Web-based handwritten chemical formula recognition

Abstract

The emergence of touch-based mobile devices such as laptops and table PCs have become a prevalent stream nowadays. Consequently, the development of web based applications will have to be designed to support efficient input methods on mobile platforms. Handwriting input method has increased its popularity since it is able to express user's handwriting into meaningful expression that can be analyzed by computing device. An effective analysis on handwritten chemical expression recognition over the web has not been suggested. As such, the objective of this project is to develop an efficient web-based online handwritten chemical expression recognition system. To implement web-based recognition system, client-server architecture is proposed to handle request from various locations and comprises three major components: HTML5 web browser, computation engine and a standard web server. This project has specifically made the following contributions: (I) Developed and designed user interface using HTML5 technology for the chemical expression editor in the web browser by adopting progressive recognition approach for dynamic chemical expression recognition. (II) Investigated and optimized communication between web browser and web server by proposing enhanced web-based models to support better overall system architecture. (III) Evaluated the performance of the models and identified the limitations of current recognition algorithms. (IV) Developed a middleware to smoothen the data exchange and integration between web browser (client processes) and web server (server processes). (V) Extended the function of handwritten chemical expression tree format which is able to display standard web chemical description language such as SMILES representative.