next up previous
Next: February 15, 2007: Registration Up: DDDAS Weekly Meetings The Previous: March 1, 2007: Laser

February 22, 2007: Image Registration

Aman Rathi presented his progress on Image Registration Problem for the project [ppt] As we know that�we would be getting�a prostate image days before the actual operation day. Hence,�the mesh computations etc would have already been carried out on that preliminary image.�On the actual operation day, we will again get images from MDA of the same prostate but they may be shifted, rotated etc due to contingent factors. To do any meaningful computations on this new image and to make sense of the computations on the previous image, we need to register this new image to the previous image. And hence, we require Image Registration of 3D images. The Image registration is of two types : Feature based and Area based. Feature based registration requires use of landmarks or features which unfortunately are hard to find in biological tissue images. Hence, we would go for Area based registration which uses FFT and other structural features of the image to co-register them. Then, we outlined the�Image registration algorithm as stated in the paper [1]. The algorithm finds out the�shift. rotation and the scale difference between the two images. We also set up a small experiment on the AVS to show that�the code works well for two 2D-images. Future work includes rigorous testing of the 2D code and converting�it to 3D.

References :

1.�R. Araiza et al. 3-D Image Registration Using Fast Fourier Transformation: Potential Applications to Geoinformatics and Bioinformatics


next up previous
Next: February 15, 2007: Registration Up: DDDAS Weekly Meetings The Previous: March 1, 2007: Laser