Object tracking method and apparatus

Abstract

Provided are an object tracking method and an object tracking device. The method comprises the following steps: an object graph is established, a group of objects having similar motion features are included in one object graph, each object is used as a vertex of the object graph, and the elasticity of each vertex in the object graph is calculated based on the representative motion feature of the corresponding object; an object group graph is established, each object group is used as a vertex of the object group graph, and the elasticity of each vertex in the object group graph is calculated based on the representative motion feature of the corresponding object group; the position range of each object group in a next frame is predicated based on the elasticity of the vortexes of the object group graph; the position range of each object in the next frame is predicated based on the elasticity of the vortexes of the object graph; and an object is detected in the next frame in a tracking time period based on the predicated position ranges of each object group and each object.

Description

technical field [0001] The present invention relates to an object tracking method and device, more particularly, to a method and device for object tracking based on a double-layer elastic graph. Background technique [0002] Currently, object tracking technology is widely used in various fields, such as intelligent transportation systems, image capture systems, intelligent monitoring systems, and so on. In simple terms, object tracking is to detect the position of the target object in the scene in each frame of the captured video image of the scene. [0003] Several object tracking methods exist in the prior art. For example, US Patent Application Publication US20120070034A1 discloses a method for detecting and tracking vehicles, which includes: a moving object detection stage, which detects candidate objects in a series of input video frames; a trajectory association stage, which uses a joint probabilistic matching framework to combine Existing trajectories are associated...

AI Technical Summary

Problems solved by technology

For example, in the previous scene, there are pedestrians walking on the road. However, in the next scene, it is difficult to identify the pedestrian in the captured scene image because the pedestrian walks next to the tree and is likely to be blocked by the tree.

Or, under the illumination of strong light, it may be difficult to recognize the vehicle in some scenes due to the reflection of the moving vehicle, resulting in inaccurate or slow object tracking results

Images