Moving target tracking method based on image

Abstract

The invention provides a moving target tracking method based on an image, and aims to the problem of tracking failure of the conventional tracking method based on a convolutional neural network duringsize or scale change of a moving target. According to the method, feature extraction is performed on the image through a pre-trained network by employing a deep learning technology, a target point istracked by employing a correlation filter method, the tracking accuracy is improved through acquisition of information of time correlation by employing a C3D network, and multi-scale tracking characteristics are provided through combination of a DenseNet network and a network structure of a feature pyramid network so that accurate tracking during change of the size or the scale of the moving target can be realized, and the method can be widely applied to tracking of the moving target based on the image.

Description

technical field [0001] The invention relates to an image-based moving target tracking method. Background technique [0002] With the decrease of image sensor cost and the development of information processing technology, image-based moving target tracking has been widely used, such as security monitoring, automatic driving, environmental detection, field investigation and observation, human-computer interaction and other fields. increasingly important role. [0003] Since the target tracking method is the basis of image-based advanced visual processing tasks, such as traffic statistics of people or vehicles, abnormal behavior detection, etc., it has been developed rapidly in recent years. Traditional target tracking methods mainly include generative-based target representation modeling methods and discriminative-based target appearance modeling methods. Most of these methods use manually extracted features and shallow classifier structures, and there are problems such as s...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the problem that the current tracking method based on convolutional neural network fails to track when the size or scale of the moving target changes, the present invention provides an image-based moving target tracking method, which applies deep learning technology and utilizes pre-training A good network extracts the features of the image, and uses the correlation filter method to track the target point

Images