Address event-driven real-time object tracking method for unstructured signals

Abstract

The invention discloses a real-time target tracking method for an unstructured signal driven by an address event. The specific implementation steps are as follows: (1) read data packets and perform preprocessing; (2) construct a plane matrix; (3) calculate each address Membrane voltage at the corresponding position of the event; (4) determine the effective event point; (5) construct the local space-time plane; (6) estimate the motion vector of the current address event; (7) determine whether the time stamp of the current address event is within the time interval ; (8) Modify the motion vector; (9) Determine the position of the target within the current address event sequence. The invention recovers the motion information and spatial structure of the object by estimating the motion vector of each read-in address event, correcting the motion vector, and improves the accuracy and stability of target tracking while reducing the calculation amount and increasing the speed.

Description

technical field [0001] The invention belongs to the technical field of physics, and further relates to a real-time target tracking method of address event-driven unstructured signal in the technical field of visual signal processing. The present invention can be used in the technical fields of intelligent transportation, human-computer interaction, video surveillance, robot navigation, virtual reality and the like to detect and track moving objects in three-dimensional space. Background technique [0002] Moving target tracking is one of the key issues in machine vision research. The main task is to estimate the trajectory of the target in the video, that is, to collect the video image sequence from the traditional image sensor, detect the moving target to be tracked, and then determine the target in each frame of image. However, traditional image sensors cannot capture fast-moving targets such as missiles and airplanes, and a large amount of redundant data brings huge compu...

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that because the search window used by the Camshift algorithm contains multiple candidate models with similar features, the Camshift algorithm cannot distinguish whether there is same-color interference, resulting in low tracking accuracy

The disadvantage of this method is that the construction of the mapping matrix requires many hyperparameters, which cannot adapt to the target tracking in any actual scene, and the generalization is low.

On the other hand, the Kalman filter continuously corrects the estimated value of the motion state through the input and output observation data of the model, combined with the actual motion parameters, which consumes a lot of memory and cannot be tracked in real time.

Images