Font Size: a A A

The Study On The Mechanism Of Humanoid Recognition For Video Image Chinese Character

Posted on:2005-10-15Degree:MasterType:Thesis
Country:ChinaCandidate:F QinFull Text:PDF
GTID:2168360122492314Subject:Detection Technology and Automation
Abstract/Summary:PDF Full Text Request
Human ocular apperceive is truly a robust character recognition system which can adapt to all kinds of noise. Using eyes and brain, man can recognize all sorts of characters and images, no matter what disturber and distortion. Obviously, it is very important for Chinese character recognition, even for image recognition that we do some research on the course of man ocular apperceive.The following are the major contents in the thesis:1. Image pretreatment. We do some research on binarization of gray image and character segmentation. An adaptive binarization filter Algorithm is presented, which combines adaptive filtering with local threshold binarization method. The algorithm overcomes the effect of uneven light and noise on binarization, especial for shade images.2. Thinning of Chinese character images and stroke extraction. A morphology-based thinning algorithm and a stroke extraction algorithm based on improved gradient is presented. The algorithm assorts Chinese character strokes as four sorts, and extracts them from Chinese character images.3. The establishment of etymon and the coding scheme for Chinese character. Some research on Chinese character structure is done, and after some coding schemes of Chinese character are summarized and trait of computer vision is considered , a new code-eight-shape code for Chinese character is presented, which indicates Chinese character structure trait and can be extracted automatically by computer from Chinese character images.4. The humanoid Chinese character recognition system. Based on eight-shape code scheme and code-extraction algorithm, humanoid Chinese character recognitionsystem is presented.The research in this article shows that the adaptive binarization filtering algorithm for gray character image is effective; the thinning algorithm for Chinese character images and the stroke-extraction algorithm can extract Hen, Su, Pie, Na four sorts stroke from Chinese character images; eight-shape code of Chinese character can effectively indicates the trait of Chinese character structure. The Chinese character recognition system based on eight-shape code can effectively improve noise immunity and correct rate of the system.
Keywords/Search Tags:adaptive filter, Chinese character stroke extraction, eight-shape Chinese character code, Chinese character recognition
PDF Full Text Request
Related items