Abstract
Visual tracking has extensive applications in intelligent monitoring and guidance systems. Among state-of-the-art tracking algorithms, Correlation Filter methods perform favorably in robustness, accuracy and speed. However, it also has shortcomings when dealing with pervasive target scale variation, motion blur and fast motion. In this paper we proposed a new real-time robust scheme based on Kernelized Correlation Filter (KCF) to significantly improve performance on motion blur and fast motion. By fusing KCF and STC trackers, our algorithm also solve the estimation of scale variation in many scenarios. We theoretically analyze the problem for CFs towards motions and utilize the point sharpness function of the target patch to evaluate the motion state of target. Then we set up an efficient scheme to handle the motion and scale variation without much time consuming. Our algorithm preserves the properties of KCF besides the ability to handle special scenarios. In the end extensive experimental results on benchmark of VOT datasets show our algorithm performs advantageously competed with the top-rank trackers.
Highlights
Visual object tracking plays an active role in military guidance, robot navigation, medical image processing, virtual augment reality and many other applications
Spatial Robustness Evaluation (SRE) evaluates the trackers by different initial annotation; Temporal Robustness Evaluation (TRE) divides the sequences into 20 pieces and evaluates we compare the center location error for each sequence with two other excellent trackers and the the trackers on different length of sequences; OPE evaluates trackers with other original Kernelized Correlation Filter (KCF) tracker
We analyze the KCF tracker when it suffers from blur or fast motion, and improve the tracking
Summary
Visual object tracking plays an active role in military guidance, robot navigation, medical image processing, virtual augment reality and many other applications. In [8], conditional random filed is exploited to evaluate the matching score of target These methods suffer from the larger variations of target or scenarios, including applications with motion blur and fast motion. The multi-channel features and the approach to integrate them together are applied in KCF to build an insensitive stronger classifier for illustration variation and appearance model variation In this way, the tracker is easier to understand and can be supported by richer powerful features e.g., CNN features and textural features, rather than just using the raw greyscale pixels. Our main contributions can be concluded as follows: (1) analyze the property of frequency-domain feature for blurred image; (2) utilize sharpness point function to evaluate the motion of target; (3) build an active and search scheme on kernelized correlation filter for fast and robust tracking of scenarios with fast and/or blur motion. Our method has achieved excellent performance competing with the current rank-top trackers on extensive experiments; in addition, our method reaches a high speed even in worst scenarios
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
