Design And Implementation Of Machine Understanding Mathematical Geometric Diagram System

With the rapid development of artificial intelligence and a series of results achieved in various industries,it is becoming more feasible for artificial intelligence to learn about human thinking and think about problems.Among them,the education industry has also increased the application of artificial intelligence in its products,such as using machine learning technology to achieve intelligent understanding,intelligent answering,intelligent correction,and so on.For the intelligent understanding of geometric diagram problems in mathematical problems,based on image processing related technologies,the algorithm research of geometric diagram meaning understanding is implemented,and then an accurate,efficient and intelligent understanding system is implemented.The field of math problems is of great significance.Through investigating and researching a large number of SAT exercises and examination questions,an intelligent understanding mathematical geometric diagram system from image to predicate representation was developed.First,the OpenCV image processing API is used to preprocess the geometric diagram received by the system.Through experiments,the method based on binary image connected area labeling is used to segment the character features and geometric features in the diagram,and the segmented character information is appropriately optimized and merged.In the process of character recognition,the system uses the existing optical character recognition function of the Tesseract open source framework to train specific language sets for character recognition that often appear in geometric diagram to recognize characters that appear in images.In the process of geometric figure recognition,the Hough transform algorithm is used to identify and optimize the straight lines and circles in the figure,and then these basic geometric elements are composed into complex geometric elements.After obtaining the results of character recognition and geometric figure recognition,the system’s result merge function combines the two results according to their original position information to obtain the final understanding result of the input geometric image.Finally,the PyQt5 framework is used to design and implement the interactive interface of the system.Through the use of various technologies,a machine understanding of the geometric image system was realized,and 80 pictures containing geometric images were randomly taken from the SAT’s real problem as a test set,and functional and non-functional tests were performed on these 80 pictures.In the end,relatively satisfactory results were achieved.