Design Manual - glasnost.itcarlow.ie
Contents
1. File Pre processing Localization Segmentation Recognition Help Do Image Figure 3 menu bar As is shown in figure 3 there are six menus in menu bar file pre processing localization segmentation recognition and help File Pre processing Localization Segmentation Recognition Help Open Ctrl O Exit Ctrl Q Figure 4 menu of file There are two options in file menu figure 4 open and exit User can use this menu to input image or exit the system Convert to Grayscale Gaussian Smoothing Figure 5 menu of pre processing There are two options in pre processing menu figure 5 which allow user to see the result after convert imputed image to grayscale or after applied Gaussian smoothing algorithm to the grayscale image B Sc Hons in Software Engineering CW228 Project Title Number Plate Recognition Name Dongfan Kuang Login ID C00131031 Supervisor Nigel Whyte Canny Edge Detection Number Plate Isolation Original Image Figure 6 menu of locolization There are two options in localization menu figure 6 which allow user to see the result after edge detection or number plate isolation Segmentation recognition and help menus don t have sub menus So when user can check the result after character segmentation and recognition by click these two menus When user clicks help menu a user manual will pop up to teach user how to use this software B Sc Hons in Software Engineering2. binarization and vertical projection Number plate area is first separated from the original image and converted into black and white form by perform Binarization algorithm onto it then the vertical projection algorithm is performed to separate single character in the number plate area 4 4 2 Detailed Description of each module Module name character segmentation Description separate isolated number plate sub image into several parts and each separated part is supposed to contains one character Input number plate sub image Output a set of sub image each of these sub images contains a character Main scenario 1 Clickthe Menu Segmentation in the main interface 2 Get the sub image from the original image as the same size and position of the resized number plate area convert the sub image into binary image 13 B Sc Hons in Software Engineering CW228 Project Title Number Plate Recognition Name Dongfan Kuang Login ID C00131031 Supervisor Nigel Whyte 3 Apply vertical projection algorithm onto the binary sub image to separate the single character 4 The isolated single character sub image will be displayed in the segmentation panel 4 5 Sub system of Character Segmentation 4 5 1 Overview Character Recognition Grid Feature Method Figure 11 modular design of character recognition sub system Character recognition contains a module called grid feature method to identify the characters in number pl
3. CW228 Project Title Number Plate Recognition Name Dongfan Kuang Login ID C00131031 Supervisor Nigel Whyte 4 Modular Design of Each Sub system 4 1 Modular Design Overview openimage file Convert Color Image into Grayscale Pre processing Gaussian Smoothing Canny Edge Detection Number Plate Localization Roughly Isolation Number Plate Isolation Plate Resizing Binarization Character Segmentation Vertical Projection Character Grid Feature Recognition Method Figure 7 module design overview The whole NPR system is composed of four sub systems pre processing number plate isolation character segmentation and character recognition Each of these sub systems also composed of several modules as shown in figure 7 B Sc Hons in Software Engineering CW228 Project Title Number Plate Recognition Name Dongfan Kuang Login ID C00131031 Supervisor Nigel Whyte 4 2 Sub system of Pre processing 4 2 1 Overview Pre processing Convert Color Open file Image into Grayscale Gaussian Smoothing Figure 8 modular design of pre processing sub system The pre processing sub system has three modules open file grayscale and Gaussian smoothing After an image file is opened it then can be converted into grayscale form and the noises in the image can be removed through Gaussian smoothing module 4 2 2 Detailed description of each module Module name open file Description load an image file Input im
4. the number plate area by using vertical projection algorithm it outputs a set of sub images and each of these sub images contains a single character Character recognition sub system identifies the characters which contains in each sub images and output them in text format B Sc Hons in Software Engineering CW228 Project Title Number Plate Recognition Name Dongfan Kuang Login ID C00131031 Supervisor Nigel Whyte 3 Interfaces Design 3 1 Main GUI Design The screenshoot of main user interface of NPR system is shown in figure 2 File Pre processing Localization Segmentation Recognition Help Original Image b Localozation Segmentation Recognition Figure 2 main user interface screenshoot On the top is menu bar which contains six menus There are four panels below the menu bar original image panel localization panel segmentation panel and recognition panel When running the application results of each main step will be shown in these panels Two buttons below the recognition panel run and clean Button run can run the whole process to output the plate number in text form by using the inputted image Button clean is used to clean all the previous changes and return the application back to original status B Sc Hons in Software Engineering CW228 Project Title Number Plate Recognition Name Dongfan Kuang Login ID C00131031 Supervisor Nigel Whyte 3 2 Menu Design A pas l A met
