High-speed CAMShift Tracking of Insect on GPU

This paper introduces a high-speed CAMShift tracking algorithm towards insect mostly carried on GPU(Graphics Processing Unit) using CUDA (Compute Unified Device Architecture). Most insects move fast and the traditional human-based observing and measuring procedure is quite time-costly. The tradition...

Full description

Saved in:
Bibliographic Details
Published in2018 IEEE International Conference on Information and Automation (ICIA) pp. 897 - 904
Main Authors Xu, Ziqiang, Nie, Yuman, Cao, Pingguo, Song, Quanjun
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.08.2018
Subjects
CAMShift
CUDA
GPU
Graphics processing units
High-speed tracking
Histograms
Image color analysis
Insect
Insects
Microsoft Windows
Target tracking
Online AccessGet full text

Cover

More Information
Summary:This paper introduces a high-speed CAMShift tracking algorithm towards insect mostly carried on GPU(Graphics Processing Unit) using CUDA (Compute Unified Device Architecture). Most insects move fast and the traditional human-based observing and measuring procedure is quite time-costly. The traditional CAMShift algorithm is fundamental and effective but still not fast enough when it comes to fast-moving tracking targets, which can't meet the high-speed target tracking requirements. In the improved algorithm, based on 16-hue color bins, a single color-pattern insect can be tracked by calculating the moments of a binary image at high speed and the motion spatiotemporal parameters, including position, size and orientation, of an individual insect can be extracted directly. Experiments on a tracking benchmark database (VOT2014 Benchmark) and a high-speed vision platform, IDP Express show that there is a great speedup and considerable performance improvement compared with the standard CAMShift algorithm, which certify that the algorithm may potentially be applied on not only insect tracking but also some other high-speed moving target tracking scenes.
DOI:10.1109/ICInfA.2018.8812482