A three-dimensional model classification method and retrieval method

Abstract

The invention discloses a three-dimensional model classification method and a retrieval method. The classification method includes establishing a neural network model; generating a training data set; inputting the training data set into the neural network model; and inputting the three-dimensional model to be tested into the neural network model Among them, the neural network model outputs classification results; the retrieval method further includes calculating the Euclidean distance between the model descriptor of the 3D model to be tested and the model descriptor of each template 3D model; the template 3D model corresponding to the smallest Euclidean distance is is the retrieval result of the 3D model to be tested. In the neural network model of the invention, the output end of the convolutional layer of the Nth structure is respectively connected to the input end of the batch normalization layer of the N+1st structure and the input end of the batch normalization layer of the N+2th structure, increasing The input of each structure can effectively improve the accuracy of data processing by the neural network model, and then improve the classification and retrieval accuracy of the three-dimensional model to be tested in this method.

Description

technical field [0001] The invention relates to the technical field of three-dimensional model classification retrieval. Background technique [0002] With the rapid development of multimedia technology and computer hardware and software, massive 3D models are widely used in many fields such as industrial design and intelligent production. Management is an urgent need in various fields. Effective classification management of 3D models can improve design productivity, increase the reuse rate of original models, and reduce time and labor costs. [0003] In recent years, artificial intelligence has achieved rapid development in both theory and application. In particular, the introduction of the theoretical method of deep learning has solved many difficult problems in artificial intelligence, and at the same time realized the promotion and application of artificial intelligence to various fields. to go. In the research field of 3D model classification and retrieval, improving...

AI Technical Summary

Problems solved by technology

At present, most of the research work on 3D model classification and retrieval focuses on rigid body models, and its algorithms are difficult to apply to non-rigid body models with variable poses. The main reasons are: (1) Non-rigid 3D models can produce a variety of irregular poses. Large differences lead to difficult classification and retrieval; (2) inconsistencies in scales increase the difficulty of classification and retrieval; (3) two 3D models may appear very similar in a certain posture, resulting in classification and retrieval difficulties. drop in accuracy

Images