Research And Implementation Of Online Handwritten Sanskrit Tibetan Feature Extraction Methods

With the popularity of the information age and electronic equipment, on-line handwriting has become a simple and convenient way of life, which is on-line handwriting proposed more stringent requirements. Compared with the modern Tibetan, Sanskrit Tibetan-shaped complex, large character sets, like the word more. Sanskrit Tibetan online handwritten recognition is a multi-class and multi-level pattern recognition problem, is now in the initial stage of experimental research. In recent years, many handwriting recognition technology used in the online Tibetan handwriting recognition, and achieved good results, which makes Sanskrit Tibetan online handwritten recognition technology has important theory significance and broad market prospects.Character recognition feature extraction is the first step in the arithmetic processing, the success of a recognition system is often determined by the features it uses and defines. This article summarizes the classical method of handwritten characters feature extraction technique, analyzes and discusses the advantages and disadvantages of the algorithm in detail. Mainly completed the following aspects of the work:First, study the structural characteristics of Sanskrit Tibetan, learn from other mature character recognition technology, find out the representative feature extraction method. Proposed extraction based on Gabor filter characteristics, features and character direction merge feature strokes.Second, according to the feature extraction algorithms and models, design and implement the feature extraction method. Construction of Sanskrit character recognition system, includes the training phase and the testing phase two subroutine. The system includes feature extraction module, LDA feature selection module, MQDF identification and classification module library modules. After the system constructed, a lot of training and testing samples are used to debug and improve the system, and the system can achieve the design goal.By the above-described methods and ideas, research and implement of Sanskrit Tibetan online handwriting recognition system. Experiments show that the Sanskrit character feature extraction method is simple and effective, achieves the basic objectives.