Visual tracking method based on hybrid hierarchical filtering and complementarity characteristics

A technology of hierarchical filtering and visual tracking, applied in character and pattern recognition, instruments, biological models, etc., can solve problems such as unsatisfactory robustness, achieve superior performance, improve tracking robustness, and improve tracking accuracy.

Active Publication Date: 2021-01-15
中国人民解放军陆军炮兵防空兵学院
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In addition, the existing multi-feature fusion aims to improve the tracking accuracy, and its robustness in complex scenes

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
  • Visual tracking method based on hybrid hierarchical filtering and complementarity characteristics
  • Visual tracking method based on hybrid hierarchical filtering and complementarity characteristics
  • Visual tracking method based on hybrid hierarchical filtering and complementarity characteristics

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0118]The algorithm of this embodiment is implemented using Matlab R2018a and MatConvNet as development tools, and is debugged on an Intel(R)Core(TM)3.19GHz CPU and GeForce GTX Titan X GPU computer. Define the prediction search area as 100×100, and the output probability map as 50×50; set the number of particles in the three stages as N1= 3000, N2=400,N3=100. The experimental data of this embodiment is the OTB100 benchmark data set released in 2015 [Y.Wu,J.Lim,and M.-H.Yang.Object tracking benchmark[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2015,37 (9):1834–1848.], the benchmark library contains 100 videos, and these video sequences are annotated as 11 challenging attributes. The most intuitive and reliable evaluation method in this benchmark library is OPE (one-pass evaluation), which runs all videos from start to finish. In addition, there are two evaluation indicators: Precision (average center error, with 20 pixels as the threshold) and Success (the coinc...

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

The invention discloses a visual tracking method based on hybrid hierarchical filtering and complementarity characteristics. The method comprises the following steps: establishing a three-stage hybridhierarchical filtering target tracking framework, and estimating a tracking result through a coarse-to-fine search strategy by using confidence output of each stage. The method specifically comprisesthe following steps: establishing a first-stage observation model as observation 1 by using deep CNN features so as to separate a target from a background and roughly position the target; establishing a second-stage observation model by using HOG features to serve as observation 2, and adjusting the target position; 3, establishing a third-stage observation model as observation by using SIFT features, and finally positioning the target. According to the method, the tracking precision and robustness are improved, and the tracking effect is excellent in the environments of rapid target movement, background mixing and the like.

Description

Technical field[0001]The invention relates to the technical field of computer vision target tracking, in particular to a visual tracking method based on hybrid hierarchical filtering and complementary features.Background technique[0002]Visual tracking is a basic research topic in the field of computer vision and an important technology in video analysis. Its main purpose is to estimate the state of a target using video sequence data. The important significance of visual tracking research lies in its broad application prospects. It has important application value in civil fields such as video analysis, vehicle navigation, human-computer interaction and intelligent transportation, as well as military fields such as visual guidance, target positioning and fire control.[0003]As we all know, the visual tracking system mainly includes five core components: motion model, feature extraction, observation model, model update and integrated post-processing. Among them, feature extraction is th...

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
IPC IPC(8): G06K9/46G06K9/62G06N3/00G06N3/04
CPCG06N3/006G06V10/462G06N3/045G06F18/253G06F18/214
Inventor 朱虹韩裕生薛模根王勇袁广林祖鸿宇吴云智吴令夏杨钒
Owner 中国人民解放军陆军炮兵防空兵学院
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap