Object pose detection method and device

Abstract

The invention discloses an object pose detection method and device. The object pose detection method comprises receiving an original image acquired by a camera, extracting and obtaining a marked code image from the original image, wherein the marked code image is an image of a marked code attached to the object to be detected; obtaining the pose of the marked code through calculation in dependence on the marked code image and a pre-generated pose calculating model, wherein the pose of the marked code includes the rotation amount and the displacement amount of the marked code relative to the reference pose; obtaining the current pose of the object to be detected through calculation in dependence on the pose of the marked code. According to the invention, the object pose detection process can be simplified, the efficiency of the detection process can be improved, and the accuracy of the detection result can be improved.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a method and device for detecting the pose of an object. Background technique [0002] In the field of intelligent robots, in the process of carrying objects, intelligent robots need to detect and judge the current position and angle of objects relative to themselves. The problem of pose estimation refers to the process of using feature correspondence information to estimate the relative rotation and translation between two space coordinate systems, which is the core technical problem to be solved to realize the function of intelligent robot to carry objects. Pose estimation is an important fundamental problem in computer vision, computer graphics, and photogrammetry. [0003] There are three main methods for estimating the pose at present: the pose estimation method based on feature points, the pose estimation method based on the model, and the pose estimation method based on...

AI Technical Summary

Problems solved by technology

[0006] Although the pose estimation method based on feature points occupies the mainstream position in visual odometry, this type of method still has the following disadvantages: First, the extraction of key points and the calculation of feature descriptors are very time-consuming

The disadvantage is that the accuracy and continuity of pose estimation cannot be guaranteed due to the inability to obtain the dense sampling required for continuous estimation in high-dimensional space

[0015] In particular, for chair pose estimation, in addition to the above shortcomings, the learning-based pose estimation method needs to collect and label a large number of samples for various chairs, which also requires a lot of manpower and material resources

Images