IRJET- Robust and Fast Detection of Moving Vechiles in Aerial Videos using Sliding Windows in Matlab

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 07 Issue: 01 | Jan 2020 www.irjet.net p-ISSN: 2395-0072
© 2020, IRJET | Impact Factor value: 7.34 | ISO 9001:2008 Certified Journal | Page 1683
Robust and Fast Detection of Moving Vechiles in Aerial Videos Using
Sliding Windows in MATLAB
Mr. Abhilash Chavhan,
Post Graduate Student,
Department of Electronics Communication Engineering,
Priyadarshini Bhagwati College of Engineering, Nagpur.
Dr. Pramod Bokde,
Assistant Professor,
Department of Electronics Communication Engineering,
Priyadarshini Bhagwati College of Engineering, Nagpur.
-------------------------------------------------------------------------------***--------------------------------------------------------------------------
Abstract— Vechile surveillance of a extensive vicinity lets us to
analyze tons about the each day sports and site visitors statistics.
With the speedy improvement of far off sensing, satellite video
has grown to be critical facts supply for vehicle detection, which
affords a broader area of surveillance. The completed work
normally makes a speciality of aerial video with fairly-sized
gadgets based totally on feature extraction. However, the shifting
automobiles in satellite TV for PC video imagery range from just a
few pixels to dozens of pixels and exhibit low contrast with
respect to the historical past, which makes it tough to get to be
had looked or shaped statistics. In this project, we inspect the
trouble of transferring vehicle detection in satellite TV for PC
imagery. To the fine of our know-how, it is the primary time to
deal with moving vehicle detection from satellite films. Our
approach consists of two degrees: first, via foreground movement
segmentation and trajectory accumulation, the scene movement
warmth map is dynamically built. Following this, a unique
saliency primarily based historical past model which intensifies
transferring items is offered to segment the vehicles within the
warm areas. The detection of vehicles driving on busy urban
streets in videos acquired by airborne cameras is challenging due
to the large distance between camera and vehicles, simultaneous
vehicle and camera motion, shadows, or low contrast due to weak
illumination. However, it is an important processing step for
applications such as automatic traffic monitoring, detection of
abnormal behaviour, border protection, or surveillance of
restricted areas.
Keywords: Satellite, Historial, Camera Motion
I INTRODUCTION
Cameras established on airplanes or Unmanned Aerial
Vehicles (UAVs) are able to examine the floor region and gather
video data in a highly effective and green way. The various
substantial quantity of capability applications are automated
traffic tracking, detection of odd behaviour, border safety, or
surveillance of restrained regions. Those packages proportion the
need for accurate detection and monitoring of all moving objects
within the camera’s area of view before the scene can be analyzed
and interpreted. There are numerous elements that complicate
the automation of transferring item detection inclusive of the
huge distance between digicam and gadgets leading to small-
sized gadgets within the photo, simultaneous object and camera
motion, shadows, or low contrast due to susceptible illumination.
