Neural Network

Image as an important means by which human being acquires information from outside world. Image processing and pattern recognition are crucial technology of computer image processing. In recent years, great developments have taken in image processing and pattern recognition technology. Image processing and pattern recognition technology have already applied in many domains, such as intelligence traffic, character recognition, remote sensing technology and so on. It has already obtained very good practical effect.In this paper image processing and pattern recognition technology were presented systemically firstly. Then a character recognition method on steel slab's serial number based on BP neural network was discussed in detail in the paper. At first, the image was transferred from RGB color space to HSI color space. Then slab's area was located from the picture based on color image's S weight's trait of HSI space. Every possible slab's serial number regions were found out by its texture feature after a serial of image pre -treatment means was used on the picture. After the possible areas were removed by slab's serial number region's Statistic features, all of slab's serial number regions were located accuracy on the image.Characters were segmented after slab's serial number area was obtained. Then characters features were picked up from the segmented characters. Finally, characters were recognized by BP neural network after the BP neural network was trained with enough stylebooks.An experiment was taken on acquired images with Microsoft Visual C++ 6.0 platform. The experiment results indicate that using Image processing and pattern recognition technology flexibility can recognize slab's serial number fast and efficiency. This method has very good accuracy and practical value.