Generation Of Circuit Diagram Based On The Sketch Recognition

In recent years, around fast, natural and efficient sketch recognition based on the hand-painted way, extensive research has been carried out in CAD field by domestic and foreign researchers. CAD technology is often used in electrical drawings to shorten design cycle, reduce design cost and improve design quality. However, most CAD systems do not support hand-painted design and concept design. Compared to print graphics, hand-painted sketch has optional and highly irregular features. However, the features have constrained the development of hand-painted recognition technology all the time. This paper focuses on the on-line recognition of hand-painted circuit diagram and integrates hand-painted recognition technology with the CAD system to develop hand-painted recognition module and promote the development of conceptual design.In the pre-processing stage of line element recognition, removing the same points, simplifying, smoothing points and stroke segmentation have been done to sampling points. This paper classifies the line element according to the possibility value of fitting line and circle based on the least square method. To the straight line which has general or specific slope, different method was adopted to fit straight line and extract the endpoint of the straight line. Further recognition algorithm has been studied to recognize circle and semicircle which belong to the circular arc.The description and extraction of geometrical feature is a prerequisite for the graphics element recognition. This paper focuses on the conditions and the algorithms to extract geometrical relationship between two line elements. Template matching method was used to complete the graphics element recognition.This paper studies overall recognition for hand-painted circuit diagram in theory. The constraints were added according to the context sketch information to output the graphics element.The proposed recognition algorithm has been verified by developed CAD recognition system module. Results show that the method based on geometrical relationship and pattern recognition is simple, targeted, high recognition accuracy, high speed. These algorithms can also be used in other areas of the sketch recognition.