Although many approaches for shifting object detection in aerial
video surveillance records exist in the literature, the ones
techniques are frequently lacking reliability, robustness, or
actual-time capability. In this project, we make aware of at the
utility of sliding windows for automobile detection in aerial
motion pictures. At the start evolved for face and human
detection that is a brute pressure or exhaustive seek technique
used to localize objects of a positive magnificence across the
entire photograph. A classifier learns an object look version to
reviews its self assurance approximately object lifestyles at each
seek step. The applicability of sliding home windows for car
detection in aerial motion pictures. But, the purpose to pick out
parameters that make a contribution maximum to each the
detection performance and the runtime and optimize them to
reap high detection rates (reliability), few false fine (FP)
detections (robustness), and actual-time processing. More than
one item monitoring can use those detections as enter, but this is
past the scope of this paper. A music-earlier than-discover (TBD)
algorithm is utilized in order to hit upon movement that is
unbiased of the digicam movement. This unbiased motion is given
by means of clustered motion vectors and does no longer
represent automobiles.
As an alternative to TBD, distinction snap shots as carried out
in wide location surveillance with low body quotes of
approximately 1 Hz may be used, but we procedure videos with
high frame costs of 15–30 Hz, wherein difference snap shots
produce extra noise in comparison to TBD. Furthermore,
difference pics do no longer provide statistics approximately
movement route and pace that we specially use to lessen the hunt
space of the sliding window. No longer only can a big amount of
FP detections be averted this manner, however additionally the
processing time is reduced. Then, we talk, evaluate, and optimize
the most critical sliding window parameters such as the desire of
the automobile look version, handling of variable object length, or
optimization techniques. In city scenes with up to 20 motors
within the digicam’s field of view, we attain detection costs of 88
% with most effective 2 % FP detections and processing times
much less than 40 ms according to frame. The main data source
of today’s intelligent transportation system is generally from
ground-fixed or aerial-based cameras or sensors. The major
drawback of these kinds of imagery is the limited spatial
coverage. As a result, researchers start to put attention to higher
space. With great potential and advantage in the field of wide
area monitoring, satellite video has become a new powerful way
for traffic management. Thus, the development of
computationally efficient and reliable detector for massive MIMO
also needs to be thoroughly addressed.
II OBJECTIVES
 Extended motion clusters are rotated upright based on
the direction of the related motion vectors. The
assumption is that the orientation of a vehicle
corresponds to its motion direction.
 We achieve rotation invariance and need to apply the
sliding window for only one orientation which is an
important search space reduction.
 The scale of the entire scene can be normalized using
the Ground Sampling Distance (GSD) that gives us the
image resolution in meters. So, image rescaling is
necessary only for different vehicle sizes and not for the
distance between camera and scene.

