TEXTURE MORPHING | |
Halfway through my time in Minnesota I started working on a second project focusing on texture
morphing. The subject is the proposed topic for Haleh Hagh-Shenas's dissertation. Haleh just started
working on the project at the beginning of the summer so I was able to discuss preliminary ideas on how
to go about the project. I was very excited to wrok with Haleh, especially since true texture morphing has not been done before. There is very little research, and most of the previous work has been on filters and histograms. Texture morphing can be compared to facial morphing in the sense that the purpose is to change the first texture into the second while keeping the integrity of both images as much as possible. For example, we have all seen ads or movies where a face has changed from one person to another. The animator moves the facial features to morph into the next frame. You can tell the image is a face at every point in time. Haleh would like to apply this same idea to textures so that you can tell the characteristics of the texture. While it is easy enough to explain, finding and implementing an algorithm to morph between any patterned textures has proven to be difficult. The problem of morphing is complicated due to the lack of similar features and interference from the texture itself. Haleh started me out reading papers on texture modeling, Gabor filters, and the Earth Mover's Distance. While I read these, I also started working with Matlab to see what kind of functions are in the Image Processing Toolkit. As of right now I have a program that can move similar colored pixels from the source texture towards the destination texture.
| |
Main -- People -- Research -- Life Outside the Lab |