Sei sulla pagina 1di 6

EXTRACTION OF LICENSE-PLATE NUMBER OF OVER SPEEDING

VEHICLES
Dr. U.B. MAHADEVASWAMY
Associate Professor, Dept. of E&C, SJCE College, ubms_sjce@yahoo.co.in
Mysore, Karnataka 570006, India

NAVEEN LAMBA
Dept. of E&C, SJCE College, naveenlmb3@gmail.com
Mysore, Karnataka 570006, India

ABSTRACT: Traffic management plays a crucial role in cities especially in metropolitan. Lots of
road mishape occurs due to over speeding of vehicles. Usually RADAR or LIDAR techniques are
opted for monitoring traffic speed, which deploy active devices for its operation. Hence lots of
extra power is getting wasted. In this paper, we propose an algorithm to detect the vehicle's speed
using image processing techniques. If the speed exceeds the prescribed limit, the license plate of
the over speeding vehicle is recognized and forwarded to the concerned authorities.

KEYWORDS: Image enhancement, object detection, foreground, blob analysis, speed detection,
contour, image segmentation, noise reduction, over speeding and license plate recognition.

INTRODUCTION

Vehicle speed monitoring is important for enforcing speed limit laws. It also broadcasts the traffic
conditions of the monitored section of the road of interest. Vehicle detection and tracking in real
time is a challenging task in traffic surveillance systems. It often acts as an initial step for further
processing such as speed of the detected object and their license plate recognition [Y.K.Ki,
D.K.Baik (2006)]. Vehicle detection from a video stream relies heavily on image processing
techniques such as motion detection, foreground modeling, image enhancement, blob analysis, and
centroid calculation. A typical moving object detection algorithm has the following feature: (a)
Estimation of the foreground for motion detection [Kunming (2008)] (b) Bounding the detected
objects (c) tracking bounded object [seizime colloque gretsi (1997)] (d) Calculation of velocity.

METHODOLOGY

The approach to create a system to detect the over speeding vehicles from a video sequence has
been put forth in this section. The block diagram of the proposed system is shown in Figure 1.

Video capturing
In the first stage of the project, the videos are captured using a fixed camera. Video cameras are
standard sensing devices deployed for monitoring purpose. The camera is positioned in such a way
that the vehicle approaching the camera can be captured clearly with the license plate visible.

Extracting image frames from the video


A video sequence is a series of still images with a small interval between two successive images.
The video sequence is converted to sequence of frames using MATLAB software. MATLAB
Fig. 1. System overview

image processing library convert the video which is in AVI format into frames. It captures the
video stream and stores the frames in the buffer. Thus we obtained the image frames for further
processing.

Foreground modeling
The moving objects in the video are to be detected for further processing. To perform this,
foreground modeling is done to detect the difference between the background and foreground
contents in an image frame. Gaussian Mixture model [Naikur Bharatkuma (2006)] is used for
foreground detection in the project.

By using the Gaussian Mixture Model (GMM) technique, the values of each particular pixel in the
image is modeled as a mixture of Gaussians. Based on the repetitiveness and variance of each of
the Gaussians of the mixture, Gaussians that correspond to the background can be determined.

Image enhancement
The recorded video may have some noise due to bad weather (light, wind, etc.). To overcome this,
a morphological operation called opening is performed to improve the image quality. The opening
is a composite operator, constructed from the two basic operators erosion and dilation. Opening of
set A by set B, the structuring element is achieved by first the eroding set A by B, then dilating the
resulting set by B. Visual explanation of the opening process is in Figure 2.

Fig 2. Opening operation

Contour tracking
Contour tracking tracks the moving objects. Moving vehicles are tracked by bounding contour and
updating them continuously with each image frame. In the project, we use a tool called Blob
analysis from MATLAB for contour tracking. A blob is defined as a region of connected pixels,

201
which constitutes an object. Blob analysis is the identification and study of these regions in an
image. The blob features usually calculated are area, perimeter, blob shape and centroid. Blobs are
defined based on the minimum area. When a moving object with the area greater than the specified
minimum area is detected in the image frame, it is bounded by a rectangle, hence tracked.

The blobs are given the required shape and color. The counts of blob present in each of the frames
give the vehicle count in that frame. The same procedure is repeated for the entire video. To go to
the next frame, step command is used and the predefined function blob is executed for each of the
frames.

Speed Calculation
Blob analysis provides us the detail of number and position of each blob in a particular frame.
Using this result, coordinates of centroid of each blob in a particular frame are calculated. Now
Euclidean distance between two corresponding centroids in successive frames is calculated. The
speed can now be calculated easily using basic mathematics relation between speed, distance and
time.

