Font Size: a A A

Hand-drawn Geometry Figures Recognition For Automatic Machine Solving

Posted on:2020-02-19Degree:MasterType:Thesis
Country:ChinaCandidate:K L ShiFull Text:PDF
GTID:2428330578452061Subject:Computer technology
Abstract/Summary:PDF Full Text Request
This paper proposes a hand-drawn geometric figure recognition algorithm based on the automatic machine solving system.This algorithm solves the hand-drawn geometric image recognition of the junior high-level plane geometric proof problem.The recognition result can be combined with the text information to answer the question in the automatic machine solving system.Recognizing the user's hand-drawn geometry and outputting recognition results to help the user understand the expression of geometric information and the transition of geometric figure from text to figure.This paper is divided into two parts,one is the identification of auxiliary marks,the auxiliary marks including the letters of the points,and the right or digital marks indicating the angle,mainly including preprocessing,symbol identifier recognition;the other is the identification of plane geometric entities,reasoning and expression,mainly to identify the geometric entities and auxiliary lines in the graph to obtain geometric information.The first part is the identification of the auxiliary mark.Firstly,through the preprocessing,the connect component labeling algorithm is used to distinguish the geometric figure and the auxiliary mark area.Then,the mark is identified by CNN,this part is prepare for the subsequent recognition of geometric figures.The second part is the identification of plane geometric entities.Firstly,the circle is detected by Hough gradient transform to obtain the circle s coordinates of the center and radius.Then the position of the turning point matches the position of the symbol identifier.According to the position calculation formulas,the geometric position information is obtained;the closed figure is input into the recognition tree to obtain the combination of the basic geometric entities and the information of the auxiliary lines,combined with the properties of the basic geometric entities,get the complementary geometric information;finally,output the expres-sion.By analyzing the characteristics of hand-drawn geometric figures,this paper identifies the an-gle marks in hand-drawn geometry figure,corrects the angle errors caused by the non-standard hand-drawn geometry figure,and constructs the recognition tree to identify the closed graph-ics,combined with the properties of the basic geometric entities.Finally,geometric informa-tion expression is obtained.
Keywords/Search Tags:hand-drawn geometric figure, Hough gradient transform, CNN recognition, recognition tree, information extraction
PDF Full Text Request
Related items