Character Image Recognition Technology And Application Research

Image processing technology is increasingly becoming the indispensable technology research areas of society. Computer vision technology is led to the rapid development of image processing technology in the field of measurement to identify. In this paper, machine vision, image processing technology to achieve the automatic processing and recognition of the label image segmentation and instrument panel image, and focus on the label of surface image segmentation, label surface information classified summarized and surface character recognition and instrument panel color segmentation handwritten digits identification.Color segmentation of the image at home and abroad have achieved considerable process, some of the more mature approach. In addition, the handwritten character recognition technology is still immature, still is field of pattern recognition of one of the most challenging tasks. According to the product label, the instrument surface of handwritten digits in the domestic and foreign research is still relatively rare, scarce information research. Therefore, based on machine vision label and dashboard surface automatic image processing and recognition has important significance.In this paper, mainly in a variety of image processing technology, the product label color segmentation, and electrical instrument panel handwritten figures were studied. The main job of the work of the following aspects:1. Research color image pretreatment and segmentation method:Based on the analysis of the camera, lighting position and brightness for the influence of color, proposes the best camera, light visual angle the optimum scheme based on the edge of all kinds of color segmentation method, analyzes the advantages and disadvantages, the paper puts forward a kind of by uniform color space CIELAB effective segmentation edge detection method, implementation of the label and badminton background segmentation.2. For the badminton labels classified summarized, including level, factory name the flight path, the model, and design of the classification; and research the label of the character recognition block surface, analyzes some OCR could not identify the cause of the characters.3. On the surface of the meter handwritten digital research, and the pretreatment and feature extraction is the key part of handwritten numeral recognition, for adhesion, ambiguous handwritten numeral characters, based on the histogram equalization homomorphic filter and the image enhancement, Otsu threshold segmentation, Hough transform tilt correction method, put forward the vertical projection curve and character size combination of segmentation method, ultimately based on the BP neural network identification algorithm instrument the handwritten digits identification research.The article used Matlab and VC6.0software to realize the algorithm and method, the label color images and instrument panel script digital simulation analysis and processing. Finally after experiment, has achieved the good effect.