NORMALIZATION METHOD OF LICENSE PLATE IMAGE IN CASE OF ITS RECOGNITION BY MEANS OF VIDEO SURVEILLANCE

Author:

Kazymyr Volodymyr, Chernihiv National University of Technology (95 Shevchenka Str., 14027 Chernihiv, Ukraine)

Zarovsky Ruslan, Chernihiv National University of Technology (95 Shevchenka Str., 14027 Chernihiv, Ukraine)

Radchenko Andrii , Chernihiv National University of Technology (95 Shevchenka Str., 14027 Chernihiv, Ukraine)

Language: ukrainian

Annotation:

Urgency of the research. Car’s plates recognition is of great practical interest because it reduces the costs of organizing parking for cars, timely detect traffic violators and quickly respond to events associated with the use of cars.

Target setting. Haar detector in the case of its application for the detection of license plates doesn't select exact boundaries of plate.

Actual scientific researches and issues analysis. Existing methods of plate normalization based on the contours allocation by using the Sobel or Laplace operators and then search for lines using Hough transform and histogram brightness on the horizontal lines of the image.

Uninvestigatigated parts of general matters defining. Existing methods of normalization is not always cope with the task that leads to a reduction in the percentage of correctly recognized car plates.

The research objectiveIn the paper, the method of plate’s normalization after detection by Haar detector for subsequent character recognition, which is based on the use of differential images, was presented.

The statement of basic materials. The proposed normalization method are based on differential image in which every pixel is represented by the module difference or the difference between the original image pixel and its horizontal or vertical neighbor, under certain conditions or without conditions. Next histogram equalization of such images is executed and then searched for vertical and horizontal straight lines. After that the group of the found lines is executed, the turning angle and plate’s boundaries is founded.

ConclusionsThe correctness of the proposed method depends on images submitted for its entrance and the parameters of the algorithms used in it. However, if we talk about the probability of which were derived from the tests for a particular sample, the percentage of correct results 99.54%, and wrong  0.46%.

Key words:

Haar detector, normalization, differential image, clustering

References:

1. Sistema raspoznavanija avtomobilnykh nomerov «NomerOK» [The car’s plate recognition system «NomerOK»]. Retrieved from http://avtonomerok.su/.

2. Operator Sobelja i Laplasa [Sobel and Laplace operator]. Retrieved from http://robocraft.ru/blog/computervision/460.html.

3. OpenCV shag za shagom. Preobrazovanie Hafa [OpenCV step by step. Hough transformation]. Retrieved from: http://robocraft.ru/blog/computervision/502.html.

4. Martinsky O. Algorithmic and mathematical principles of automatic number plate recognition systems. Retrieved from http://javaanpr.sourceforge.net/anpr.pdf.

5. Rasheed S., Naeem A., Ishaq O. Automated number plate recognition Using Hough Lines and template matching. Retrieved from http://www.iaeng.org/publication/WCECS2012/WCECS2012_pp199-203.pdf.

6. Raspoznavaniia avtomobilnykh nomerov v detaliakh [License plate recognition details]. Retrieved from https://habrahabr.ru/company/recognitor/blog/225913/.

7. haarcascade_russian_plate_number.xml. Retrieved from: https://github.com/opencv/opencv/blob/master/data/haarcascades/haarcascade_russian_plate_number.xml.

8. OpenCV shag za shagom. Obrabotka izobrazheniia – sviortka [OpenCV step by step. Image processing – convolution]. Retrieved from http://robocraft.ru/blog/computervision/427.html.

9. Fisenko, V.T. (2008). Kompiuternaia obrabotka i raspoznavanie izobrazhenii [Computer image processing and recognition]. Saint-Petersburg: SPbGUITMO Publisher (In Russian).

10. Histogram Equalization. Retrieved from http://docs.opencv.org/2.4/doc/tutorials/imgproc/histograms/histogram_equalization/histogram_equalization.html.

11. Klasterizatsiia [Clusterization]. Retrieved from http://www.machinelearning.ru/wiki/index.php?title=Кластеризация.

12. Metody klasternogo analiza. Ierarkhicheskie metody [Methods of cluster analysis. Hierarchical methods]. Retrieved from http://www.intuit.ru/studies/courses/6/6/lecture/182?page=2.

13. Berikov, V.B., Lbov, G.S. Sovremennye tendentsii v klasternom analize [Current trends in the cluster analysis]. Retrieved from http://www.ict.edu.ru/ft/005638/62315e1-st02.pdf.

14. OBJECT ORIENTATION, PRINCIPAL COMPONENT ANALYSIS & OPENCV. Retrieved from https://robospace.wordpress.com/2013/10/09/object-orientation-principal-component-analysis-opencv/.

15. OpenCV. Retrieved from http://opencv.org/.

Download