Handwritten character recognition based on SVMs for a multiplication table game

In this paper, we present a multiplication table game that automatically detects and recognizes handwritten digits written on the white board. The result of the multiplication of two numbers would not only be displayed on the monitor, but it would be also spoken by the Text To Speech (TTS). The proposed application is divided into three stages: 1) detect a white board; 2) segment candidate regions of handwritten characters on the white board; 3) recognition using SVMs. In the experiment, the MNIST database of handwritten digits is used to evaluate the performance of the proposed application and very promising results are achieved.