Font Size: a A A

Based On Combination Strategy Online Uyghur Handwriting Recognition

Posted on:2013-01-17Degree:MasterType:Thesis
Country:ChinaCandidate:G W QuFull Text:PDF
GTID:2218330374966707Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
How to use the computer to process and recognize handwriting, is a very activeresearch topic in the field of pattern recognition. With the popularity of a variety ofmobile devices, traditional Uyghur keyboard input methods can not meet people'sneeds, while on-line handwriting the Uyghur recognition technology for its real-time,quick and easy loved by the people, thus becoming the handwriting recognition areasof the Uyghur the new direction of development and research topics.Discussed indetail and research-based portfolio strategy (HMM and rules) Uyghur handwritingrecognition and on-line handwriting Uyghur recognition system, preprocessing,feature extraction and combination of classification carried out a detailed description,and a certain number of the Uyghur handwritten word experiment, the specific workare as follows:1,this article an overview of the status of online Uyghur handwriting recognitiontechnology, a detailed analysis of the characteristics of online Uyghur handwritingbody, and combine commonly used in the HMM preprocessing and characteristicsextraction method, a certain degree of improvement the Uyghur text handwriting bodyof baseline Find and pre-classification based on HMM, extracted three characteristics.2, Delayed strokes for online Uyghur handwritten, a new treatment ideas,judged in the preprocessing phase first delayed strokes and extract certaincharacteristics, the results of the main strokes of the HMM classification, further fineclassification.3,the combination of classification strategy is to first use of the HMM Uyghurhandwriting body of the main strokes in accordance with the front of thecharacteristics of the classification, and then re-use extract the Uyghur handwritingbody of delayed strokes of the information, according to certain rules of the table afterthe HMM classification results, re-classification of this strategy to avoid the complexproblems caused by delayed strokes, step-by-step use of different information to achieve the recognition of handwritten Uyghur.4,the combination of the previously proposed classification strategy thinking,handwritten word samples of a certain number of Uyghur experiment, experimentsshow that this idea is feasible, although some still needs further exploration andresearch.
Keywords/Search Tags:online Uyghur, hidden Markov model, delayed stroke, combination ofclassification strategy, rule table
PDF Full Text Request
Related items