Document Details

Document Type : Thesis 
Document Title :
Vehicle Plate Recognition
التعرف على لوحة المركبة
 
Subject : Faculty of Computing and Information Technology-Computing Sciences 
Document Language : Arabic 
Abstract : In this research, a new genetic algorithm (GA) technique is introduced to detect the location of a License Plate (LP) depending on the layout of its symbols. An adaptive threshold method has been applied to overcome the dynamic changes of the lighting conditions when converting the image into binary. Detection of all objects inside the unknown image is performed by the connected component analysis technique. A scale-independent Geometric Relationship Matrix (GRM) has been introduced to model the layout of the internal symbols of any LP to simplify the process of system adaptation. The introduced GA can be used in the localization problem of any 2-D compound object in plane images. Moreover, A new crossover operator, based on sorting, has been introduced which greatly improved the convergence speed of the system. The system has been implemented using MATLAB and various types of image samples have been experimented to verify the distinction of the proposed system. Encouraging results have been reported for many cases having variability in orientation, scaling, plate location, lighting conditions and the presence of different types of objects such as textures or edges. Examples of distorted plate images were successfully detected due to the independency on the shape and location of the plate. After the detection phase, symbols (or license plate digits and characters) are sent to the support vector machine classifier which recognizes these symbols and produces ASCII codes for the digits and characters represented by the symbol shapes of the license plate. Features representing each symbol shape are extracted after taking contour of each shape. Then, the centroid of the shape is calculated and the shape is divided into sixteen angular zones. Pixels are counting for each angular zone to obtain a feature vector of sixteen quantities. The feature vector of each symbol shape is used in both the training and recognition phases of the SVM. A high recognition rate has been recorded in our experiments that reach to a degree that permits using it in real applications. 
Supervisor : DR.Gibrael Al Amin Mohammad Abo Samra 
Thesis Type : Master Thesis 
Publishing Year : 1433 AH
2012 AD 
Added Date : Sunday, November 18, 2012 

Researchers

Researcher Name (Arabic)Researcher Name (English)Researcher TypeDr GradeEmail
فراس صالح خليفةKhalefah, Feras SalehResearcherMaster 

Files

File NameTypeDescription
 34324.pdf pdf 

Back To Researches Page