Vehicle track segmentation using higher order random fields

Quach, Tu -Thach

doi:10.1109/LGRS.2016.2643564

Title: Vehicle track segmentation using higher order random fields

Full Record
Other Related Research

Abstract

Here, we present an approach to segment vehicle tracks in coherent change detection images, a product of combining two synthetic aperture radar images taken at different times. The approach uses multiscale higher order random field models to capture track statistics, such as curvatures and their parallel nature, that are not currently utilized in existing methods. These statistics are encoded as 3-by-3 patterns at different scales. The model can complete disconnected tracks often caused by sensor noise and various environmental effects. Coupling the model with a simple classifier, our approach is effective at segmenting salient tracks. We improve the F-measure on a standard vehicle track data set to 0.963, up from 0.897 obtained by the current state-of-the-art method.

Authors:

^[1]

Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)

Publication Date:: Mon Jan 09 00:00:00 EST 2017

Research Org.:: Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)

Sponsoring Org.:: USDOE National Nuclear Security Administration (NNSA)

OSTI Identifier:: 1338399

Report Number(s):: SAND-2016-12839J
Journal ID: ISSN 1545-598X; 650058

Grant/Contract Number:: AC04-94AL85000

Resource Type:: Accepted Manuscript

Journal Name:: IEEE Geoscience and Remote Sensing Letters

Additional Journal Information:: Journal Volume: PP; Journal Issue: 99; Journal ID: ISSN 1545-598X

Publisher:: IEEE

Country of Publication:: United States

Language:: English

Subject:: 97 MATHEMATICS AND COMPUTING; vehicle track; image segmentation; random fields; synthetic aperture imaging

Citation Formats


                    Quach, Tu -Thach. Vehicle track segmentation using higher order random fields.  United States: N. p., 2017. 
Web.  doi:10.1109/LGRS.2016.2643564.

Copy to clipboard


                    Quach, Tu -Thach. Vehicle track segmentation using higher order random fields.  United States.  https://doi.org/10.1109/LGRS.2016.2643564

Copy to clipboard


                    Quach, Tu -Thach. Mon .  
"Vehicle track segmentation using higher order random fields".  United States.  https://doi.org/10.1109/LGRS.2016.2643564.  https://www.osti.gov/servlets/purl/1338399.

Copy to clipboard


                    
@article{osti_1338399,

  title        = {Vehicle track segmentation using higher order random fields},

  author       = {Quach, Tu -Thach},

  abstractNote = {Here, we present an approach to segment vehicle tracks in coherent change detection images, a product of combining two synthetic aperture radar images taken at different times. The approach uses multiscale higher order random field models to capture track statistics, such as curvatures and their parallel nature, that are not currently utilized in existing methods. These statistics are encoded as 3-by-3 patterns at different scales. The model can complete disconnected tracks often caused by sensor noise and various environmental effects. Coupling the model with a simple classifier, our approach is effective at segmenting salient tracks. We improve the F-measure on a standard vehicle track data set to 0.963, up from 0.897 obtained by the current state-of-the-art method.},

  doi          = {10.1109/LGRS.2016.2643564},

  journal      = {IEEE Geoscience and Remote Sensing Letters},

  number       = 99,

  volume       = PP,

  place        = {United States},

  year         = {Mon Jan 09 00:00:00 EST 2017},

  month        = {Mon Jan 09 00:00:00 EST 2017}

}

Copy to clipboard

Journal Article:

Free Publicly Available Full Text

Accepted Manuscript (DOE)

Publisher's Version of Record

https://doi.org/10.1109/LGRS.2016.2643564

Other availability

Search WorldCat to find libraries that may hold this journal

Citation Metrics:

Cited by: 3 works

Citation information provided by
Web of Science

Save / Share:

Export Metadata

Save to My Library

Similar Records in DOE PAGES and OSTI.GOV collections:

Convolutional networks for vehicle track segmentation

Journal Article Quach, Tu-Thach - Jounral of Applied Remote Sensing

Existing methods to detect vehicle tracks in coherent change detection images, a product of combining two synthetic aperture radar images taken at different times of the same scene, rely on simple, fast models to label track pixels. These models, however, are unable to capture natural track features such as continuity and parallelism. More powerful, but computationally expensive models can be used in offline settings. We present an approach that uses dilated convolutional networks consisting of a series of 3-by-3 convolutions to segment vehicle tracks. The design of our networks considers the fact that remote sensing applications tend to operate inmore »« less
Cited by 3
https://doi.org/10.1117/1.JRS.11.042603

Full Text Available
Vehicle track identification in synthetic aperture radar images

Patent Chow, James G.

Various technologies pertaining to identification of vehicle tracks in synthetic aperture radar coherent change detection image data are described herein. Coherent change detection images are analyzed in a parameter space using Radon transforms. Peaks of the Radon transforms correspond to features of interest, including vehicle tracks, which are identified and classified. New coherent change detection images in which the features of interest and their characteristics are signified are then generated using inverse Radon transforms.
Full Text Available
SAR Imagery Segmentation by Statistical Region Growing and Hierarchical Merging

Journal Article Ushizima, Daniela Mayumi ; Carvalho, E A ; Medeiros, F N.S. ; ... - Digital Signal Processing

This paper presents an approach to accomplish synthetic aperture radar (SAR) image segmentation, which are corrupted by speckle noise. Some ordinary segmentation techniques may require speckle filtering previously. Our approach performs radar image segmentation using the original noisy pixels as input data, eliminating preprocessing steps, an advantage over most of the current methods. The algorithm comprises a statistical region growing procedure combined with hierarchical region merging to extract regions of interest from SAR images. The region growing step over-segments the input image to enable region aggregation by employing a combination of the Kolmogorov-Smirnov (KS) test with a hierarchical stepwise optimizationmore »« less
Full Text Available
Sensor Fusion-Based Vehicle Detection and Tracking Using a Single Camera and Radar at a Traffic Intersection

Journal Article Li, Shenglin ; Yoon, Hwan-Sik - Sensors

Recent advancements in sensor technologies, in conjunction with signal processing and machine learning, have enabled real-time traffic control systems to adapt to varying traffic conditions. This paper introduces a new sensor fusion approach that combines data from a single camera and radar to achieve cost-effective and efficient vehicle detection and tracking. Initially, vehicles are independently detected and classified using the camera and radar. Then, the constant-velocity model within a Kalman filter is employed to predict vehicle locations, while the Hungarian algorithm is used to associate these predictions with sensor measurements. Finally, vehicle tracking is accomplished by merging kinematic information frommore »« less
https://doi.org/10.3390/s23104888
Scalable Track Detection in SAR CCD Images

Technical Report Chow, James G ; Quach, Tu-Thach

Existing methods to detect vehicle tracks in coherent change detection images, a product of combining two synthetic aperture radar images ta ken at different times of the same scene, rely on simple, fast models to label track pixels. These models, however, are often too simple to capture natural track features such as continuity and parallelism. We present a simple convolutional network architecture consisting of a series of 3-by-3 convolutions to detect tracks. The network is trained end-to-end to learn natural track features entirely from data. The network is computationally efficient and improves the F-score on a standard dataset to 0.988,more »« less
https://doi.org/10.2172/1347496

Full Text Available

Similar Records