Kinect Depth Reconstruction Method Based on Camera Motion and Image Shading

Abstract

The invention discloses a camera motion and image brightness-based Kinect depth reconstruction algorithm. The algorithm comprises the steps of 1) uploading data collected by Kinect to a computer through a third-party interface under the condition that a Kinect depth camera and an RGB camera are calibrated and aligned; 2) recovering a three-dimensional scene structure and a motion track of the kinect RGB camera from an RGB video sequence, and obtaining a relationship between point cloud and a camera motion; and 3) reconstructing image depth by utilizing brightness status information of an image in combination with the relationship between the point cloud and the camera motion, obtained in the step 2). According to the algorithm, the depth camera does not need to be improved physically, a complex apparatus combination does not need to be designed, and an illumination calibration step which is often used in a conventional depth reconstruction method, generally only can be limited in laboratory conditions, does not have a practical application value and is complex and strict in condition is not needed, so that compared with the conventional method, the algorithm has higher practical application value and significance.

Description

technical field [0001] The invention relates to the field of depth reconstruction in computer image processing, in particular to a Kinect depth reconstruction method based on camera motion and image light and shade. Background technique [0002] With the introduction and promotion of some relatively cheap civil depth cameras in recent years, such as Microsoft Kinect and AsusXtion Pro, depth information is widely used in somatosensory games, real-time 3D reconstruction, augmented reality, virtual reality and other fields. The application of information has become an important support for the development of new human-computer interaction methods. However, most of the more popular civilian depth cameras currently on the market have the problems of insufficient depth detection accuracy and too much interference noise, which seriously affects the quality of application products based on depth information. Therefore, how to obtain more accurate depth information is of great signi...

AI Technical Summary

Problems solved by technology

[0006] The object of the present invention is to overcome the deficiency that the depth detection accuracy of existing civil depth camera is not enough, provide a kind of Kinect depth reconstruction method based on camera motion and image light and shade, this method does not need to carry out physical improvement to depth camera, does not need It needs to design a complex combination of devices, and does not need the complex and harsh illumination calibration steps that are often used in traditional depth reconstruction methods, which are generally limited to laboratory conditions and have no practical application value. Compared with traditional methods , has greater practical application value and significance

Images