3D model construction method based on non-isocentre C-shaped arm 2D projected image

Abstract

The invention discloses a 3D model construction method based on a non-isocentre C-shaped arm 2D projected image. The 3D model construction method based on the non-isocentre C-shaped arm 2D projected image comprises the following steps that a 2D ball-lattice image is revised based on a 2D ball-lattice image; internal / external parameters in a C-shaped arm imaging system in any posture are confirmed based on the revised 2D ball-lattice image; displacement parameters of a C-shaped arm rotary center in any posture are confirmed based on the revised 2D ball-lattice image; a 2D anatomy image within a missing angle is virtually reconstructed based on the revised 2D ball-lattice image, and a virtual 2D anatomy image is obtained; the 2D anatomy image and the virtual 2D anatomy image are combined to form a complete full-angle 2D projected image set, and a 3D model is constructed. The 3D model construction method based on the non-isocentre C-shaped arm 2D projected image is based on ordinary C-shaped arm machines possessed by the majority of hospitals at present, and reconstruct a skeleton 3D model.

Description

technical field [0001] The invention relates to computer imaging technology, in particular to a method for constructing a 3D model based on a non-isocentric C-arm 2D projection image. Background technique [0002] At present, the research methods of 3D reconstruction are mainly divided into two categories: one is active methods, such as laser scanners, which can obtain higher reconstruction accuracy; the other is passive methods, such as methods based on image sequences, which The 3D information of the scene is reconstructed by using the 2D image of the scene under natural lighting. [0003] In the 3D reconstruction system based on image sequences, since the 3D data obtained at a certain viewing angle is only the 3D information corresponding to a certain local surface of the reconstructed object, in order to realize the 360-degree surface reconstruction of the object, it is necessary to collect Reconstruct multi-view images of objects. There are currently two methods. One ...

AI Technical Summary

Problems solved by technology

There are currently two methods. One method is to use the turntable to rotate and reconstruct the object, and record the angle of rotation. This method can obtain higher reconstruction accuracy, but requires high-precision rotation angle control; the other method is to reconstruct the object around the camera. Objects are acquired in a regular or free manner from multi-view images. The hardware of this method is relatively simple, but there are certain limitations in application, that is, the image acquisition and object reconstruction are divided into two independent steps, so the objects missed during acquisition cannot be found in time Viewing angle, which may lead to the loss of three-dimensional information on some viewing angles of the object

Images