Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

A Fast and Robust Target Tracking Method Based on Sparse Compact Correlation Filter

A correlation filter and target tracking technology, applied in the field of computer vision, can solve problems such as precision dependence, and achieve the effect of alleviating the problem of over-fitting, high precision and fast speed

Active Publication Date: 2022-05-10
XIAMEN UNIV
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Most of this offline training method can achieve real-time, but its accuracy depends on the network and data used for training

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A Fast and Robust Target Tracking Method Based on Sparse Compact Correlation Filter
  • A Fast and Robust Target Tracking Method Based on Sparse Compact Correlation Filter
  • A Fast and Robust Target Tracking Method Based on Sparse Compact Correlation Filter

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0058] The present invention belongs to the target tracking method of the correlation filtering class, and the following embodiments will further illustrate the present invention.

[0059] Embodiments of the present invention include the following steps:

[0060] A. At frame t, for a given target, a basic sample is constructed from the target and its context. The training sample is composed of all circular translation samples of this basic sample. The labels of these circular translation samples are determined by the Gaussian function, and DCF training is more The loss function of the channel correlation filter is defined as follows:

[0061]

[0062] in, is the circular convolution operation symbol, X t ∈R M×N×D and W t ∈R M×N×D is the basic sample and filter of frame t, Y∈R M×N is the label determined by the Gaussian function, M, N and D represent the width, height and channel number respectively, and ξ is the regular term parameter; the goal of filter learning is ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

A fast and robust object tracking method based on sparse compact correlation filters, involving computer vision technology. The basic sample is constructed from the target and its context, and the training sample is composed of all the circular translation samples of the basic sample. DCF trains the loss function of the multi-channel correlation filter; the exclusive sparse regular term and the group sparse regular term are integrated in multi-task learning Construct intra-group sparse regularization items, introduce time consistency constraints in target tracking to alleviate the problem of DCF degradation over time, introduce intra-group sparse regularization items and time regularization items to define regression loss functions, and learn sparse correlation filters ; Channel pruning removes the redundant filters as a whole, sorts the D channel filters according to the degree of importance, and selects the top channel filters for tracking; constructs a Lagrangian function, and uses the ADMM algorithm to optimize the regression loss. Effectively improve the discriminative and interpretability of the filter, with high precision and fast speed.

Description

technical field [0001] The invention relates to computer vision technology, in particular to a fast and robust tracking method based on sparse compact correlation filters. Background technique [0002] Human beings have a high visual perception ability for external videos, and the brain can quickly and accurately locate moving targets in the video. The computer must imitate the visual perception ability of the human brain, and must reach the human level in speed and precision. Visual tracking is a basic problem in computer vision and the basic content of visual perception. Its speed and accuracy determine the real-time and accuracy of visual perception. Object tracking is one of the important research directions in the field of computer vision, and it plays an important role in intelligent video surveillance, human-computer interaction, robot navigation, virtual reality, medical diagnosis, public safety and other fields. The task first selects the object of interest in the...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06T7/277
CPCG06T7/277G06T2207/20081
Inventor 王菡子梁艳杰熊逻
Owner XIAMEN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com