Off-line Handwritten Chinese Character Recognition Of Little Class Set Research Based On SVM-GA

Chinese characters are the results of Chinese cultural accumulation, which has a long history, and they represent Chinese people's wisdom. Chinese character recognition is a kind of pattern recognition with great difficulty. Additionally, concerning office automation and machine translation, off-line handwritten Chinese character recognition has a huge potential in application, which has caught attentions over the world. To draw a conclusion, the research on off-line handwritten Chinese character recognition not only is of great importance in theoretical value, but also contains significant value in use.This thesis mainly focuses on off-line unconstrained handwritten Chinese character recognition of little class set. This experiment selected 100 font types of Chinese characters which are commonly used from GB2312-80 database. Collecting 108 samples for each type, and the total number of samples is 10800, The main parts of this paper are as follows: Firstly, Special forms are designed to collect samples from people who are from different kinds of job, sex and degree, and their ages are between 18 and 60. Using the form can not only meet the requirement of recognition but also simplify some of preprocessing steps and improve the efficiency of the preprocessing.Feature extraction is a very important step for Chinese character recognition. According to the feature extraction of the off-line handwritten Chinese character recognition, four directional line element decomposition feature was extracted on the basis of elastic meshes partition method. Both stroke-based directional decomposition approach and a fuzzy sub-stroke extraction method are used to extract the four sub-strokes of hand-written Chinese character. They can not only avoid the disadvantage of contour based method being sensitive to different widths and distortions of the stroke, but also improve the blurring strokes and the losing of the important information of the stroke caused by skeleton based methods.Support Vector Machine is a high-performance learning machine on the basis of the statistical theory. SVM based on the information of limited samples to search for the best compromise between the complexity of the model and the learning ability, with a view to obtain the best generalization ability. There is a problem in the SVM that it depends on the performance of the parameter settings, including penalties and kernel parameters, but no suitable theory can guide to find adapted parameters. Combined adaptive genetic algorithm with SVM, design an automatic parameter selection method for SVM. This method selects crossover probability and mutation probability according to the fitness values of the object function, therefore reduces the convergence time and improves the precision of GA. This method was applied to off-line handwriting Chinese character recognition. Experimental results demonstrate an improvement of the generalization performance for SVM.