A target tracking and positioning method based on azure Kinect

Abstract

The invention discloses a target tracking and positioning method based on Azure Kinect, comprising the following steps: step 1: extracting three-dimensional point cloud data of the target object through Azure Kinect; step 2: mapping the target object point cloud to a color map; step 3: establishing The "color-shape" model of the target object fused with depth information; step 4: collect color images and depth images in real time; step 5: use depth variance to segment the connected domain of the target object and similar color background; step 6: use the target object shape deviation Ratio, histogram Bhattacharyachian distance to complete target recognition; Step 7: If the target recognition is successful, calculate the target mass center P0, and complete target positioning with the depth mean value of the target mass center area; Step 8: Output the target object coordinates and update the target color histogram , length and width values ​​and depth values; Step 9: If target recognition fails, repeat steps 4 to 8. The invention can overcome the problem that the object tracking method based on the color features is easily interfered by similar color backgrounds.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a target tracking and positioning method based on Azure Kinect. Background technique [0002] Kinect is a 3D somatosensory camera launched by Microsoft. It has a color camera and a depth sensor. Since Kinect can obtain the depth information of objects, it also shows unique advantages in the field of target tracking. Azure Kinect provides developers with four types of sensor SDKs: depth, vision, sound, and orientation, including a 1-megapixel TOF depth camera, a 12-megapixel RGB high-definition camera, a 7-microphone circular array, and an inertial measurement unit IMU. A control access mode and custom installation to provide cross-platform low-level access to Azure Kinect device configuration and hardware sensing streams. [0003] The depth camera can choose wide or narrow FOV field of view when controlling access, so the data obtained in the actual space will be more ac...

AI Technical Summary

Problems solved by technology

[0007] At present, most of the target tracking methods using Kinect are based on color features. This tracking method needs to mark the tracking target in advance, which is inconvenient to operate. When the target object and the background color are similar, it is easily interfered by similar colors, and the tracking effect is not ideal.

Images