Research on 3D Model Generation for Shaft Parts Based on Image Recognition

It has been nearly half a century since the proposal of transforming 2D engineering drawings to 3D solid models, but there are still controversies on the harsh hardware requirements, excessive constraints during conversion and low efficiency. Directing at the above-mentioned issues, a method of transforming 2D engineering drawings to 3D solid models based on image recognition is proposed by taking shaft parts as an example. In this paper, the engineering drawings were firstly reprocessed through scanning, denoising, segmentation, and thinning; secondly, the characteristic parameters of shaft parts were obtained based on the recognition of corners and circles, as well as the calculation of the BP neural network; finally, the transformation from 2D engineering drawings to 3D solid models for shaft parts was realized. This paper has found a new goal for image recognition and discovered a new method for transforming 2D engineering drawings to 3D solid models, which is of certain guiding significance to subsequent research.