AER image sensor-based real-time multi-object tracking method and apparatus

Abstract

The invention relates to the technical field of image processing and provides an AER (Address-Event Representation) image sensor-based real-time multi-object tracking method and apparatus for processing information of an image acquired by an AER image sensor in real time, detecting multiple objects simultaneously and accurately providing information of position parameters and the like of each object. The method comprises the steps of acquiring data by utilizing the AER image sensor, performing real-time processing, monitoring light intensity change, performing output when the change reaches a specified threshold, and filtering out useless background information by utilizing the sensor to reduce data transmission quantity and redundant information; transmitting sensor data into an arbitration mechanism in parallel, judging a transmission sequence through the arbitration mechanism, and forming a time tag, wherein the time tag contains time t, polarity pol and position information p=(x,y)<T>, and the value of pol is +1 or -1; and then processing the time tag, thereby realizing real-time tracking. The method and the apparatus are mainly applied to image processing and tracking.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to an event-driven real-time processing algorithm, which automatically and accurately captures useful information, has the advantages of low data volume, high processing speed, and high recognition accuracy, and realizes intelligent vision for multi-object recognition and tracking image application. In particular, it relates to a real-time multi-object tracking method and device based on an AER image sensor. Background technique [0002] With the development of communication, network, computer and microelectronics technology, computer vision technology is applied in various fields, and the application of monitoring system in production and life is gradually increasing. The traditional monitoring system requires personnel to be on duty, which consumes manpower and has certain Therefore, it is particularly important to study the intelligent monitoring system. In the...

AI Technical Summary

Problems solved by technology

For example, the image sensor transmits a large amount of raw image data collected to the processing system for tracking. Due to the serial transmission, there is a serious bandwidth limitation, and the processing speed cannot meet the high-speed real-time requirements.

Moreover, in the face of images with complex targets and background noise interference, a large amount of redundant information will be generated, resulting in unsatisfactory tracking results

In addition, the shape of the object being tracked is diverse, and it may be accompanied by a series of changes during the movement process, such as deformation, color change, movement direction change, etc., which will affect the tracking results, making the research of moving object tracking algorithms exist in many aspects. certain challenge

Images