Depth video space plane detection method and system

Abstract

The invention discloses a depth video space plane detection method and system. The method comprises the following steps: an RGB image and a depth image of a current frame in a depth video are preprocessed; multiple groups of candidate points meeting conditions are selected from the point cloud data, each group of candidate points comprises three candidate points, multiple candidate planes are obtained through fitting, and an initial plane is obtained from the multiple candidate planes; based on the initial plane and the inner points thereof, spatial clustering is carried out on all the inner points, a class with the most points is obtained, and the points in the class are detection plane data of the detection plane of the current frame; and a current output plane is obtained according to the historical output plane data and the detection plane. The plane can be rapidly detected, the efficiency is high, and the stability is good.

Description

technical field [0001] The present invention relates to the technical field of depth video space plane detection, in particular to a depth video space plane detection method and system. Background technique [0002] In an application scenario of Augmented Reality (AR for short), it is often necessary to place a virtual object on a spatial plane (ie, a real plane object in a shooting scene) for interaction with a user. In addition, for daily 3D scenes, planes are the main component or form of items or objects, that is, planes are common features in daily 3D scenes, and can be used in image processing algorithms such as scene and object recognition to improve algorithm accuracy and speed, such as , in the SLAM algorithm, the plane can be used as a feature to improve the efficiency of pose estimation. [0003] Therefore, in order to obtain the plane information of the spatial plane, the plane detection methods in the prior art are divided into two categories according to the f...

AI Technical Summary

Problems solved by technology

However, the extraction of key points is often based on texture, and it is impossible to obtain a texture-free plane, resulting in missed detection.

[0005] The plane detection method based on the depth map can be divided into iterative fitting method, Hough transform method and clustering method according to the implementation method. Although the iterative fitting method and the Hough transform method are implemented in different ways, the results are similar. When there are multiple planes, the problem of false detection is still prone to occur; however, the existing clustering method has a huge amount of computation and cannot output the range of the plane, so the existing clustering method cannot be directly applied to AR devices

Images