Three-dimensional model classification method based on feature fusion and voting mechanism

• Main Authors: ZHANG CHUNXIANG, YAN SHAOKANG, GAO XUEYAO
• Format: Patent
• Language: Chinese; English
• Published: 21.07.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention relates to a three-dimensional model classification method based on feature fusion and a voting mechanism. The method comprises the following steps: firstly, preprocessing a training model and a test model in a three-dimensional model data set to obtain a group of two-dimensional views representing the three-dimensional model, and calculating to obtain low-layer shape features of the views: D1, D2, D3, a Zernike moment, a Fourier descriptor and a KSP feature; and realizing extraction of view high-level semantic features by using a RegNet design space. And fusing the low-level shape features and the high-level semantic features of the view obtained from the training model by using Concat to obtain training data. And fusing the low-level shape features and the high-level semantic features of the view obtained from the test model by using Concat to obtain test data. And then, optimizing the RegNet design space and the LSTM by using the training data, extracting discrimination features from fusion f