International Journal of Advanced Information Technologies (IJAIT) Abstract In this paper, the c

Size: px

Start display at page:

Download "International Journal of Advanced Information Technologies (IJAIT) Abstract In this paper, the c"

毛龚
4 years ago
Views:

1 Abstract In this paper, the complete procedure of image-based license plate number recognition was presented, and we addressed the improvement for the common problems in each parts of license plate recognition system. The whole procedure was divided into four sections that were separately preprocessing, locating of vehicle plate area, character segmentation and character recognition. The locating of vehicle plate area was achieved by the method base on fuzzy theory. The higher brightness areas in the fuzzy image were taken as the candidates for evaluating the exact license plate location. In the section of character segmentation, the fast, effective clustering algorithm and the tilt corrective algorithm were used for getting the correct and modified character connective components. Furthermore, in order to make the performance of character recognition higher, the module of mixture of experts was adopted. The self-organization map acted the distributor and gating network in this module, assigned overall classes to several experts called experts. Thus, the amount of classes which a experts in charged will be reduced and then the recognition rate will be promoted. After the experiment, we found that has improved recognition rate of 5%. Keyword: license plate recognition, Mixture of Experts, fuzzy theory, self-organization map, committee machine. 1. ( RFID microwave[13][15] ) (PC) 100

2 Input Vector Expert 1 Expert 2 Expert K g k Gating Network y 1 g 1 y 2 g 2 y k 1 2 PDA [12] HSI Chang[2] (Hough Transform) [6][14] 2 (Gating Network) (Self-Organizing Map, SOM)[8][9] 3 101

3 Expert 1 Expert 2 2 x 1 x 2 Expert 3 Expert k k x n Input Vector Topology Layer 3 SOM (a) (b) (c) (d) (e) 4 (a) (b) (c) (d) (e) (f) (f) 1% 5% 3. [11][17] Chang 102

4 (Membership Function) ( S I E ) 3.1 (Saturation) [5] µ ( s ) = exp( α s ) S ( 1) s α 3.2 (Intensity) µ ( i) = 1 exp( β( i I )) I ( 2) i I β I 3.3 ( ) i d d µ E L 2 d 2 2 µ ( Ei ( )) = Ei ( + d) e σ ( 3) E d= L E(i+d) E(i+d)=1 0 σ L L 50 ( ) ( Prewitt) ( ) ( 4b) 103

5 1. E 1 2. E 2 3. E 1 E 2 µ ( x ) min[ µ ( x ), µ ( x )] µ ( x ) E = = ( 4) E1 E2 E1 E2 3.4 S I E µ ( x) = min[ µ ( x), µ ( x), µ ( x)] ( 5) M E S I 4d (moving average filter) M 4e 3.5 ( 4f) (Connective Components) score m T ( fi fi)( g j g j) = max j T 2 ( 6) i ( fi fi)( g j g j) score m m fi f i i gi g i j

6 - [7][16] i (x i,y i ) j (x j,y j ) ( y ) 1 j y i θ ji = tan i j ( x x ) j i ( 7) -180 ~ ( ) θ ( ) 1. y (0,1) X i C j d ij 4. X i C g d gi r i c g d gi r

7 5. (divide and conquer) (experts) [14] (Mixture of Experts, ME) (Dynamic Structure) 2 (Gating Network) (Self-Organizing Map, SOM) (full connection) δ = µ η ( k) [ x( k) w ( k)] ( 8) j t qj j w ( k+ 1) = w ( k) + δ ( 9) j j j k k µ t η qj j g (Mean Square Error) SOM S 8 B E F C O 106

5 5.2 (a) (b) SOM 2 5 2 3 6 8 9 S J 6 E F L S (mapping

8 5 5.2 (a) (b) SOM S J 6 E F L S (mapping error) 1. SOM 2. ( K-Means) 3. classes table classes

9 5. mapping error classes table 6. classes table SOM c ij = ( X X)( G G ) T ( X X)( Gji Gji ) ji ji T 2 ( 10) c ij X i j i c i C=[ c 0 c 1 c 2 c N-1 ] N 3. C u u g k = exp( u ) ( 11) k 4. Y N 1 = y g ( 12) k = 0 k k bit ( 96.1%) 5 ( 97.2%) 2 ( 98.8%) 6.2 ( 2) 4f