 In order to detect independent motion, it is crucial to
compensate the videos for camera motion first. We will
detect Harris Corners with sub-pixel accuracy and track
them over time by a gradient based search in a local
image region
 Corresponding corners between subsequent images are
used to estimate homographies as global image
transformations for image registration
III BLOCK DIAGRAM
Fig. Block diagram of Aerial Video content detection
In the above block diagram shown in the figure, we have
streamed the video and has done frame extraction for pre
processing using K-means clustering algorithm through which we
will extract the feature of thr aerial video, which is inputed to the
value based containt for detection purpose. In this we have
generated a SIFT Matching Algorithm pattern for randomly
checking the videos that are played for extracting the image in
sequential manner.
IV RESEARCH METHODOLOGY / PLANNING OF WORK
In order to demonstrate the effectiveness of the sliding
window approach compared to methods taken from the
literature, we evaluate detection approaches based on object
segmentation. We use the same TBD algorithm with extended
motion clusters as search space and apply one algorithm for blob
extraction based on the tophat transform [26] and one algorithm
for edge based contour extraction based on clustering of Canny
edges and Harris corners [3]. Each blob or cluster is considered
as one detected vehicle. The authors of the second method also
propose to perform color segmentation and fuse the information
in a Bayesian network. As we do not have color information in
our sequences SEQ 1, SEQ 2, and SEQ 3, we skip these processing
steps in our evaluation. The detection performance is compared
using the f-score as visualized. Motion vector clustering is
considered as baseline approach and is clearly improved by all
three methods. Inner vehicle structures such as trunks or engine
hoods cause split detections (i.e. FP detections) for blob and
contour based segmentation. Merged detections (i.e. FN
detections) often occur in SEQ 1 and cause the large gap between
the sliding window approach and the segmentation methods. We
also evaluate the average overlap between detection and ground
truth rectangles. This is given by the N-MODP that lies between
0.6 and 0.7 for the sliding window and between 0.5 and 0.6 for
the segmentation methods which suffer from under segmentation
due to street texture or sidewalk.
The vehicle detection approaches in aerial imagery have been
well studied in recent research work. In aerial images; the
resolution is high enough to utilize the shape or appearance
models of vehicles to satisfy the demand for detection. Even the
component-based vehicle detection approach can be applied for
the aerial imagery. However, satellite video sequences cannot
provide the detailed information of vehicles because of the
limited resolution. Though less appearance information of
vehicles can be utilized for detection, some methods are still
proposed for object detection in high-resolution satellite imagery
to detect vehicles by using an elliptical blob detection strategy
and separating vehicles from non-vehicular objects by using a k-
Nearest Neighbor (KNN) classifier with various classical features.
V CONCLUSIONS
In this work, we had scene-adaptive algorithm for automobile
detection in aerial photos using an improved framework. Rather
than directly the use of the network, we built a new structure
with fewer layers to enhance the detection performance. By
introducing the context-aware-based function map fusion, we
mixed the characteristic maps of the adjacent frames with the
contemporary frame to boom the accuracy of automobile
detection. Rather than making use of the traditional bounding
container, a sloping bounding field is used to make certain the
rotation in variance of the proposed framework, especially for
motors at a high density and those with big length–width ratio.
Finally, the experimental outcomes display the prevalence of our
proposed set of rules for vehicle detection as compared to the
alternative latest technique.
VI REFERENCES
[1] R. Benenson, M. Mathias, R. Timofte, and L. Van Gool.
Pedestrian detection at 100 frames per second. In Proceedings of
the IEEE International Conference on Computer Vision and
Pattern Recognition (CVPR), 2012.
[2] X. Cao, C. Wu, J. Lan, P. Yan, and X. Li. Vehicle Detection and
Motion Analysis in Low-Altitude Airborne Video Under Urban
Environment. IEEE Transactions on Circuits and Systems for
Video Technology, 21(10):1522–1533, Oct. 2011.
[3] H.-Y. Cheng, C.-C. Weng, and Y.-Y. Chen. Vehicle Detection in
Aerial Surveillance Using Dynamic Bayesian Networks. IEEE
Transactions on Image Processing, 21(4):2152–2159, Apr. 2012.
[4] M.-M. Cheng, Z. Zhang, W.-Y. Lin, and P. H. S. Torr. BING:
Binarized Normed Gradients for Objectness Estimation at 300fps.
In Proceedings of the IEEE International Conference on Computer
Vision and Pattern Recognition (CVPR), 2014.
[5] N. Dalal and B. Triggs. Histograms of Oriented Gradients for
Human Detection. In Proceedings of the IEEE International
Conference on Computer Vision and Pattern Recognition (CVPR),
pages 886–893, 2005.
[6] S. J. Davey, M. G. Rutten, and B. Cheung. A Comparison of
Detection Performance for Several Track-Before-Detect
Algorithms. In Proceedings of the 11th International Conference
on Information Fusion (FUSION), 2008.
[7] P. Dollar, S. Belongie, and P. Perona. The Fastest Pedestrian ´
Detector in the West. In Proceedings of the British Machine Vision
Conference (BMVC), 2010.

