Object recognition method based on combination of 2D and 3D features

• Main Authors: CHEN XUEFANG, XIAO WUYI, LIU HUAZHU
• Format: Patent
• Language: Chinese; English
• Published: 11.09.2020
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention discloses an object recognition method based on combination of 2D and 3D features. The method comprises: obtaining point cloud data for a single model object, and carrying out the reconstruction of a three-dimensional model; performing dimension reduction on the point cloud data acquired in the step 1 to obtain an image of which the dimension is reduced into two dimensions after point cloud processing, and preprocessing an acquired two-dimensional image data set; training a classification network based on the MobileNetV2; and carrying out online object recognition. According to the method, the shielding problem in the prior art is solved by extracting the three-dimensional local features for classification, and then two-dimensional classification and three-dimensional recognition are carried out in sequence, so that the calculation difficulty is reduced, the real-time recognition efficiency is improved, and the final recognition accuracy is high. 本发明公开了一种一种基于2D与3D特征结合的物体识别方法，对单个模型物体，获取点云数据进行三维模型