The Study On A Real-time Digital Recognition System Based On Image Processing

Because image detection system can acquire quickly a lot of message and is easy to process automatically, at the same time it is easy to integrate design message and machining control message, so it is applied in many fields of modem automation manufacture, such as product monitor, finished product inspection and quality control. Image detection technique can improve greatly production efficiency and autoimmunization degree. It is the base of computer integration manufacture.Many methods have been proposed for digital character recognition these days, but they are often subjected to substantial constraints. In this paper, we have developed a practical vision system for recognizing the digits on steel billets, and we have also developed a numerical recognition system for Automated Guided Vehicle.(1)In chapter 1 and 2 of this thesis, we introduced these image processing algorithm in detail: Image Segmentation,Image Filter, and Character Recognition, and the theory of image-processing is introduced too.(2)A system is established to recognize the numbers on steel billets in chapter 3. The Infrastructure of this system, such as the selection of CCD and lighting, is being stated at first. According to the characteristics of the numbers to be recognized, a new approach is proposed. Based on image processing technique, candidate regions of digits are extracted from an image through using the message of its stationary position. After using Mart edge detector,auto-threshold algorithm and grayscale area filter, noise was filtered and characters were extracted from the image. An improved template match technique was used to recognize the characters. The experimental results show that the system is efficient, robust and reliable for recognizing the digits.(3)The research work of chapter 4 relies on the new type of visual navigation vehicle platform which is developed independently by the Institute of Automated Instrumentation Engineering in Zhejiang University. It presents a real time digits recognition system for Automatic Guided Vehicle, which includes the overall Infrastructure of this system,the selection of some instruments, such as CCD,the digits,lighting. Based on machine vision, candidate regions of digits are extracted from an image through filtering approach using color components and threshold segmentation algorithm based on HIS color space. Since the digits in the image often have skew and slant, the digits are recognized after the skew and slant correction. The skew and the slant is corrected using the calculated angles. To recognize the digits, an efficient digit recognition method based on structural and statistical characteristics of the numbers is proposed. In experiments, we obtained a correct digit recognition rate of more than 90% when it running and can recognize all the numbers when it is stationary, and the total processing time is 63ms.The experimental results show that the system is efficient, robust and reliable for recognizing the digits.(4)In chapter 5, some conclusions and future researches are drawn in this dissertation.