The algorithm to detect the speed is as follows:


1. Two frames are considered from the video sequence. They are continuously monitored to detect
the moving vehicles.
2. Once the vehicle is detected at frame 1, it is bound with the blob and the centroid of the
corresponding vehicle is determined.
3. The centroid of the same blob in the frame 2 is determined.
4. The Euclidean distance i.e. the pixel length between the centroid coordinates is calculated and
the time taken to cover the distance is obtained from the video.
5. The speed is then calculated using the formula:

(1)

Where K is calibration constant.

License Plate Recognition


The License Plate of the vehicle is recognized if the speed exceeds the limit. The steps for
implementing License Plate Recognition algorithm in MATLAB are described below:
1. Conversion of a colored image into gray image.
2. Image is enhanced to improvise the given image by filling holes, sharpening the edges of
objects and join the broken lines and increases the brightness of an image.
3. Horizontal and Vertical Edge Processing of the image. In License Plate Recognition algorithm,
the horizontal and vertical histogram represents the column-wise and row-wise histogram
respectively.
4. The obtained histograms are passed through a low pass filter to remove the noise present in the
image.
5. Segmentation of the image to find all the regions in an image that has high probability of
containing a license plate.
6. The segmented regions are processed to obtain the region having highest probability of
containing a license plate.

202
SIMULATION AND RESULTS

For the implementation of our algorithm we took a video with a frame rate of 30. Using MATLAB
the video was successfully read and the frames were chosen for the operations to be performed.
Foreground of the frame was obtained and it was filtered to get clear description of the boundary
of the vehicle. The outcome is as in Figure 3.

Fig 3. A video frame, its foreground and filtered foreground

Using blob analysis, the boundary was provided to the vehicles and the number of vehicles present
in current frame was displayed on the left top corner. The result is displayed in Figure 4.

Fig 4. Blob analysis and Detection of number of vehicles on screen

203
The speed was calculated as proposed in the algorithm and successful results were obtained as in
Figure 5. When the speed exceeded the prescribed limit, license plate of the over speeding vehicle
was recognized. The outputs in the intermediate stages are in Figure 6 and 7. The final recognized
license plate shown in figure 8 can be sent to the concerned authorities.

Fig 5. Display of the measured speed

Fig 6. Sharpened gray image

Fig 7. Segmented image

204
Fig 8. Recognized license plate

CONCLUSION AND FUTURE SCOPES

The proposed algorithm detected the moving vehicles from the video. The images of foreground
objects were enhanced for the smoothening of further processing. The vehicles were bounded by a
blob .The speed of the vehicles were computed using the proposed algorithm. The license plate is
recognized once the vehicle exceeds the speed limit. The results were comparable with the true
speed of the vehicles.

Future work will be directed towards achieving the following issues:


1. Better camera control to enable smooth object tracking at high zoom, in case video is vibrating.
Video stabilization algorithm is required.
2. Improvement in the algorithm to implement the system in double lane road.
3. Improvements to be made in the algorithm so that the system does not fail in the case of
different background and different environmental conditions.

REFERENCES

Y.K.Ki, D.K.Baik, " Model for accurate speed measurement using double loop detectors" IEEE
Trans. Veh. Technol., vol. 55, no. 4, pp. 10941101, Jul. 2006.
H.Cheng, B. Shaw, J. Palen, B. Lin, B. Chen, Z.Wang, "Development and field test of a laser
based nonintrusive detection system for identification of vehicles on the highway" IEEE Trans
Intell. Transp. Syst.
S. Gupte, O. Masoud, R. Martin, N. Papanikolopoulos, "Detection and classification of vehicles,
IEEE Trans. Intell. Transp. Syst., vol. 3, no. 1, pp. 3747, Mar. 2002.
R. Canals, A. Roussel, J.L. Famechon, S. Treuillet, A biprocessor oriented vision - based target
tracking system, IEEE Trans. Ind. Electron., vol. 49, no. 2, pp. 500506, Apr. 2002.
T. Schoepflin, D. Dailey, Dynamic camera calibration of roadside traffic management cameras
for vehicle speed estimation, IEEE Trans. Intell. Transp. Syst., vol. 4, no. 2, pp. 90-98, Jun.
2003.
S. Messelodi, C. Modena, and M. Zanin. " A computer vision system for detection and
classification of vehicles at urban road intersections." Pattern Analysis and Applications,
8(12):1731, 2005.
C. Stauer and W. Grimson. "Adaptive background mixture models for real time tracking." In
Proceedings of the Conference on Computer Vision and Pattern Recognition, pages 246252,
1999.
D. Hari Hara Santosh, P. Venkatesh, P.Poornesh, L. Narayana Rao, N. Arun Kumar Tracking
Multiple Moving Objects Using Gaussian Mixture Model International Journal of Soft
Computing and Engineering (IJSCE), May 2013.

205

Potrebbero piacerti anche