Research On Character Tecognition Technique Based On Machine Vision

As a important section of development of Machine vision, Character Recognition Technique have applied very widely in many fields. But in industrial locale, the complication of environment around bring lots of difficulties to exactly recognize the characters. The paper have mainly made research on Character Recognition Technique in train wheel steel characters , which is shot in industrial locale.The paper uses common Image Pretreatment method to eliminate noise and heighten an objective. Analyse the feature of steel characters picture, compare two Line Inspection methods and adopt Hough transformation to rectify the picture. Through comparing various Edge Inspection Operator, Combines Canny operator with Closing Edge, acquire the outline of character, and separate characters from background. Analyze the difference of contiguous pixels in gray image , make use of Zone Continuance method,and merge each character region by the character's outline. Approach a circle ,read radially a cirque area from inner radius to outer radius,and rectify the freakish characters. Use Projection method and abstract each character. Compare various Character Recognition methods, adopts a method named Crossing and extract features of NMI, and make Template Matching.Through six main steps , the paper complete whole recognition process to steel characters.The experiment result manifest that ,the method is easy and realization is fast in the VC program designed by myself, and it can come up to 90% above in recognition rate, it can be fully meet inspection demand in industrial locale after been improved.