An augmented reality three-dimensional tracking registration method based on LINE-MOD template matching

Abstract

The invention claims an augmented reality three-dimensional tracking registration method based on LINE-MOD template matching. depth information is introduced, so that LINE- MOD template matching method can complete object identification and matching under any scale. And on the basis of an Improved LINE MOD template matching, through matching of the current view angle image and offline different view angle reference views, a key frame closest to the current view angle and a pose corresponding to the key frame are obtained. And the pose of the part is accurately estimated by using an interpolation method through the topological relation between the reference views. The method is superior to the texture-free target object tracking and pose estimation precision proposed by the Hinterstostat onthe ACCV, the real-time performance of the algorithm is good, and the tracking and registering requirements of the AR system can be well met.

Description

technical field [0001] The invention belongs to the technical field of augmented reality, and in particular relates to an augmented reality three-dimensional tracking registration method based on LINE-MOD template matching. Background technique [0002] Rigid object detection and pose estimation are the hotspots and difficulties of research in augmented reality technology in recent years. According to the appearance characteristics of the object, it can be divided into non-textured rigid object detection and pose estimation and textured rigid object detection and pose estimation. When the surface texture of a rigid object is rich, a stable number of feature points and feature descriptors can be extracted. Based on these stable feature points and feature descriptors, the pose of the object can be accurately detected and described. Among them, SIFT (Scale Invariant Feature Transform), PCA-SIFT, SURF (Speeded Up Robust Features) and other feature points and feature descriptors...

AI Technical Summary

Problems solved by technology

Among them, the LINE-MOD method is one of the most advanced template matching methods in recent years. The rigid object detection and pose estimation in this paper are inspired by LINE-MOD, but this method can only be used for rigid objects at a fixed scale. Detection and pose estimation, while this method has low accuracy for object pose estimation

Images