REAL-TIME NUMBER PLATE READING

J. Bulas-Cruz (*), J. Barroso (**), A. Rafael (***) e E. L. Dagless (****)

(*) Professor Auxiliar (**)Assistente

Universidade de Trás-os-Montes e Alto Douro, Secção de Engenharias Apt.202, 5001 Vila Real Codex, Portugal

 (***) Professor Associado

Universidade de Aveiro, Departamento de Electrónica e Telecomunicações Apt. , Aveiro, Portugal

(****) Professor of Microeletronics

University of Bristol, Dep. of Elect. & Electronic Eng. University Walk, Queen's Building Bristol BS8 1TR, United Kingdom

Abstract: Vehicle number plate recognition systems are expected to have numerous applications in traffic surveying and monitoring, e.g. finding stolen cars and controlling access to car parks. In this paper algorithmic improvements to a previous version of a number plate reading system are described. The work builds on the experience gained by the Computer Vision Group at the University of Bristol. This paper addresses the problem of locating the number plate area in the image and proposes a new line based method for number plate location, which is suitable for real-time number plate recognition.

Keywords: computer vision, image segmentation, optical character recognition, automatic recognition, traffic control, real time systems.

1. INTRODUCTION

There is a growing demand for traffic data concerning traffic flow and automatic vehicle identification (Inigo, 1983; Ali, 1993). Increa-sing levels of road traffic have led researchers around the world to adopt advanced electronic and advanced computer vision technology to monitor and control traffic. In order to maximise the use of vision technology the images of the traffic scene should be captured and analysed in real time.

In a typical situation, a number plate reading system will analyse the images captured by a camera at the road side or at the entrance of a car park (Barroso, 1995a, 1995b). Figure 1 shows typical images captured at the entrance of a car park, while Figure 2 shows typical images captured at the road side. The system should find and recognise number plate

 characters in a video frame, in order to extract and decode the number plate, under varying weather conditions. When a car is detected, e.g. by the induction loop, a trigger signal is sent to the number plate reader in order for it to grab and analyse the image(s) of the detected car. The average processing time should be under 1 second per number plate. Experience shows this is usually enough to cope with heavy traffic.

The main tasks of a number plate recognition system are the segmentation of the characters and their identification. This paper addresses the second problem and suggest a line-based method to solve it. The method, instead of looking for character like shapes in the image, takes advantage of the "signature" of the number plate area in a horizontal cross-section of the image.

Fig. 1. Typical images captured at the entrance of a car park.

Fig. 2. Typical images captured at the road side.

 The paper is organised as follows. Next section points to related work and describes the method for number plate location and character segmentation. Section 3 shows some test results. Section 4 contains the conclusions.

2. NUMBER PLATE LOCATION AND CHARACTER SEGMENTATION

Text segmentation techniques have been developed for applications such as the processing of documents for facsimile transmission, and the location of post code on envelopes. Many of these techniques are based on contextual information. This information might be that the text forms horizontal stripes in the image, or that characters are of a similar size and can be discriminated from other forms in the image (Elliman, 1990).

The segmentation methods used for printed text tend to be inadequate for number plate segmentation, because of the reduced number of characters, typically from 6 to 8.

Elliman (Elliman, 1990) analyses the use of Fourier spectra to discriminate different areas of an image, and concludes that some information can be drawn from there. This approach has been tried and the results for the number plate case are described later.

Number plate reading systems have been developed by Elsydel (Elsydel, 1988) and Kato (Kato, 1991). Very little information is given on the segmentation methods they use. Another number plate reader, developed by Nijhuis and co-workers (Nijhuis, 1995), uses a colour based method for the segmentation of the number plate area. This system has been tested on 10.000 different images, without checking for any unreadable or illegal license plates, and in 75.4% of the images the system has decided that a number plate was present.

In order to locate the number plate area in a previous prototype (Bulas-Cruz, 1995; Storer 1994; Lotufo, 1990), a search for character like shapes was conduced. When three or more such shapes were found in similar horizontal positions, the system would look in the neighbourhood for other similar shapes. At a second stage, if the shapes were identified as number plate characters, the system would assume a number plate region had been found.

In order to improve the performance of the location process, a new technique has been developed. This technique is based on the fact that the lines where the number plate is located in the image have a clear "signature" which makes it usually possible to distinguish them from other lines in the image, or at least to pre-select some positions where to look further. Figure 3 shows two such lines. The top image shows the positions of the cross sections (the white lines). The "signature" of the number plate can be observed in the bottom cross section. It corresponds to strong grey level variations at somehow "regular" intervals.

Fig. 3. Cross sections of a car image (the "signa-ture" is shown in the bottom image).

The analysis of the image lines in order to identify which lines "cut" the number plate has been conduced both in the spatial and in the Fourier domain. In the Fourier domain the analysis has proved to be very difficult and so the work continued using the spatial information. An algorithm which analyses the maxima and minima of the cross section has been developed. The algorithm searches for a set of continuous maxima and minima that have some predefined characteristics (number, relative distances, amplitude, etc.). These characteristics are dynamically chosen from a set of predefined values, using statistical information.

