Tuesday, April 13, 2004

Several problems in Beads simulation and tracking

1. When save the BMP file, video is gone. OpenGL picks a pointer to the device at one time: m_hDC = ::GetDC(this->m_hWnd); or m_hDC = CreateCompatibleDC(NULL); The second one sets up a Null pattern and saves the values of every pixel.
2. I didn't get the data in the memory "m_data". The format of m_data is B G R B G R... with character values. The data in the memory is not from regular left-top, seems it is from left-bottom, just need to flip it. Rewrite all the data in the memory to BMPPIXEL variable and use BMP2DISPLAY to show it on the screen, that means we don't need to use OpenGL to show the pictures.
3. I have no confidence with the "force attraction model", the single beam may do something such like drag the bead from the votex a little bit.
The program can save the image data to matrix "m_data" and at the same time the video is shown. We process the image data and put the tracking marker to the image. I need to try "Correlation" today. How to get the nano-bead locations in the first image is the first problem, then choose "trust regions" and do FFT, multiply with the FFT of nano-bead model, iFFT, choose the peak postion. Twice FFts may drive the program slow. It is easy to try in Matlab first.

No comments:

Post a Comment