Fast automatic registration method and system based on multiple kinects

Abstract

The present invention is based on the fast automatic registration method and system of multiple Kinects. Three Kinect sensors are arranged around the human body. Each Kinect sensor acquires the depth data of the human body, the color data of the human body and the bone point data of the human body. The internal reference of each Kinect is used to obtain the initial transformation matrix between two adjacent Kinect sensors; the human body depth data and human color data are processed, and the internal reference of each Kinect is used to generate human body point cloud data; the initial transformation matrix is ​​used to transform the human body point cloud Convert the data to a unified world coordinate system; use the bone point data to classify the human body point cloud data, and calculate the rigid body registration transformation matrix; use the least square method to optimize the rigid body registration transformation matrix to obtain the final transformation matrix; through the final transformation matrix, The human body point cloud data obtained by the three Kinect sensors are registered together to obtain the final registration result.

Description

technical field [0001] The present disclosure relates to the technical field of image processing, in particular to a fast automatic registration method and system based on multiple Kinects. Background technique [0002] Reconstruction of physical objects and scenes can help people process information better. With the development of technologies such as computers and somatosensory devices, reconstruction of real objects and scenes is playing an increasingly important role in the fields of robot control, medical care, education, and computer applications. [0003] With the release of Kinect, a somatosensory device with fast imaging speed, low price, easy operation, and the ability to obtain object depth information, how to quickly and realistically reconstruct real objects and scenes with Kinect is becoming a hot research issue. The key to the scene is how to quickly and accurately register multiple Kinects. Among the algorithms based on depth image registration, the iterativ...

AI Technical Summary

Problems solved by technology

[0007] (1) The algorithm assumes that one point set is a subset of another point set, which is often difficult to meet;

[0008] (2) In the process of searching for corresponding points, the algorithm has a high computational cost;

However, judging from the actual application status and results of calibration, sometimes there will be shortcomings such as low calibration accuracy, unstable calibration system, complicated, cumbersome, and time-consuming calibration operations, resulting in unsatisfactory calibration results.

[0015] Due to the lack of versatility and accuracy of traditional 3D registration methods, it is difficult to meet the various needs of users. The reconstruction process of the model, the complicated process and the time consumed are all issues that need to be paid attention to

Images