Recognition of Handwritten Mathematical Symbols with PHOG Features
نویسندگان
چکیده
Converting handwritten formulas to LaTex is a challenging machine learning problem. An essential step in the recognition of mathematical formulas is the symbol recognition. In this paper we show that pyramids of oriented gradients (PHOG) are effective features for recognizing mathematical symbols. Our best results are obtained using PHOG features along with a one-againstone SVM classifier. We train our classifier using images extracted from XY coordinates of online data from the CHROHME dataset, which contains 22000 character samples. We limit our analysis to 59 characters. The classifier achieves a 96% generalization accuracy on these characters and makes reasonable mistakes. We also demonstrate that our classifier is able to generalize gracefully to phone images of mathematical symbols written by a new user. On a small experiment performed on images of 75 handwritten symbols, the symbol recognition rates is 92 %. The code is available at: https://github.com/nicodjimenez/
منابع مشابه
Prototype Pruning by Feature Extraction for Handwritten Mathematical Symbol Recognition
Successful mathematical handwriting recognition will require recognizers for large sets of handwritten symbols. This paper presents a recognition system for such handwritten mathematical symbols. The recognizer can provide a component of a handwritten interface for computer algebra systems such as Maple. Large sets of similar symbols present new challenges in the area of handwriting recognition...
متن کاملRecognition of Online Handwritten Math Symbols using Density Features
We explore the application of shape context features and 2D histograms for recognition of handwritten math symbols. Density features are used to extract the contextual information from the shape of the symbol. Both, the shape context feature and 2D histograms are implemented using three different strategiesCounting points, Parzen windows and Inverse distance. Using the dataset from the Competit...
متن کاملStructural analysis of online handwritten mathematical symbols based on support vector machines
Mathematical expression recognition is still a very challenging task for the research community mainly because of the two-dimensional (2d) structure of mathematical expressions (MEs). In this paper, we present a novel approach for the structural analysis between two on-line handwritten mathematical symbols of a ME, based on spatial features of the symbols. We introduce six features to represent...
متن کامل