Project1: [2007-2009] Ege University Campus Car Entrance Tracking System with LPR Project,
[Budget: $7200]
Advisor: Asst.Prof.Dr. Muhammet Gokhan Cinsdikici
In this project, LPR project is adopted for Ege University. This project aims to support security systems in campus. University security department has reported several violations related with unknown automobiles. They keep track of cars manually. This project aims to report all cars automatically and alerts the unidentified cars to security department.
Project2: [2004-2008] [TÜBİTAK – 104 S 236] Designing and Implementing Non-invasive Electrochemical Measurement Equipment Hardware & Software Systems for Glucoses Inspection.
[Budget: $100000]
Advisor: Asst.Prof.Dr. Muhammet Gokhan Cinsdikici
In this project, Pharmacy department has designed new non-invasive equipment to measure glucoses. This hardware is located in wrest watch. The base of the watch is covered with special chemicals that react under certain micro voltage energy levels with human tears. My position in this project is constructing a neural network model for estimating threshold level of the glucoses for patient by just reading the measurements.
Project3: [1999-2003] LPR (License Plate Recognition) Project – Supported by Aselsan Military Res./Dev.Co.,
[Budget: $3000]
Advisor: Prof.Dr.Turhan Tunali Project Res: Res.Ass.Muhammed Gokhan Cinsdikici
In this project, the triggered image (including car and environmental info) is taken as input data. Over this input, we are trying to extract plate-like regions (candidates). First, image is scaled by Kaiser filtering (Also simple interpolation methods are tried). Then vertical edge detection is done. Vertical edges give some clues about plate character density (Some other techniques like symbol searching, line projection tracking are also studied). After isolating all candidate regions (skewed/rotated), they are rotated automatically by Radon/Hough transformation. After rotation is corrected, we segment all candidates into small objects (character-like shapes) with our method of Bidirectional Vertical Thresholding (based on simple gradient direction). Number of objects defines the candidate as real plate. After all character objects are extracted, they are fed into recognition module. This module is composite module of feature extractor (principle component analysis unit) and recognizer (backpropagation neural network). After characters are recognized, the plate conversion is accomplished.
Project4: [2002-2004] GIS System for Ege University – Supported by Ege University
[Budget $25000, Project ID: 02-BTAM-001]
Advisor: Prof.Dr.Fikret IKIZ As Project Res: Res.Ass.Muhammed Gokhan Cinsdikici
This project is designed and implemented for developing GIS maps for Ege University infrastructure. In this work, I have studied on exporting network topology of the Ege University. The fiber optic cabling and Utp Cat5 cable topology for GIS system.
Project5: [2001-2003] E-Learning and ECDL Adaptation for Education in Ege University – Supported by Ege University
[Budget $25000, Project ID: 02-BTAM-001]
Advisor: Prof.Dr.Fikret IKIZ As Project Res: Res.Ass.Muhammed Gokhan Cinsdikici
This project is designed and implemented for developing GIS maps for Ege University infrastructure. In this work, I have studied on exporting network topology of the Ege University. The fiber optic cabling and Utp Cat5 cable topology for GIS system.
Project6: [1996-1997] Neural Network for Pattern Recognition – Supported by Ege University Engineering Department
[Budget $5000, Project ID: 94-MUH-003]
Advisor: Assoc.Prof.Dr.Yusuf Ozturk As Project Res: Res.Ass.Muhammed Gokhan Cinsdikici
Document analysis is the main subject of project. In the project, the text areas and image areas are tried to be differentiated. MAN and MAREN neural network models are studied in the project. The aim of these neural networks is to recognize characters. If the characters show a straight line and the lines are forming paragraph, then the area is marked as text area. If not, the area is marked as image area. Image segmentation and pattern recognition techniques are studied.