5. Institute of Technology Carlow B Sc Hons in Software Engineering CW228 Design Manual Project Title Number Plate Recognition Dongfan Kuang Login ID C00131031 Supervisor Nigel Whyte 18 January 2010 B Sc Hons in Software Engineering CW228 Project Title Number Plate Recognition Name Dongfan Kuang Login ID C00131031 Supervisor Nigel Whyte Table of Contents 1 2 3 alaeo Ule CI a aaee EA E E E T 3 Architectural DesSigN iseseisana nanan a aena haia aa a Aaa a saian Eerie aiaa 4 interfaces DESI Me aes waa de inina EEEE E EE EE ERT ARATAT 5 SA Main GUI Desig Miirennatmecasiccoieinuweini n a dunes vdbaadeeual 5 32 Menu DESIGN noseia yecsscoduenaveveuces vavaucedvedeudeavevercdach cceacuanadesseavedeacusdaceasedecdeacuadeceaceanes 6 Modular Design Of Each SUS VS tein vie ccasevied sxuihieceevadatvecsduSenuiisiacnesaduedvausboncentiuncvasuiadenediacntaes 8 4 1 Modular Design OVerview cccccccccccessssssssscaeeececeeecsesesseaeeeeeeseessueseseeaeeeseeeeesseseasaeeeeees 8 4 2 Swb SyStem Of Pre PrOCeSs IG acco vivccdennics cacasadectecsasanceeisiecuesadavvcunysccveiiniaed sudeamavaabeacentagues 9 4 2 1 OVEMIEW noninsani daieed tisecvananchecn ed a AEE A EAEAN ENE ESAS EA 9 4 2 2 Detailed description of each module ssssssssssssrssesesssssrsrrssrrrrsserrentesrennssrennsssnt 9 AS Sub system of LOCAMIZAU ON sicrcssrccsictastesiceczsacicnusaiaaaacaatsanizadadasdasauadsayetiendseetladlapaatec
6. age file jpg or bmp Main success scenario 1 Click the Menu File inthe main interface 2 Click the Open option 3 The file open window will appear user can choose the directory select the file or type the file name into the file name column and then click OK button The selected image will be loaded and displayed in the original panel in the main user interface the default file type is bmp B Sc Hons in Software Engineering CW228 Project Title Number Plate Recognition Name Dongfan Kuang Login ID C00131031 Supervisor Nigel Whyte Extensions if user chose a file which is not jpg or omp type or typed a wrong file name an error window will appear to tell user they made a mistake Module name convert color image into grayscale Description change the inputted image into grayscale format Input original image Output grayscale image Main success scenario 1 Click the Menu Pre processing in the main interface 2 Clickthe Convert to grayscale option 3 Original image displayed in the original image panel will be changed into grayscale image Module name Gaussian smoothing Description remove noises in the image Input grayscale image Output grayscale image with fewer noises Main success scenario 1 Click the Menu Pre processing in the main interface 2 Clickthe Gaussian smoothing option 3 The grayscale image displayed in the original panel will be ch
7. anged into the grayscale image with fewer noises 10 B Sc Hons in Software Engineering CW228 Project Title Number Plate Recognition Name Dongfan Kuang Login ID C00131031 Supervisor Nigel Whyte 4 3 Sub system of Localization 4 3 1 Overview Number Plate Localization Canny Edge Number Plate Detection Isolation Non maximum Edge Roughly Suppression Thresholding Isolation Differentiation Resizing Figure 9 modular design of number plate localization sub system Sub system of number plate localization contains two main modules Canny edge detection and number plate isolation The grayscale image first converted into edge detected form through Canny edge detection module and then the number plate area is identified through number plate isolation module from the edge detected image Each of these two modules is contains several small modules 4 3 2 Detailed Description of Each Module Module name Canny edge detection Description perform the edge detection algorithm to remove all the other information except the edges in the image Input grayscale image after Gaussian smoothing Output edge detected image Main success scenario 11 1 2 3 4 5 6 B Sc Hons in Software Engineering CW228 Project Title Number Plate Recognition Name Dongfan Kuang Login ID C00131031 Supervisor Nigel Whyte Click the Menu Localization in the main interface Click the Canny edge detection option P
