A method, system and storage medium for building an optical model based on deep learning

Abstract

The present invention provides a method, system and storage medium for building an optical model based on deep learning, including: collecting a two-dimensional plane image of an operating tool and then screening out a characteristic pattern of a marking point, constructing a characteristic recognition self-learning model to obtain a characteristic recognition algorithm, calculating the three-dimensional coordinate data of the marking point of the operating tool, establishing a point cloud data model of the marking point, marking the point cloud corresponding to the marking point in the operating tool to be modeled in the point cloud data model, constructing a characteristic marking self-learning model to obtain a characteristic marking algorithm, collecting a two-dimensional plane image of all the marking points of the operating tool to be modeled, and obtaining an optical model of the operating tool to be modeled based on the characteristic recognition algorithm and the characteristic marking algorithm. By constructing a characteristic recognition self-learning model and a characteristic marking self-learning model to perform characteristic recognition and characteristic marking, and then collecting the two-dimensional plane images of all the marking points in real time, the optical model of the operating tool to be modeled can be obtained according to the characteristic recognition algorithm and the characteristic marking algorithm.

Images