Research On Automatic Identification System Of Various Information Label Based On Vision

Image identification and processing is one of the main researching contents of machine vision and application, and has the extensive application prospect in the robot and the industrial automation domain, automatic identification technology of label is one of the important domains of vision application, this article putted forward identification problem of product label of various information, and attempted to construct the actual label identification system according to the development of image processing technology in the pattern identification and the application of image identification technology in the automatic production line. Information of label generally consist of digits, letters, specific Chinese characters, barcode, people have done much research on the identification of digits, letters, Chinese characters based vision method, but these present methods are't competent for the synchronization identification of various information of label, so in order to answer requirement of real time quality and accuracy, and need to develop simple algorithm according to the actual situation;hardware method was used for the tradition identification method of barcode , but software method was used in this system.The main researching contents and productions of the article are as follows:1. The method of image processing and image segmentation has been studied, and putted forward the method of segmentation of label image based on segmentation problem of various information of label. This system can exactly make sure label position in entire image all the same, and extracted label from the entire image even if when label has rotated and size of label image has changed, images of various information which are need to recognize were extracted according to the position of various information (digits, letters, Chinese characters, barcode) in label.2. The methods of digits identification have been analyzed based on the basis theory of image identification technology, aiming at the problem which possibly exist conglutination between digits in label and digit structure exist deformity when printing, this article putted forward the method which the digits outline characteristic and statistical characteristic combined, it not only overcome many probledms but also can realize accuratelly identification of digits of many kinds of typefaces, and enhanced the identification speed at the same time.3. Because of the problem that structure and statistical characteristic of letters and Chinese characters are complicate and it's very diffcute to exact characteristic, this article putted forward to use the method of template match to identification of letters and Chinese characters, and studied all kinds of methods of template matching, Based on requirement of real time quality and accuracy of this system, the method which RST central invariant moment and the traditional template match combined was used, not only enhance speed of identification, but also ensure perfection of identification.4. Studied the basis theory of barcode and the concrete coding manner of EAN-13 barcode.Analyzed the software method of barcode identification and proposed to use the similar sides distance to recognize barcode based on coding manner of barcode, that identification system still may act the correct solution according to the similar sides distance even if the barcode printing exist flaw or the scanning is insufficiently, causes between the actual observed value and theoretics value has the big deviation.