License Plate Recognition

Emerging Trends in Computer Science and Information Technology -2012(ETCSIT2012) Proceedings published in International Journal of Computer Applications® (IJCA) Automatic Vehicle Identification Using License Plate Recognition for Indian Vehicles Sandra Sivanandan Department of Computer Engineering K. K. Wagh Institute Of Engineering Education & Research, Hirabai Haridas Vidyanagari Amrut-Dham, Panchavati, Nashik-422003 University of Pune, Maharashtra Ashwini Dhanait Department of Computer Engineering K. K.
Wagh Institute Of Engineering Education & Research, Hirabai Haridas Vidyanagari Amrut-Dham, Panchavati, Nashik-422003 University of Pune, Maharashtra Yogita Dhepale Department of Computer Engineering K. K. Wagh Institute Of Engineering Education & Research, Hirabai Haridas Vidyanagari Amrut-Dham, Panchavati, Nashik-422003. Yasmin Saiyyad Department of Computer Engineering K. K. Wagh Institute Of Engineering Education & Research, Hirabai Haridas Vidyanagari Amrut-Dham, Panchavati, Nashik-422003. ABSTRACT In this study, a smart and simple algorithm is presented for vehicle’s license plate recognition system.
The proposed algorithm consists of three major parts: Extraction of plate region, segmentation of characters and recognition of plate characters. For extracting the plate region edge detection and morphological operations are used. In segmentation part scan line algorithm is used. Character Segmentation for Devanagari Number Plates is also presented. Optical character recognition technique is used for the character recognition. The objective is to design an efficient automatic authorized vehicle identification system by using the vehicle number plate.

Here we are presenting a smart and simple algorithm for vehicle’s license plate recognition system for Indian Vehicles. In this study, the proposed algorithm is based on extraction of plate region, segmentation of plate characters and recognition of characters. In India we find plates having Devanagari fonts as well (though according to rules it is not allowed). Character extraction for Devanagari font is slightly different as compared to English font because of the header line (shirorekha). We propose algorithm for character extraction for Devanagari font. The recognized plate an be then compared with police hotlist database to identify stolen vehicles. The paper is organized as follows: Section II provides an overview of the overall system. Extracting the plate region is explained in Section III. Section IV gives the segmentation of individual plate characters. Section V deals with recognition of characters using optical character recognition based on statistical based template matching algorithm which uses correlation and section VI deals with verification of plate according to Indian rules. The paper concludes with Section VII. Keywords
Devanagari, Edge detection, License plate recognition, Optical character recognition, segmentation. 1. INTRODUCTION License plate recognition (LPR) is a form of Automatic Vehicle Identification. It is an image processing technology used to identify vehicles by only their license plates. Real time LPR plays a major role in automatic monitoring of traffic rules and maintaining law enforcement on public roads. The LPR system’s significant advantage is that the system can keep an image record of the vehicle which is useful in order to fight crime and fraud (“an image is worth a thousand words”).
Early LPR systems suffered from a low recognition rate, lower than required by practical systems. The external effects (sun and headlights, bad plates, wide number of plate types) and the limited level of the recognition software and vision hardware yielded low quality systems. However, recent improvements in the software and hardware have made the LPR systems much more reliable and wide spread. 23 Emerging Trends in Computer Science and Information Technology -2012(ETCSIT2012) Proceedings published in International Journal of Computer Applications® (IJCA) in night condition, contrast enhancement is important before further processing [1]. . STRUCTURE OF LPR SYSTEM Fig. 1) Original Image Fig. 2) Gray Scale Image Flowchart of Proposed System The algorithm proposed in this paper is designed to recognize license plates of vehicles automatically. Input of the system is the image of a vehicle captured by a camera. The captured image taken from 3-5 meters away is first converted to gray scale. We apply vertical edge detection algorithm and morphological operation i. e. open and close for plate extraction. After applying morphological operations image is filtered out to get exact plate region. Plate region is cropped.
Row segmentation separates row in plate and column separation separates characters from row. Finally recognition part OCR recognizes the characters giving the result as the plate number in ASCII format. The result in ASCII format is can be verified on the basis of rules followed in India. Fig. 3) Gray image after contrast enhancement 3. 2 Vertical Edge Detection Before applying edge detection median filter is to be applied to image for removing noise. The main idea of median filter is to run through the signal, entry by entry, replacing each entry with the median of neighboring entries.
