Damir Arh's Corner

During the Computer Vision course at university I designed a small program that recognises time from images of a particular analog clock. With minor changes the program could be adapted for images of any analog clock. The application is written in Delphi for Win32 platform.

The program along with sample images and complete source code is available for download. It might be of interest to those who would like to take a look at an implementation of some of the classic computer vision algorithms. To achieve the time recognition I had to implement linear filtering with image convolution, an approximation to the Canny edge detector and two adaptions of Hough transform for recognition of circles and clock hands.

As a part of the Artificial Intelligence and Symbolic Programming course at university me and a group of fellow students (in alphabetical order: Uros Cibej, Matija Jekovec, Gregor Leban, Mitja Lustrek, Martin Znidarsic) under the supervision of Aleks Jakulin got involved in a project which tried to achieve sensible tactical behavior of a group of soldiers controlled by a human player at a higher abstraction level.

Although the project was only meant as a learning polygon for different approaches to real time tactical artificial intelligence, we still achieved quite satisfactory results. A short presentation at the end of the course classes that was extremely well received got us an invitation to the Solomon seminar at Jozef Stefan Institute.

The complete project can be found in the downloads section of my site. It consists of the source codes which are due to the experimental nature of project pretty messy and need Simple DirectMedia Layer libraries and Visual C++ 6 to be compiled, a brief technical report which describes the used algorithms, a Power Point presentation for the Solomon seminar and a set of animations in DivX format which demonstrate the interesting behavior of the soldiers.

Unfortunately most of the above materials are in Slovenian language and will therefore be of limited use to the non-Slovenian speakers. The only exceptions to this are probably the source codes and the animations.