Representing and Characterizing Handwritten Mathematical Symbols through Succinct Functional Approximation
Authors:
- 1 September 2007
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Ninth International Conference on Document Analysis and Recognition (ICDAR 2007)
- Vol. 2 (15205363) , 1198-1202
Abstract
We model on-line ink traces for a set of 219 symbols to "best fit" low-degree polynomial series. Using a collection of mathematical writing samples, we find that in many cases this provides a succinct way to model the stylus movements of actual test users. Furthermore, even without further similarity-processing, the polynomial coefficients from the writing samples form clusters which often contain the same character as written by different users. We find this style of characterization to be an attractive tool due to the suitability of the representation to computation and mathematical analysis.Keywords
This publication has 8 references indexed in Scilit:
- Hybrid feature extraction and feature selection for improving recognition accuracy of handwritten numerals2005
- HMM-based handwritten symbol recognition using on-line and off-line features2002
- Multiresolution recognition of offline handwritten Chinese characters with wavelet transform2002
- A local search approximation algorithm for k-means clustering2002
- A study of representations for pen based handwriting recognition of Tamil characters1999
- A novel approach to optical character recognition based on ring-projection-wavelet-fractal signatures1996
- Wavelet descriptors for multiresolution recognition of handprinted charactersPattern Recognition, 1995
- The state of the art in online handwriting recognitionIEEE Transactions on Pattern Analysis and Machine Intelligence, 1990