Hand-Written Recognition Words By Using Quad-Vector CodingBy Using Quad-Vector Coding

• Main Authors: Jhih-Yóu,Yi, 易志由
• Other Authors: Chien-Chou,Lin
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/89138513878089149886

碩士 === 樹德科技大學 === 資訊工程學系 === 94 === In this thesis, a novel recognition algorithm of handwritten Chinese characters is proposed for embedded systems. In general, the recognition of handwritten Chinese characters needs a respectable computation for extracting features of the input character and finding a match character in a large character database. The correctness rate of a recognition algorithm is proportional to its computation time. Since the embedded system usually has a fewer resource and lower computing power, it is not easy to port these complex algorithms to embedded systems. In order to reducing the computation of searching database, we proposed a quad-vector coding for each stroke of a Chinese character. Each Chinese character is coded as a string with quad values instead of eight values. The adopted matching algorithm is substring matching algorithm which is robust to error strokes for a complex character. The simulation is based on a personal computer with GRAPHIRE4 CTE-440 digital board made by Wacom. The recognition system is also implemented on an aJile 100Evbm embedded system. The recognitions of the two systems are efficient and the correctness rates are high up to 90%.