The student Ekhlas Falih obtained her PhD degree from the Department of Computer Sciences / University of Technology for her dissertation. That proposed a system used the BOVW to understand the event in which the video data is extracted and compiled to generate a group Of the visual words that form the visual dictionary. As a pre-processing phase, the video images are converted to gray and then converted using the Haar filter for the purpose of implementing the proposed algorithms with minimal complexity. The first stage of the proposed visual system is to extract important points from the images. The proposed system used FAST characteristics to extract angles as important points. The FAST has been improved by proposing an adaptive threshold rather than a fixed threshold to reduce the space of attributes and increase the speed of implementation of the proposed system. After discovering the important points, the proposed system uses SURF as a second stage of the video bag system to build the local descriptor for these important points and generate a signature for the image that is fixed with rotation and zoom.

This was on Sunday 22/4/2018 and on the discussion hall in the department. The discussion committee consisted of Prof. Ahmed Tariq Sadiq, Dr. Matheel Imad Eddin Abdel Moneim, (Department of Computer Sciences), Dr. Esraa Hadi Ali (University of Babylon), Faculty of Information Technology, Dr. Mohamed Saheb Mehdi) from the University of Nahrain / Faculty of Science / Computer Science Department in the presence of the student supervisor (Dr. Abdul Amir Abdullah Karim).

Download Abstract