8. ate area and output them as text format 4 5 2 Detailed Description of Each Module Module name character recognition Description identify the characters in the different part of number plate sub image Input segmented sub image Output plate number in text form Main scenario 1 Clickthe Menu Recognition in the main interface 14 B Sc Hons in Software Engineering CW228 Project Title Number Plate Recognition Name Dongfan Kuang Login ID C00131031 Supervisor Nigel Whyte 2 Apply grid feature method onto each segmented sub image 3 The registration number on the number plate will be displayed in the recognition panel Extensions if some character cannot be identified it will be displayed as a question mark 5 Conclusions The whole development process can be divided into four main steps pre processing number plate localization character segmentation and character recognition An executable version will be delivered and then tested at the end of each step to check the performance and fix the bugs 15
9. erform the differentiation algorithm onto the grayscale image to find the edge strength by calculating the gradient of the image Perform non maximum suppression algorithm Perform edge thresholding algorithm onto the image to remove all the other information in the image except edges The grayscale image displayed in the original panel will be changed into the edge detected image Module name number plate isolation Description separate the number plate area from the edge detected image Input edge detected Image Output location information of number plate area Main success scenario 1 2 3 4 5 Click the Menu Localization in the main interface Click the number plate isolation option Defines a slide window which has the same ratio of length and width of number plate but larger than it And then move this window around in the edge detected image to find the area has most edge information Perform plate resizing algorithm to find the exact boundaries of the number plate The isolated sub image will be displayed in the localization panel 12 B Sc Hons in Software Engineering CW228 Project Title Number Plate Recognition Name Dongfan Kuang Login ID C00131031 Supervisor Nigel Whyte 4 4Sub system of Segmentation 4 4 1 Overview Character Segmentation Binarization Vertical Projection Figure 10 modular design of characte segmentation sub system Character segmentation has two modules
10. nacadt 11 4 3 1 OVENI EW niinen AE A a ash ca EAA AEE E EEA 11 4 3 2 Detailed Description of Each Module ccccssscecssstececesssseeeeeessneeeeessaeeeeseeea 11 4 4 Sub system of Segmentation ccccssscccessscsccesssssescecsesssseceesssecesssoonseeceesseseceessesecesess 13 AAAs SOVEIVICW noniine aiena Eea E E Ep Ea Saa E EEA a 13 4 4 2 Detailed Description of each module s sssssssssesssssesssesssssrrsssrrernsereessesrensssreenese 13 4 5 Sub system of Character Segmentation ceccccesssscececsssteceecssseeeecesseeeeeecseneeeeeesseeneeees 14 AIDA E O A EA E E A 14 4 5 2 Detailed Description of Each Module cccccsscecssssececesssseeeceesseeeeeesenaeeeenseea 14 EIU LUTTO AAA E E E E E EA TAT E dedemea sean awaeap adenine 15 B Sc Hons in Software Engineering CW228 Project Title Number Plate Recognition Name Dongfan Kuang Login ID C00131031 Supervisor Nigel Whyte 1 Introduction This document describes the requirements for developing the NPR system The NPR system can read the registration number automatically from the number plate area in digital image and change the number into plain ASCII text This project is only deal with the software part of NPR it will analysis the inputted digital image and output the registration number of a vehicle The steps involved in NPR include image pre processing which contains the grayscale and Gaussian smoothing Next step is called plate isolation separate number
11. plate area from the whole image After find the number plate area the individual characters in the number plate are separated and identified through character segmentation and optical character recognition by using vertical projection algorithm and grid feature method B Sc Hons in Software Engineering CW228 Project Title Number Plate Recognition Name Dongfan Kuang Login ID C00131031 Supervisor Nigel Whyte 2 Architectural Design As is shown in figure 1 the NPR system has four main components Pre processing component Localization component Segmentation component and Recognition component NPR System Pre Number Plate Character Character processing Localization Segmentation Recognition Figure 1 NPR system architectural Pre processing sub system first applies grayscale algorithm to the inputted image to change the color image into grayscale format and then applies Gaussian smoothing algorithm to the inputted image to removes the noises Pre processing can improves the success rate of number plate localization and character recognition Number plate localization sub system first takes the pre processed image provided by pre processing sub system and then applies Canny edge detection algorithm to the pre processed image to detect edges in the image The last step is found out the number plate area by using a number plate isolation algorithm Character segmentation sub system is used to separate every single character in
Download Pdf Manuals
Related Search