A target object tracking method, device, computer equipment and storage medium
By performing depth fitting on the original image and fusing it with point cloud data, the features of the target object are extracted, which solves the problem of false detection and missed detection of lidar under adverse weather conditions and realizes efficient tracking of the target object.
CN116681730BActive Publication Date: 2026-05-29CHINA AUTOMOTIVE INNOVATION CORP
Patent Information
- Authority / Receiving Office
- CN Β· China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA AUTOMOTIVE INNOVATION CORP
- Filing Date
- 2023-06-14
- Publication Date
- 2026-05-29
AI Technical Summary
Technical Problem
In adverse weather conditions, point cloud data detection by lidar is prone to false detections and missed detections, affecting the real-time tracking performance of targets.
Method used
A depth image is obtained by depth fitting the original image, and then fused with point cloud data to extract the features of the target object. The image is then updated by combining the tracking trajectory at historical moments, thereby improving the tracking accuracy of the target object.
Benefits of technology
In scenarios where lidar detection is not feasible, it improves the accuracy of target identification and tracking, reduces false matching and missed detection events, and improves target tracking efficiency.
β¦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN116681730B_ABST
Abstract
The application relates to a target tracking method and device, computer equipment, a storage medium and a computer program product. The method comprises the following steps: acquiring original images and point cloud data at a current moment in a target scene; performing depth fitting on the original images to obtain depth images of the original images; performing data fusion on the depth images and the point cloud data to obtain fusion data at the current moment; extracting features of a target from the fusion data at the current moment; and updating a tracking trajectory of the target extracted at a historical moment based on the features of the target extracted at the current moment to obtain the tracking trajectory of the target extracted at the current moment. The method can better extract feature information of a three-dimensional target in a scene that is not conducive to laser radar detection, improve the accuracy of correlation matching, reduce false matching and missed detection in multi-target matching caused by the fact that a laser radar cannot reach a normal working state, and improve target tracking efficiency.
Need to check novelty before this filing date? Find Prior Art