Whitepaper: M2M Telematics for Vehicle Detection Using Computer Vision - Happiest Minds

Jan 2014, HAPPIEST MINDS TECHNOLOGIES
M2M Telematics for
Vehicle Detection Using
Computer Vision
Navya M
SHARING. MINDFUL. INTEGRITY. LEARNING. EXCELLENCE. SOCIAL RESPONSIBILITY.

2
© 2013 Happiest Minds Technologies Pvt. Ltd. All Rights Reserved
Copyright Information
This document is exclusive property of Happiest Minds Technologies Pvt. Ltd. It is
intended for limited circulation.

3
Contents
Copyright Information....................................................................................................... 2
Contents ........................................................................................................................... 3
1 Abstract ................................................................................................................. 4
2 Introduction........................................................................................................... 4
3 Services in Telematics............................................................................................. 5
4 Road Safety and Vehicle Detection ......................................................................... 5
4.1 Vehicle detection using Stereo Vision .....................................................................7
4.2 Vehicle Detection using Monocular vision...............................................................8
4.3 Vehicle Detection during Day and Night..................................................................8
5 Conclusion: .......................................................................................................... 11
6 References: .......................................................................................................... 13
About Happiest Minds..................................................................................................... 14
About the author............................................................................................................. 14

4
1 Abstract
This paper after a brief survey of computer vision in telematics, describes the various steps involved in
vehicle detection through image processing. The paper also covers Monocular and stereo based vision
for image detection and also discuss about how the system can work under various lighting conditions
(day and night) and different approaches adopted to achieve it. And finally end with the recent
development and implications of computer vision in telematics.
2 Introduction
Increasing collaboration between vehicle manufacturing companies and telematics vendors is one of
the major reasons contributing to the growth of telematics services industry. Next generation vehicle
telematics application such as automatic vehicle identifier systems, fleet operation management
systems, remote vehicle diagnostics, incident detection systems, and in-vehicle terminal assistance
are expected to increase the demand of telematics services in vehicle manufacturing industry.
Rising awareness amongst consumers over the need and importance of telematics services will drive
the market demand in upcoming years. Need of effective vehicle tracking systems is further expected
to drive the demand of telematics services in diverse application areas such as emergency service
vehicles, public transport (taxis), cargo delivery, and others. Growing consumer demand for electric
vehicles will lead to investment in telematics.
Growing popularity of GPS enabled smart phones can pose a challenge to the growth of this industry.
Smart phones are consuming an important share in world telematics market and these phones
provide multiple functionalities of telematics systems. Currently, the U.S. and Europe accounts for
the majority of global telematics services market due to stringent regulations from government to
have vehicles equipped with emergency services. Further growth of this industry is expected to be
driven by widespread use of this technology in emerging economies and extended support from
governments to promote vehicular telematics.

5
3 Services in Telematics
At its core, automotive telematics deals with services provided to vehicles over a telecommunications
device. Some of the key services presently available in the domain of automotive telematics include:
• Automatic Crash Notification
• Roadside Assistance Services
• Vehicle Tracking
• Remote Door Services
• Navigation Assistance
• Traffic Assistance
• Concierge Services
• Infotainment Services
• Fleet Management
• Diagnostics
The future of the Telematics would be affected by the increased availability of bandwidth and the
penetration of the wireless network infrastructure.
4 Road Safety and Vehicle Detection
Road safety is of major concern today. This is particularly true of two major groups: motor insurers and
the drivers they insure, especially the youth and employers and their work force who drive to work. In-
vehicle monitoring and road safety needs have become increasingly critical driving the use of vehicle
telematics and/or driver telematics. The research and development of advanced sensing,
environmental perception, and intelligent driver assistance systems presents an opportunity to help
save lives and reduce the number of on-road injuries [1]. A variety of sensing modalities have become
available for on-road vehicle detection, including radar, lidar(illuminating a target with laser and
analyzing the reflected light), and computer vision. Imaging technology has progressed immensely in
recent years. Cameras are cheaper, smaller, and of higher quality than ever before. Concurrently,
computing power has increased dramatically [1]. Consequently, developing on-board automotive
driver assistance systems aiming to alert a driver about driving environments, and possible collision
with other vehicles has attracted a lot of attention. In these systems, robust and reliable vehicle
detection is the first step — a successful vehicle detection algorithm will pave the way for vehicle
recognition, vehicle tracking, and collision avoidance [2].Vehicle detection can be classified into various

6
categories, one of them being Monocular and Stereo vision .The following table depicts the work done
in the recent years in vehicle detection using computer vision (Monocular and Stereo) [1].
There are three kinds of methods for locating potential vehicles: 1) knowledge-based methods; 2)
motion-based methods; and 3) vision based methods. Knowledge-based methods make use of prior
knowledge to locate candidate vehicles in an image. Such prior knowledge includes symmetry, colour,
shadow, geometrical features and texture, etc. As one of the main signatures of man-made objects,
symmetry has been used for object detection and recognition in computer vision. In general,

