Research On Characters Extraction And Recognition Method Of Off-line Handwritten Chinese Character Based On Procedure Neural Networks

Off-line handwritten Chinese character recognition is a popular research pointin recognition field, and it has important theoretical significance and practicalvalue. Chinese character recognition is a super large-scale classificationandrecognition problem, caused by the randomness of writing and irregularities,Chinese characters have a deformation and connection with each other, but nohandwriting process information. Consequently, Off-line handwritten Chinesecharacter recognition is one of difficult problems in Chinese character recognitionfield. In this paper, the theory of procedure neural network and the process ofhuman in congizing characters are connected, characters extraction and recognitionmethod of off-line handwritten Chinese character based on procedure neuralnetworks is studied, the new machine’s knowledge of off-line handwritten Chinesecharacter is explored.The main works in this paper are as follows:1. A way of fault-tolerant features for Chinese characters is analyzed andproposed. By imitating the writing habits of Chinese character, the fault-tolerantcharacteristics of Chinese characters are respresented as the features of stroketype’s number, stroke type’s location, topology type’s number, the information ofstroke order, a date table of off-line handwritten Chinese character features isgiven.2. The model of process neural network is constructed for extracting strokecharacteristics of Chinese characters. The two-dimensional lattice image ofhandwritten Chinese characters are transformed into a sequence of spatialaggregation and polymerization time, because of the capability of dealing withspace-time information of procedure neural network, the model of procedure neuralnetwork is established for extracting stroke characteristics of Chinese characters,and particle swarm optimization algorithm is used to train the network model. Thetrained process neural network is used to extract the characters of Off-linehandwritten Chinese character, then the data table of Off-line handwritten Chinesecharacter features is constructed. 3.The fault-tolerant matching rules of Off-line handwritten Chinese characterare studied. The fault-tolerant matching rules are made by imitating the humanability of learning and comparison judgment and analysis. The data tables ofOff-line handwritten Chinese character features and sample library Chinesecharacter features are matched.Choose handwritten Chinese sample characters in SCUT-IRAC HCCLIBwhich have every kind structure. MATLAB7.0simulation software platform isselected to test the characters extraction and recognition method of Off-linehandwritten Chinese character based on procedure neural networks. Experimentalresults show that the method preferably eliminate the distortion impact on featureextraction,and the recognition rate is high.