Line Handwritten Chemical Formulas To Address Key Issues Research

As a hot area, research on formula processing faces a lot of difficulties due to its complicated structure, expressive meanings and rich forms. With the development of mobile devices, the online handwriting input mode provides a novel method to handle this problem. The related works are based on several research areas which contain pattern recognition, document image processing, human-computer interface and so on.Mathematical and chemical formulas are two most important fields of science. Many researchers have focused a lot on mathematical formula processing and related software products have been developed. However, research on chemical formula is still emerging and has not yet matured. In this paper, we focus on the chemical formulas and symbols written on a Tablet PC and propose some methods to recognize, analyze and understand the object. Specifically, the research work in this thesis mainly includes:1. Recognition of online handwritten chemical symbolsIn view of the freestyle of handwriting and various deformations of online handwritten chemical symbols, we firstly readjusted the spatial distribution of the sampling points based on the reinforcement of symbol quality and standardization of the font appearance, and then designed a two-level classifier for recognition of 102 kinds of chemical symbols.2. Structure analysis of chemical formulasIn this paper, we segment a chemical formula using its spatial information. Some techniques such as projection, connected component and character's width are composed to analyze the formula's structure. Specifically, we handle the breakout and connected strokes in formulas and the experiment results show that the approach proposed in this paper works well on both normative and unconstrained data.3. Understanding of chemical formulasIn this paper, we extract the main components from a formula based-on the chemical grammar. The processing sequence contains generator extraction, separator extraction, text area location and organic loop recognition. We locate the atoms in a loop and find the bonds among them. An organic loop is stored as an adjacency matrix and the formula is transformed into a chemical grammar tree, which will be used for display and information retrieval. At last, we check the semantic validity using4. Pen-based HCI for chemical applicationsWe design 15 gestures for recognition, analyzing and understanding of online handwritten chemical formulas. Two principles are considered, which are the designed gestures must be accepted by users and easily distinguished by computer. Besides, we can demonstrate some simple chemical experiments in the discussed interface.5. Setting up a chemical datasetWe set up a general dataset for the research on processing of online handwritten chemical formulas. In this paper, we introduce the whole progress and analyze both characteristics and properties of the data.