[8] P. Dollar, Z. Tu, P. Perona, and S. Belongie. Integral Chan- ´ nel
Features. In Proceedings of the British Machine Vision Conference
(BMVC), 2009.
[9] P. F. Felzenszwalb, R. B. Girshick, D. McAllester, and D.
Ramanan. Object Detection with Discriminatively Trained Part
Based Models. IEEE Transactions on Pattern Analysis and
Machine Intelligence, 32(9):1627–1645, Sept. 2010.
[10] U. Franke, C. Rabe, H. Badino, and S. Gehrig. 6D-Vision:
Fusion of Stereo and Motion for Robust Environment Perception.
In W. G. Kropatsch, R. Sablatnig, and A. Hanbury, editors, Pattern
Recognition, volume 3663 of Lecture Notes in Computer Science
(LNCS), pages 216–223. Springer Berlin / Heidelberg, Sept. 2005.
[11] Y. Freund and R. E. Schapire. A decision-theoretic
generalization of on-line learning and an application to boosting.
Journal of Computer and System Sciences, 55(1):119–139, Aug.
1997.
[12] A. Gaszczak, T. P. Breckon, and J. Han. Real-time people and
vehicle detection from UAV imagery. In Proceedings of SPIE Vol.
7878, Intelligent Robots and Computer Vision XXVIII: Algorithms
and Techniques, 2011.
[13] G. Gualdi, A. Prati, and R. Cucchiara. Multistage Particle
Windows for Fast and Accurate Object Detection. IEEE
Transactions on Pattern Analysis and Machine Intelligence,
34(8):1589–1604, Aug. 2012.
[14] C. Harris and M. Stephens. A combined corner and edge
detector. In In Proceedings of the Fourth Alvey Vision Conference,
1988.
[15] R. Hartley and A. Zisserman. Multiple-View Geometry in
Computer Vision. Cambridge University Press, 2 edition, Mar.
2004.
[16] C. Heng, S. Yokomitsu, Y. Matsumoto, and H. Tamura. Shrink
boost for selecting multi-LBP histogram features in object
detection. In Proceedings of the IEEE International Conference on
Computer Vision and Pattern Recognition (CVPR), 2012.
[17] N. Japkowicz and M. Shah. Evaluating Learning Algorithms: A
Classification Perspective. Cambridge University Press, 2011.
[18] R. Kasturi, D. Goldgof, P. Soundararajan, V. Manohar, J.
Garofolo, M. Boonstra, V. Korzhova, and J. Zhang. Framework for
Performance Evaluation of Face, Text, and Vehicle Detection and
Tracking in Video: Data, Metrics, and Protocol. IEEE Transactions
on Pattern Analysis and Machine Intelligence, 31(2):319–336,
2009.
[19] B. Leibe, A. Leonardis, and B. Schiele. Robust object detection
with interleaved categorization and segmentation. International
Journal of Computer Vision, 77:259–289, May 2008.
[20] T. T. Nguyen, H. Grabner, H. Bischof, and B. Gruber. Online
Boosting for Car Detection from Aerial Images. In Proceedings of
the IEEE International Conference on Research, Innovation and
Vision for the Future (RIVF), 2007.
[21] C. Papageorgiou and T. Poggio. A Trainable System for
Object Detection. International Journal of Computer Vision (IJCV),
38(1):15–33, June 2000.
[22] V. Reilly, H. Idrees, and M. Shah. Detection and Tracking of
Large Number of Targets in Wide Area Surveillance. In
Proceedings of the European Conference on Computer Vision
(ECCV), 2010.
[23] I. Saleemi and M. Shah. Multiframe Many-Many Point
Correspondence for Vehicle Tracking in High Density Wide Area
Aerial Videos. International Journal of Computer Vision (IJCV),
104(2):198–219, Sept. 2013.
[24] J. Shi and C. Tomasi. Good features to track. In Proceedings
of the IEEE International Conference on Computer Vision and
Pattern Recognition (CVPR), 1994.
[25] M. Siam and M. ElHelw. Robust Autonomous Visual
Detection and Tracking of Moving Targets in UAV Imagery. In
Proceedings of the IEEE International Conference on Signal
Processing (ICSP), 2012.
[26 ] Z. Zheng, G. Zhou, Y. Wang, Y. Liu, X. Li, X. Wang, and L. Jiang.
A Novel Vehicle Detection Method With High Resolution Highway
Aerial Image. IEEE Journal of Selected Topics in Applied Earth
Observations and Remote Sensing (JSTARS), 6(6):2338–2343,
2013.

IRJET- Robust and Fast Detection of Moving Vechiles in Aerial Videos using Sliding Windows in Matlab

More Related Content

What's hot (19)

Similar to IRJET- Robust and Fast Detection of Moving Vechiles in Aerial Videos using Sliding Windows in Matlab (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET- Robust and Fast Detection of Moving Vechiles in Aerial Videos using Sliding Windows in Matlab