Dissertation > Excellent graduate degree dissertation topics show

Research and Realization of Vehicle Logo Detection Technology Based on Video Image

Author: JiangBo
Tutor: PengQiang
School: Southwest Jiaotong University
Course: Traffic Information Engineering \u0026 Control
Keywords: Rough location of the logo Precise location of the logo Edge directionhistogram Profile characteristic Texture feature BP neural network Credibility Classifiercombination
CLC: TP391.41
Type: Master's thesis
Year: 2012
Downloads: 145
Quote: 1
Read: Download Dissertation

Abstract


Statistics show that, in China, about one hundred thousands vehicles are stolen each year and more than three hundreds are stolen each day. According to the survey of past criminal assaults from public security organs, the original color of vehicles is often changed by criminals after their steal and make false license plate or License plate cover in order to hide it’s crimes. It’s very difficult to identify vehicles which had been stolen only using license plate information and vehicle recognition technology. However, an important imformation of vehicles is the logo that can be used to indentify vechiles, because logos are not easy to be replaced, and most criminals do not want to replace the logo of luxury vehicles which would reduce the value of the vehicles. It has great value for polices catch the criminals with assistant of vehicles’logo identification. This article’s is main purpose to solve problems of logo location and logo identification in low resolution vehicles videos. Because the traditional mthods only use one kind of vision features to classify different vehicles, they can not achieve good performance.This paper focuses on how to roughly locate logo, how to precisely locate logo, feature extraction and identification. The main tasks include the following aspects:First, approaches about how to locate logo are studied. The author get the vehicle image by using background update method and the frame difference method, The author get the rough area of logo by using the relative position between the license plate and the logo, and the author get the precise area of logo by edges strength and mprphological processing. By logo located, we can reduce the logo of the surround environment to impact of identification, because of accurately position of the logo is prerequisite of logo correct identification.Secondly, the paper used direction histogram, contour and texture features to describe the logo’s features. Edge direction histogram can reflect shape information of the logo and be low impacted by light, Contour feature can directly reflect external contour information of vehicle logo, and2D Gabor filter texture features extraction can reflect grayscale changes of the vehicle logo from the overall. The author get template features of dge direction histograms and contour feature by using the means method, which not only can improve the recognition rate but also makes the recognition result more stability.2D Gabor filter parameters are set by comparing the experimental method and optimization method, which made the textures features of logo well difference between categories.Finally, this paper analyzes the advantages and disadvantages of the three feature extraction methods, using multiple features extraction methods and combinational classifiers to make up the lack thraditional methods. First, using the template matching method to match the logo edge direction histograms and contour, then, using the credibility to judge the match results whether meet the requirements, if the results do not meet the requirements, the author will do the identification again. In this paper, though using the combination identification of classifiers and multiple features extraction methods, the author get geate improvement in recognition rate.

Related Dissertations

  1. Research on Feature Extraction and Classification of Tongue Shape and Tooth-Marked Tongue in TCM Tongue Diagnosis,TP391.41
  2. Research on Visual Servo System of Mechanical ARM,TP242.6
  3. Municipal tourism land use planning environmental impact assessment,X820.3
  4. Study on Taste Characteristic of Taste Peptide Enzymatic Production from Oyster Base on A Neural Network Method,TS254.4
  5. Analysis on Registration System of Forest Right,D923.2
  6. The Research on Evaluation of Living Status Systems of Expressway Relocated People,D523
  7. The Research on Texture Synthesis Technology from Cloud Theory & Been Evolution Genetic Algorithm,TP391.41
  8. Analysis on the Credibility of Government in Present China’s Society,B82-05
  9. Mine Risk Information Integration and Intelligent Early Warning,X936
  10. Real Property in the Protection of Third Party,D913
  11. Research of Virus Detection Methods Based on Multiple Anti-virus Softwares Collaboration,TP309.5
  12. The Optimization Research of Feedforward Neural Network Based on Genetic Algorithms,TP183
  13. JSYJ Company Purchasing Risk Management Research,F426.92
  14. The Research on the Highway Pavement System with Neural Network and Combination Forecast,TP315
  15. Research of Orange Quality Classification Technology Based on Computer Vision,TP391.41
  16. Researches on Urban Rail Transportation Operation Management System Test and Evaluation Method,TP311.52
  17. A Study on the Township Government’s Execution in China,D625
  18. Optimization Study on Gating System and Molding Process Parameters of Injection Mold Based on Simulation,TQ320.662
  19. A Research on Monitoring and Early Warning System of Zhejiang Marine Economy,F127
  20. Study on Fabric Defect Detection and Sutomati Grad-ing System,TP391.41
  21. Research on Listed Tourism Companies Financial Crisis Prewarning Model Based on the Theories of Grey Neural Network,F224

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net  Mobile