The Research And Applying Of Industry Inspecting Based On Two-dimensional Vision

The application in industry online detection using Machine Vision Technology (MVT) has been investigated largely, The MVT can reduce the inspection cost largely and improve product quality, speed and efficiency, so it has been used widely in industry inspection and control field. Based on MVT, this paper does deeply research for printed character recognition algorithm through image process.First, this paper introduces the function ,struct and character of industry vision inspecting system, and studies the principles and methods of MVT. And Based on the order of process flow of this system, this paper expresses the algorithms and principles of system applying, the key techniques absolutely, and compares with other related techniques including image preprocessing's Histogram Equalization, Binary, Noise Removal and character Slant Correction and Syncopate, and Character Normalization. During the image preprocessing of the digital character number recognition system, we proposes some new methods such as image discrete noise removal, character slant and syncopate, and solve the problems of image slant, disturbance graveness and multi-type of character that fronted in system. These methods refers to the advanced techniques of the current image processing field, such as Ostu Binary , Niblack Binary, Connecting Area Analysis and Image Segmentation of character and image, And we discuss the identity of character abstracting method which used widely in digital character recognition, and compare the recognition algorithms, and express the error principle, method, sampling and training of Neural Network Using Back Propagating. And then this paper expresses the design and implementation of system simply.Lastly, we summarize the system and give the orientation and aim of future work.