Once an horizontal line that crosses the number plate has been located, this information is used to define an area which should contain the number plate image. This area is shown in Figure 4, for the image in Figure 3.

The following step is to locate the number plate area more accurately. This is achieved by using the vertical and horizontal projections of the binary version of the previous image, as shown in Figure 4. These projections will normally indicate the precise position of the number plate set of characters. The threshold value is dynamically chosen, based on the value of the grey level maxima and minima.

Fig. 4. Accurate location of the number plate area.

In order to segment the characters in the number plate image a method based on a technique first proposed by Lu (Lu, 1980, 1983, 1989), named peak-to-valley, is used. The method searches for valleys in the vertical projection of the binary image.

Fig. 5. Character segmentation.

3. TEST RESULTS

In order to evaluate the proposed methods, a set of tests has been conduced. Three image databases have been created: D1, of images taken at the entrance of a car park; D2, of images of vehicles travelling on a road under normal weather conditions; and D3, of images of vehicles travelling on a road under bat weather conditions (e.g., fog, rain).

The results are as follows:

D1 - 100.0% correct number plate location;

D2 - 99.5% correct number plate location;

D3 - 88.5% correct number plate location.

These results show an improvement in performance when compared with the previous method. The images in D3 are of poor quality and the previous method would normally fail. In images of good quality, the improvement comes mainly from the fact that this method is much faster. Typical execution times vary from 5 ms to 100 ms, while the previous method would typically take from 200 ms to 500 ms.

4. CONCLUSIONS

In this paper a new method to locate the number plate area in frontal or rear car images is proposed. The method is based in the observation that an horizontal cross section of the car image shows a clear "signature" when the cross section cuts the number plate area. The tests which have been conduced to evaluate the proposed method indicate that the method is both robust and fast, which makes it suitable for real-time implementations.

Further work should enable the authors to establish how to link this method to previous work on high speed character recognition, in order to significantly improve the speed of the recognition process.

ACKOWLEDGEMENTS

The authors would like to thank the students of Electrical Engineering António Viana, Hugo Fiuza, Argentina Leite e Luís Carvalho for their collaboration in the number plate project.

REFERENCES

Ali A T, Bulas-Cruz J & Dagless E L (1993), Vision Based Road Traffic Data Collection, Proc. ISATA 26th International Conference, (ATT & IVHS), Aachen, Germany.

Barroso J (1995a), Identificação Automática de Placas de Matrícula Automóveis, Tese de Mestrado em Engenharia Electrónica e Telecomunicações , Universidade de Aveiro, Portugal.

Barroso J, Bulas-Cruz J, Rafael A & Dagless L, (1995b), Identificação Automática de Placas de Matrícula Automóveis, Proc. 4.ªas Jornadas Luso-Espanholas de Engenharia Electrotécnica, Porto, Portugal.

Bulas-Cruz J (1995), Image Processing Applications using a Transputer-based System, PhD Thesis, University of Bristol, UK.

Elliman, D G, I T Lancaster (1990), A Review of Segmentation and Contextual Analysis Techniques for Text Recognition, Pattern Recognition Society, pp. 337-346.

Elsydel. (1988), Information leaflet. Elsydel Head Office. 63, Buolevard Bessiers, 75017 Paris. France.

Inigo R M, Traffic Monitoring and Control using Machine Vision: A Survey', IEEE Transaction on Industrial Electronics, Volume No IE-32, No. 3, August (1985), pp.177-185.

Kato, K., Hineroya, T., Mitani, A. and Deguchi, M. (1991). Automatic Licence Plate Reader Utilizing Image Processing.Proc. ISATA, 24th Int. Conf. Road Transport Informatics and Intelligent Vehicle-Highway Systems. Pages 435-442. Florence. Italy.

Lotufo R A, Morgan A D, Johnson A S and Thomas B T (1990). A Transputer Based Automatic Number-Plate Recognition System, Proc. of the 2nd Int. Conf. on Applications of Transputers, University of Southampton, UK.

Lu, Y (1995), Machine printed character segmen-tation, Pattern Recognition, Volume No 28, n. 1, pp 67-80, Elsevier Science Ltd, UK.

Nijhuis J A G, M. H. Ter Brugge, K. A. Helmholt, J. P. W. Pluim, L. Spaanenburg, R.S. Venema, M. A. Westenberg(1995), Car License Plate Recognition whith Neural Networks and Fuzzy Logic. EPIA’95, Portuguese Conference on AI, pp. 25-34.

Storer R, Milford D J, Bulas-Cruz J & Dagless E L (1994), Developing Embedded Appli-cations in an Array of Specialised Transputer Modules', Proc. WOTUG, Bristol, March (1994).