7
knowledge-based methods are effective in relatively simple environments but cannot work in complex
environments because the prior knowledge is susceptible to illumination, image quality and the
complexity of background. Motion-based methods commonly use optical flow to estimate the relative
motion of vehicles. The disadvantages of this method are time-consuming and cannot detect static
obstacles. Vision-based vehicle recognition methods normally follow two basic steps: 1) hypothesis
generation (HG), which hypothesizes the regions in an image where possible vehicles are; and 2)
hypothesis verification (HV), which verifies the correctness of the hypothesis in step 1) by using certain
algorithms. In general, HG is based on some simple features.
4.1 Vehicle detection using Stereo Vision
This method basically uses 2 cameras, set up in different angle to get a disparity map and then calculate
the depth i.e distance of the obstacle in front. The schematic diagram of the system is illustrated in Fig.
1. Firstly, a stereovision rig corroborated with an edge-indexed stereo matching algorithm is employed
to produce an edge-based disparity map. The regions of interest (ROIs) in the left image are created
according to disparities between the objects and they comprise the potential vehicle objects in the
scene. Secondly, symmetry properties are applied to eliminate the non-boundary points so that an
initial object contour can be extracted. Subject to noise, the initial contour obtained can be
discontinuous and incomplete, containing some noise points within the vehicle body. Therefore, the
initial contour is refined with the snake model so that a closed and complete object contour curve is
generated. Thirdly, two parameters including an object aspect ratio and an area ratio are calculated
from the contour curve and accordingly the objects within the ROIs are classified into two classes -
vehicles or other objects.
[Source: “On-Road Vehicle Recognition Using the Symmetry Property and Snake Models”, International Journal of
Advanced Robotic Systems, 2013]

8
4.2 Vehicle Detection using Monocular vision
Here a single camera mounted on a vehicle is used to detect an obstacle. The Previous image obtained
by the camera is compared with the current image and the movement of an obstacle (Vehicle) is
calculated based on the change in position of the obstacle in the images and the location of the vehicle
is predicted. The method is roughly divided into four principal steps: road motion estimation, obstacle
detection, obstacle motion estimation, obstacle segmentation and prediction as shown in the figure
below.
[Reference: C. Demonceaux , D.Kachi-Akkouche, “Robust obstacle detection with monocular vision based on motion analysis”,
Intelligent Vehicles Symposium, 2004 IEEE.]
4.3 Vehicle Detection during Day and Night
Most of the previous vehicle detection and tracking method assumed favorable light conditions where
one can easily acquire the vehicle template in the image and cannot apply the dead zone at a close
range. In the night time, we cannot acquire favorable clear images of the headlight pair, taillight pair
and turn lights of other vehicles around the ego-vehicle. Thus, a robust algorithm must be developed
to handle these difficult situations.
The two cameras that can detect vehicles in the medium and far range are installed by the side of a
rear-view mirror and at the ceiling above the back seat and the two sonar sensors (its’ beam angle is
15°) that can measure the distance in the near range are installed at the front and rear bumpers.

9
Because the environment of the vehicle changes relatively fast as the speed of an ego-vehicle is high,
we acquire 2 images of 1 field and acquire 2 signals of sonar sensors successively. The overall block
diagram is shown below [5].
[Source: “Front and Rear Vehicle Detection and Tracking in the Day and Night Times Using Vision and Sonar Sensor Fusion”,
Intelligent Robots and Systems, 2005 (IROS 2005)]
Vehicle Detection in the Day Time
The detection system in the day time consists of 4 parts that is, the preprocessing module working on
the input raw image, the vehicle candidate extraction module by a shadow region and a template, the
validation module by a prior knowledge, and the fusion module for fusing sonar and image data.

10
Vehicle Detection in the Night Time
The bright regions generated by headlights, taillights, brake lights and reflected lights around light
sources are used as the vehicle feature in the night time. Overall system is shown

