电气原理草图语义理解系统开发

Development of the Semantic Understanding System of the Electric Principle Sketch

摘要:

针对目前市面上现有的手绘草图识别系统对用户的手绘习惯仍有较大限制的现状,提出了一种基于图元以及图元数量的单个电气元件的识别方法；系统使用MATLAB做为软件开发平台；首先通过在线采集的方式来获取手绘图每个笔画的始终点坐标以及每个笔画上所有的采样点的坐标,然后通过简单的平滑处理以及去重处理来降低因输入时的抖动噪音等因素对识别结果造成的影响；其次针对识别对象的结构特征相对稳定的特点,并且考虑到图库中所有的电气元件也都是由相同的几个基本图元组成的,实现了基于图元以及图元数量的单个电气元件的识别方法；实验结果表明,系统能够有效地减少用户在手动绘制时的限制,降低了因用户手绘习惯而造成的识别出错率；并且能够实现较高准确率的识别效果。

Abstract:

At present, there are still some limitations on the existing hand drawn sketch recognition systems. We presented a recognition method of a single electrical components based on primitives and the number of primitives. The system uses MATLAB as software platform. Firstly, we get the coordinates of each stroke of the hand drawn map and the coordinate of the sampling points on each stroke. Then, we reduce the influence of the noise by simple smoothing and repeat removing processes. In addition, our recognition object's structure features are relatively stable, and all electrical components are consisting of a single or some primitives. So we realized the recognition method based on primitives and the number of primitives. Experimental results show that the system can effectively reduce the limitations of users , and reduce the identification error rate. In addition, this system can achieve a higher accuracy of recognition results.