Recognition Technology, Based On The Mechanical Operations Graphics

The homework is an important link in the teaching process, the fast development in the modern education technique and the backwardness of the means of traditional homework technique makes the problem of the reformation of homework link means put forward. The long-distance homework system carries on the beneficial trial in the reformation of homework technique means, which can automatically recognize mathematics expression and solve the hard nut to crack for making corrections. However, for such foundation course of the science and engineering as the foundation mechanics and the like, usually some graphics contents are included in the homework subject and the student usually needs to draw the homework graphics. This present thesis carries on the relatively in depth research in the automatic correction of homework graphics.First, this thesis builds up an integrated answer graphs and the topological relations of student's graphs. The whole graphics element is regarded as an integrated whole, and a method is proposed for the construction of the table of properties of space relationships posed by 11 kinds of adjacency relations and 12 kinds of position relations. By using this table, an effective description can be made about the geometrical relationships between graphic elements.This thesis puts forward the standard of recognition and evaluation based on four types of criteria, namely the type of the graphic, the property of lines, parameters of the graph, and spatial relations including the adjacency relations and the position relations . The student, 5 drawing mistakes can be classified into matching mistake and property mistake. The recognition of the two kinds of homework mistakes corresponds to two recognition modules, namely the graphics match module and the property-examining module.This thesis aims at the four indefinite features of homework graphics , namely no plot jumping -off point, no plot sequence , no accurate size , and no fixed proportion scale , and puts forward the recognition arithmetic for the graphics that make use of the adjacency relations and the position relations, and the control strategy based on the couple of the benchmark graphics.This thesis designs the assistant recognition module for homework graphics, including the graphics mark recognition, the special designation recognition, the graded arithmetic of graphics recognition, and the module for the manifestation of grade points (including the mistake manifestation and the final score).On the basis of the above work, this thesis writes program for graphic recognition and evaluation, and develops the homework graphic recognition model in the Internet context. The trial run of the models turns out to be effective in recognition.