Such noise reduction is a typical preprocessing step to improve the results of later processing (edge detection) [2]. 3. EXTRACTION OF PLATE REGION Plate Extraction is done in following steps 3. 1 Convert image to Gray Scale 3. 2 Apply Vertical Edge detection 3. 3 Candidate Plate Area Detection ? Morphologically Close image ? Fill holes in image ? Morphologically Open image 3. 3 Filtration of non Plate region 3. 1 Conversion To Gray Scale This is pre-processing step for plate extraction. We apply Formula: I( i, j) = 0. 114*A( i, j,1) + 0. 587*A(i, j, 2) + 0. 99* A(i, j,3) where, I(i,j) is the array of gray image, A(i,j,1), A(i,j,2), A(i,j,3) are the R,G,B value of original image respectively. Sometimes the image may be too dark, contain blur, thereby making the task of extracting the license plate difficult. In order to recognize the license plate even In ascending order of values: 0, 2, 3, 3, 4, 6, 10, 15, 97. Center value (previously 97) is replaced by the median of all nine values (4). Edge detection is performed on the given image, which aims at identifying points in digital image at which image brightness changes sharply or, more formally, has discontinuities.
There mainly exists several edge detection methods (Sobel, Prewitt, Roberts, Canny). We use here Sobel operator for vertical edge detection. If we define A as the source image, and Gx and Gy are two images which at each point contain the horizontal and vertical derivative approximations, the computations are as follows: 24 Emerging Trends in Computer Science and Information Technology -2012(ETCSIT2012) Proceedings published in International Journal of Computer Applications® (IJCA) Where * is 2D convolution operation. Fig. 5) Closed Image Fig. 4) Sobel Vertical Edge detection Fig. 6) Filled Image 3. Candidate Plate Area Detection A morphological operator is applied to the image for specifying the plate location. We build a morphological operator that is sensitive to a specific shape in the input image. In our system rectangular box is employed as a structural element to detect the car plates. In mathematical morphology structuring element are represented as matrices. Structuring element is a characteristic of certain structure and features to measure the shape of an image and is used to carry out other image processing operations [4]. Typical rectangular structuring element is shown in figure. Fig. ) Opened Image 3. 4 Filtration Of Non Plate Region After identify the ROI, image is then filtered using following filtering techniques. First find the connected components in image. The first technique involves removing of all white patches which has more or less area than the threshold. For instance components having area < 2000 or >20000 are eliminated. Using Bounding Box method, draw Bounding Box around components and fill the image. According to the height values, for instance, only the objects with a height greater than Tmin_h and less than Tmax_h are retained, and eliminate the other objects.
After that, if the width values of the retained objects are greater than Tmin_w and less than Tmax_w, the objects are retained; otherwise, the objects are removed, and so on. Where: Tmin_h : Minimum height of the object. Tmax_h : Maximum height of the object. Tmin_w : Minimum width of the object. Tmax_w : Maximum width of the object [6]. After filtering plate region is cropped by searching for the first and last white pixels starting from top left corner of an image. Plate is cropped from original image after getting coordinates. Using two basic operation of morphology (erosion and dilation), opening and closing of image is done.
The opening of A by B is obtained by the erosion of A by B, followed by dilation of the resulting image by B. The closing of A by B is obtained by the dilation of A by B, followed by erosion of the resulting structure by B. For closing image 10*20 rectangular structuring element is used. After closing image we have to fill the holes in this image. A hole is a set of background pixels that cannot be reached by filling in the background from the edge of the image [3]. Then image is opened using 5*10 rectangular structural element. Values are determined according to the size of the image.
Here we have used 1280X980 resolution images. 25 Emerging Trends in Computer Science and Information Technology -2012(ETCSIT2012) Proceedings published in International Journal of Computer Applications® (IJCA) 4. SEGMENTATION OF PLATE CHARACTERS Before applying the OCR, the individual lines in the text are separated using line separation process and individual characters from separated lines. Steps for Character Segmentation: 4. 1 Binarization of Plate image 4. 2 Scan Line Algorithm for row segmentation 4. 3 Vertical Projection for column segmentation
Fig 6) Filtered Image on basis of area Fig. 7) Bounding Box and filled image 4. 1 Binarization Of Plate Image Binarize the plate image. Threshold for binarization must be such that characters are displayed well. For that we take average of all pixel values in plate image and calculate threshold. Fig. 10) Binarized image Fig. 8) Image after filtration on basis of height &width of objects 4. 2 Scan Line Algorithm The scan line algorithm is based on the feature that there is transition from 1 to 0 and 0 to 1 transition in character region in a binary image.
Thus the total number of transition in character region is more than the total number of transition in other region. There are at least seven characters in license plate region and every character has more than two Jumps[7]. We can choose twelve as the threshold value. If the total number of transitions in a certain line is greater than twelve, this line may be in character region. Otherwise, it is not in character region. Algorithm: 1) Let H be height and W be Width of Plate image. 2) for(i=H/2 to 0) { Count no of transitions ie 0 to 1 and 1 to 0 in cnt; if cnt