11
By extracting the pairs of lights, filtering the noise region, and reducing the size according to the size
and shape of the bright region by light classification and light extraction, we extract vehicle candidates.
For pairing by light shape, we use the light classification step. First, we calculate possible vehicle pixel
width (PW: 1.5m is used) at the center of gravity (COG) points of each bright region clusters by IPT.
Vehicle light shape that can be appeared in the night time image is divided into 3 shapes.
- Small light: Light source by tail lights and brake lights without spreading.
- Large light: Reflected light appeared in a vehicle by other light sources
- Huge light: Light source by headlight
These lights are classified by the ratio of a vehicle size using PW to a light size.
- Small light: light size ≤ (PW/5) × (PW/5)
- Large light: small light th ≤ light size ≤ (PW/2) × (PW/2)
- Huge light: otherwise case
5 Conclusion:
The computer vision and image processing can be applied in many aspects for the traffic parameter
extraction (Vehicle counting, speed measurement etc.). While vehicle detection has been an active
research area for quite some time, open challenges still remain. In this paper I have discussed briefly
the various techniques used in vehicle detection. Almost all the major Automobiles Giants like BMW,
Ferrari, and Toyota are working to build advanced vehicle and driver telematics that aims to
significantly improve road safety and lead the innovations towards making driverless cars a future
reality. Even though the Idea of Autonomous cars existed way back, it is only been in the past decade
that this field has evolved tremendously. Respondents to this year’s survey feel that emerging nations
offer the best hope for expansion. Eighty-five percent say that growth in the BRICs (Brazil, Russia, India
and China) and other up-and coming nations is the biggest single industry trend up to 2025, which is
consistent with the 2013 survey. The following shows the key Automotive trend up to 2025 by a survey
conducted by KPMG Global Automotive.

12

13
6 References:
[1] Sayanan Sivaraman, Mohan M. Trivedi, “A Review of Recent Developments in Vision-Based Vehicle
Detection”, IEEE Intelligent Vehicles Symposium (IV) June 23-26, 2013.
[2] Zehang Sun, George Bebis, Ronald Miller,” On-Road Vehicle Detection Using Optical Sensors: A
Review”, Computer Vision Laboratory, University of Nevada, Reno, NV, March 2013.
[3] Mustafa Kisa, Fatih Mehmet Botsali, “A real-time Computer Vision System for Vehicle Tracking and
Collision Detection”, World Academy of Science, Engineering and Technology, 2012.
[4] M. Bertozzi, A. Broggi , A. Fascioli, S. Nichele , “Stereo Vision-based Vehicle Detection”, Proceedings
of the IEEE Intelligent Vehicles Symposium,2000.
[5] SamYong Kim, Se-Young Oh, JeongKwan Kang and YoungWoo Ryu, “Front and Rear Vehicle
Detection and Tracking in the Day and Night Times Using Vision and Sonar Sensor Fusion”, Intelligent
Robots and Systems, 2005 (IROS 2005).
[6] D.M. Gavrila,V. Philomin,” Real-Time Object Detection for "Smart" Vehicles”, Computer Vision, 1999.
The Proceedings of the Seventh IEEE International Conference (Volume: 1).
[7] C. Demonceaux , D.Kachi-Akkouche, “Robust obstacle detection with monocular vision based on
motion analysis”, Intelligent Vehicles Symposium, 2004 IEEE.
[8] Iwan Ulrich, Illah Nourbakhsh “Appearance-Based Obstacle Detection with Monocular Color Vision”,
Proceedings of the AAAI National Conference on Artificial Intelligence, Austin, TX, July/August 2000.
[9] Shumin Liu, Yingping Huang, and Renjie Zhang ,“On-Road Vehicle Recognition Using the Symmetry
Property and Snake Models”, International Journal of Advanced Robotic Systems, 2013.

14
About Happiest Minds
Happiest Minds is a next-generation IT services company helping clients differentiate and win with a
Unique blend of innovative solutions and services based on the core technology pillars of cloud
Computing, social computing, mobility and analytics. We combine an unparalleled experience,
comprehensive capabilities in the following industries: Retail, Media, CPG, Manufacturing, Banking
and Financial services, Travel and Hospitality and Hi-Tech with pragmatic, forward-thinking advisory
capabilities for the world’s top businesses, governments and organizations. Founded in 2011,
Happiest Minds is privately held with headquarters in Bangalore, India and offices in the USA and UK.
For further information, please contact business@happiestminds.com
Corporate Office
Happiest Minds Technologies Pvt. Ltd.
Block II, Velankani Tech Park
43 Electronics City
Hosur Road, Bangalore 560100, INDIA
Phone: +91 80 332 03333
Fax: +91 80 332 03000
United States
116 Village Boulevard, Suite 200
Princeton, New Jersey, 08540
Phone:+1 609 951 2296
2018 156th Avenue NE #224
Bellevue, WA 98007
United Kingdom
200 Brook Drive, Green Park, Reading
Berkshire, RG2 6UB
Phone: +44 11892 56072
Fax: + 44 11892 56073
About the author
Navya Manoj is a consultant with the Happiest Minds Technologies Product Engineering Services
business and is very passionate about telematics. As part of this team, she is involved with telematics
design framework development for M2M / IoT solutions. She holds a Masters’ Degree in Mechatronics,
a very specialized field of engineering

Whitepaper: M2M Telematics for Vehicle Detection Using Computer Vision - Happiest Minds

More Related Content

What's hot (20)

Similar to Whitepaper: M2M Telematics for Vehicle Detection Using Computer Vision - Happiest Minds (20)

More from Happiest Minds Technologies (20)

Recently uploaded (20)

Whitepaper: M2M Telematics for Vehicle Detection Using Computer Vision - Happiest Minds