/* * Author: Lucian Carata, calucian[]yahoo.co.uk * * This was a project towards the fulfilment of my bachelor thesis * in the Automatic Control and Computer Engineering Faculty, * Techical University of Iasi, Romania * * B.Sc. Theme: * () object recognition from video stream, * (future work:) 3D object reconstruction from camera motion (SfM) * B.Sc. Deadline: 20.07.2009 * Supervisor: Ph.D. Vasile Manta * * Current Status: Frozen (B.Sc Thesis Mark: 10/10) */ *** Note: This code is written as part of my B.Sc. Thesis. Please feel free to comment/send me suggestions or develop your own branches from my code. I will try to post here the progress of my work, as well as references to important articles in the field. My B.Sc Thesis focuses on real-time object recognition (in video data), as an aplication of the SURF algorithm, implemented on CPU, as well as GPU - CUDA). The SURF algorithm should be used in the end as a first step in 3D object modelling. [out of scope for the time being] The interface of the application will be written in WPF/C# , and as such will not be portable to linux environments. (This is due to my lack of knowledge/ experience in the GTK++ area, and the lack of time) Despite this, the library implementing the SURF algorithm is written in C++, and will be tested on both windows/linux before release. Practically, for making the whole application work under linux, one must write two OS-specific layers: (i) the Camera HAL Layer (image acquisition; using DirectShow on Windows) (ii) the Presentation Layer (GUI; using WPF on Windows). ***