Dynamic target identification tracking method under complex background

Abstract

The invention discloses a dynamic target identification and tracking method under a complex background, and belongs to the field of space on-orbit services. The method comprises the following steps: collecting video data of a dynamic target motion area by using a depth camera, separating the video data into image data and voice data which are stored according to a time sequence, processing the image data by using an improved dynamic target detection method, and marking a candidate bounding box area of the frame of image by using a minimum rectangular frame; performing feature extraction, image segmentation, category determination and bounding box regression correction of the dynamic target by using the marked bounding box region; and obtaining the center of mass of the target through the detected type of the dynamic target and position and depth data measured by the depth camera, and conneccting the centers of mass obtained each time into a motion track, wherein the motion track is used for predicting the motion state of the dynamic target, and the mechanical arm grabs the dynamic target in real time. The method is small in calculation amount, high in calculation speed, good in real-time performance of grabbing the dynamic target by the mechanical arm and high in success rate.

Description

technical field [0001] The invention relates to the field of space on-orbit service, in particular to a method for identifying and tracking dynamic targets under complex backgrounds. Background technique [0002] With the rapid development of space on-orbit services, space manipulator technology has become one of the key technologies for implementing on-orbit service and maintenance tasks. There is still a huge gap in the use of manipulators to capture space non-cooperative targets, especially moving targets. challenge. In order to ensure the stable and efficient operation of the spacecraft in orbit in the complex space environment, the space manipulator is required to realize the on-orbit assembly of the spacecraft, refueling, and the capture, recovery, maintenance, and update of modules of the failed spacecraft. [0003] Because the space background environment where the spacecraft is located is complex and the area to be detected is wide, it is necessary to simplify the ...

AI Technical Summary

Problems solved by technology

[0007] The technical problem solved by the present invention is: Aiming at the problem that the dynamic target recognition and tracking method using the convolutional neural network under the complex background has a large amount of calculation and slow operation speed, a dynamic target recognition and tracking method under the complex background is proposed to realize the robot arm in the complex background. Real-time, efficient and fast capture of dynamic targets under complex backgrounds

Images