Hand-drawn Electric Circuit Diagram Understanding Using 2D Dynamic Programming

A difficult task of a sketch understanding system is that it should always try to balance between the drawing freedom and the complexity of recognition. Most online existing works are based on the assumption that people will not start to draw a new symbol before the last one has been finished. As obviously it is not always the case, we propose in this paper a method which relaxes this constraint. The proposed methodology relies on a twodimensional dynamic programming technique (2D-DP) allowing symbol hypothesis generation, which can correctly locate symbols even when they are drawn temporally overlapped with each other. A neural network classifier, embedded in the 2D-DP, is used to label each hypothesis. Besides, tolerant connectivity constraints are introduced to help increasing the performance as well as the efficiency. Experiments show that this novel approach achieves an accuracy of more than 90 percent.