| With the rapid development of economy,China’s distribution network is constantly moving towards a strong smart grid.The State Development and Reform Commission and the State Grid Corporation are investing heavily in various kinds of power grid reconstruction projects,and the task of material submission in the early stage of the project is more arduous.At present,as a whole,the quality of material submission is still good because most of the staff are conscientious and responsible.However,there are still some shortcomings such as low manual efficiency and uneven quality of personnel,which require a machine to provide an auxiliary way.In recent years,with the continuous development of artificial intelligence,machine vision technology has made great progress.The computer can complete image recognition,surface detection,robot vision and size measurement by replacing human eyes,which greatly improves the work efficiency.In view of the insufficiency of material reporting in distribution network and the premise of the rapid development of machine vision,this paper proposes a material reporting system based on machine vision recognition and support vector machine learning,which aims to be applied in production engineering practice,save labor costs and improve work efficiency.In order to better recognize electronic drawings or drawings in engineering,the system first preprocesses the original image,obtains the gray and smooth image,then extracts and recognizes the target elements through edge recognition,image fitting and image cutting.At the same time,it compares the extracted targets according to the training data of the support training machine,and then compares the objects.Sorting and recognition of standard elements.Finally,according to the relevant standards of Liaoning Electric Power Co.,Ltd.,the research and judgment logic is formulated.The identified information is converted into material information table,and database is established through MFC platform.The material information is stored in the database,so that it can be directly extracted or transferred later. |