Design And Implementation Of Recogniton Of Online Handwritten Mathematical Expression

With the development of information technology, the demand of digitalmathematical documents is increasing. Mathematical document contains a large numberof mathematical expressions which should be recorded in the computer. But, becausdeof the complex2D results of mathematical expressions, it’s harder to be recorded thanordinary words. With the popularity of handwriting input device in recent years, on-linehandwritten mathematical expression is rapidly becoming a promising way to recordmathematical expression.The purpose of the thesis is to design and implement an on-line handwrittenmathematical expression recognizer, which is based on the related theory andtechnology of mathematical expression recognition. The main contents are as follows:1) A real-time recognition method, which is based on strokes affected areaspositioning, has been improved. Some restrictions are needed by the original method.There are some limits in traditional stroke recognition, which affects the verticalstructure of mathematical expression when processing symbols, for example, ahorizontal stroke must be recorded first. Based on the existing method, we divide thearea around the characters into ten domains and propose a method to judge the spcialrelationship between two characters. In recognition process, the result of recognition isstored in mathematical expression tree. According to the proposed structural analysisalgorithm, relative nodes affected by the new input stroke can be found by travelling themathematical expression tree. Mathematical tree is updated through the relationshipbetween the new input stoke and its relative nodes.And this is the way how theproposed recognition of mathematical expression works.2) In the VS2010development environment, Silverlight technology and MathMLtechnology are adopted to develop the recognizer proposed here. The recognizerprovides functions to recognize the single character and mathematical expression aswell as to modify characters and expressions dynamically. In order to achieve anintuitive recognition result, a display function of mathematical expression is added. Thisfunction can visually display the complex two-dimensional structure of mathematicalexpression. The recognizer is tested on the test cases, which including different spatialstructures. Experiments show that the recognizer performs very well on simplehorizontal relationship and simple nested structure, but just OK when performs on too complex nested structure.