Dear CCSE Community,

You are all cordially invited to attend this week’s two seminars by our CCSE graduate students. The seminars will take place December 10 at 2:30 pm in Building 22: Room 119. This week’s seminars are:

Seminar 1

Title: Recognition of Arabic Bank Check Digits Using Structural Features

Abstract: In this work we present a technique for automatic recognition of Arabic (Indian) bank check digits using novel structural features that are based on the structure of Arabic digits. In addition, a rule-based classifier is implemented. Moreover, Support Vector Machine (SVM), LogitBoost, and RandomForest classifiers are used in this work. CENPARMI Arabic check database is used. Classifiers’ fusion with majority voting is used with and without the rule-based classifier. Recognition rates of 98.95% and 99.08% are achieved using fusion of the statistical classifiers alone and with the rule-based classifier, respectively. The achieved recognition rates outperform the published work using the same database. The experimental results indicate the effectiveness of the structural features and the rule-based classifier for recognizing Arabic digits.

Speaker: Mr. Sameh A. Bellegdi

Bio: Mr. Sameh A. Bellegdi received his B.S. degree in computer science from Al-Ahgaff University, Yemen in 2007 and his M.S. degree in computer science from King Fahd University of Petroleum and Minerals (KFUPM) in 2013. He is currently pursuing his Ph.D. degree in computer science and engineering at KFUPM. His research interests include Pattern Recognition, Machine Learning, Arabic Computing, and Computational Biomedicine.

Seminar 2

Title:  Fractal Image Compression using Modified Iterated Function Systems

Abstract: Image data Compression based on fractal theory is fundamentally different from conventional compression methods, its idea is to generate a contraction operator whose fixed point approximates the original image in a complete metric space of images. The specification of such operator can be stored as the fractal code for the original image. The contraction mapping principle implies that the iteration of the stored operator starting from arbitrary initial image will recover its fixed point which is an approximation for the original image. This Contraction mapping is usually constructed using the partitioned IFS(PIFS) technique which relies on the assertion that parts of the image resemble other parts of the same image. It then, finds the fractal code for each part by searching for another larger similar part. This high costly search makes fractal image compression difficult to be implemented in practice, even it has the advantages of a high compression ratio, a low loss ratio, and the resolution independence of the compression rate.

In this seminar, we present the fractal image compression(FIC) using Iterated Function Systems(IFS). We then propose a modified IFS and its corresponding algorithm proving its properties that make it not only a fractal operator but also more effective than the standard one. The experimental results are presented and the performance of the proposed algorithm is discussed.

Speaker:  Mr. Baligh Mohammed Al-Helali

Bio: Mr. Baligh is a master student in computer science department at KFUPM. He received his undergraduate degree in Computer and Mathematics from Ibb University Ibb-Yemen 2003, and master in Applied Mathematics from Taiz University Taiz-Yemen 2010. His research interests focus on Computer Vision, Fractal geometry, Artificial Intelligence.

Regards,

Dr. Mohamed El-Attar

Associate Professor :: Software Engineering

Information and Computer Science Department

King Fahd University of Petroleum and Minerals

Kingdom of Saudi Arabia

​