10 classes table (40 20) SOM 121 (11 11) K-Means classes table classes table 2 (Back Propagation, BP) 2 classes table Experts Classes 1 4 A V 1 M B C D E G L O Q U 9 R H B E F J P R S 6 O D Q H J P R S 6 B Q O D U I T V Y R W X Z E A F H K P R V X B E M N Y 7 7 N U W J 8 H K M N R U 8 W P I T Z B E F K L R M 5 U P 3 Back Propagation 91.0% (91/100) 98.5% (591/600) Mixture of Experts 96.0% (96/100) 99.3% (596/600) 4 Ref. Stage 1. Stage 2. Classifier Chang [2] 97.9% 95.6% SOM Comelli [3] N.A. 91.0% Template Anagnostopoulos [1] 96.5% 89.1% Probability Neural Network Nijhuis [11] * 75.4% 98.5% Multi-Layer Perceptron Koval [10] N.A. 95.0% Multi-Layer Perceptron Ge [4] * N.A. 94.0% Support Vector Machine this 96.1% 96.0% Committee Machine 3 5% 4 Stage 1 Stage 2 * 109

11 7. [1] Anagnostopoulos, C.N.E., I.E., Loumos, V., and Kayafas E., A License Plate-Recognition Algorithm for Intelligent Transportation System Applications, IEEE Trans. on Intelligent Transportation Systems, Vol. 7, pp , [2] Chang, S.L., Chen, L.S., Chung, Y.C., and Chen, S.W., Automatic license plate recognition, IEEE Trans. on Intelligent Transportation Systems, Vol. 5, pp , [3] Comelli, P., Ferragina, P., Granieri, M.N., and Stabile, F., Optical recognition of motor vehicle license plates, IEEE Trans. on Vehicular Technology, Vol. 44, pp , [4] Ge, G., Xu, J. and Wang, M., On the Study of Image Characters Location, Segmentation and Pattern Recognition using LS-SVM, WCICA 2006 on Intelligent Control and Automation, Vol. 2, pp [5] Gonzalez, R.C. and Woods, R. E., Digital Image Processing(2 nd ), Prentice Hall, [6] Haykin, S., Neural Networks: A Comprehensive Foundation, Prentice Hall, [7] He, M.G., Harvey, A.L., and Vinay, T., Hough Transform In Car Number Plate Skew Detection, Int. symposium on Signal Processing and Its Applications, Volume 2, pp , [8] Hsu, C.C., Generalizing self-organizing map for categorical data, IEEE trans. on Neural Networks, vol. 17, no. 2, pp , [9] Kohonen T., The self-organizing map, IEEE Proc. Digital Object Identifier, vol. 78, no. 9, pp , sep [10] Koval, V., Turchenko, V. and Kochan, V., A. Sachenko, G. Markowsky, Smart. License Plate Recognition System Based on Image Processing Using Neural Network, Proc. IEEE Int. Workshop on Intelligent Data Acquisition and Advanced Computing Systems, pp , [11] Nijhuis, J.A.G., Ter Brugge, M.H., Helmholt, K.A., Pluim, J.P.W., Spaanenburg, L. and R.S., Venema, Car license plate recognition with neural networks and fuzzy logic, Proc. of IEEE Int. Conf. Neural Networks, Vol. 5, pp , [12] Otsu, N., A Threshold Selection Method from Gray-Level Histograms, IEEE Trans. on System, Man, and, Cyber-netics, Vol. SMC-9, pp , [13] Qiu, J., Sun, B. and You, Q., Study on RFID Antenna for Railway Vehicle Identification, Int. Conf. on ITS Telecommunications Proceedings, pp , [14] Tang, H.M., Face recognition committee machine, Proc. Int. Conf. Multimedia and Expo 2003, vol. 3, pp. III , Jul

12 [15] Urazghildiiev, I.R., Ragnarsson, R., Wallin, K., Rydberg, A., Ridderstrom, P. and Ojefors, E., A vehicle classification system based on microwave radar measurement of height profiles, RADAR 2002, pp , [16] Zhang, Y. and Zhang, C., A new algorithm for character segmentation of license plate, Proc. IEEE Intelligent Vehicles Symposium 2003, pp , Jun [17] Zimic, N., Ficzko, J., Mraz, M., Virant, J., The fuzzy logic approach to the car number plate locating problem, Proc. Intelligent Information Systems, pp ,

P(x,y) P(x-1,y) P(x,y-1) P(x,y+1) P(x+1,y) 1 3.2 Sobel LaplacePrewittRoberts Sobel [2] Sobel 3.3 2 [6] 0 1 1: P(x,y) t (4-connectivity) 2: P(x,y) t 3:

P(x,y) P(x-1,y) P(x,y-1) P(x,y+1) P(x+1,y) 1 3.2 Sobel LaplacePrewittRoberts Sobel [2] Sobel 3.3 2 [6] 0 1 1: P(x,y) t (4-connectivity) 2: P(x,y) t 3: chdong@npic.edu.tw rola@mail.stu.edu.tw (ITS) (LPR) ITS 2. [1][8] [3][9][12] [9] frequency table Sobel [3][12] RGB HSB color model [11] Hough transformation Sobel [4] [10] 1. Intelligent Transportation