CS 143 / Project 1 / Image Filtering and Hybrid Images
Hybrid image of a cat and a dog
The objective of this project is to create a hybrid image by combining two images with menthod mentioned in a simplified version of the SIGGRAPH 2006 paper by Oliva, Torralba, and Schyns. The hybrid image looks like one image at a distance. When you look up close, then it looks like the other one.
Approach
The algorithm mentioned in the paper basically use a low-frequency filter to create a low pass filterd image and a high-frequency to create a high pass filtered image. Two filtered images are then combined together to create the hybrid image.
 |
 |
| cutoff frequency = 7 |
The paper suggests using two cut-off frequencies for two images repectively. In this project, I use one frequency only for both filters. I generated grayscal results and by comparing them with colored ones, I find some of them looks better in grayscale when the color of the low frequency image doesn't match the high frequency one. Because in colored images, that makes low frequency image stands out even in a short distance.
 |
 |
| cutoff frequency = 7 |
my_imfilter Implementation
I pad the image by reflecting the pixels over the boundry and generate a new image. Then we can iterate through the new image apply the filter.
Results
Dog and Cat
Fish and Submarine
Bird and Plane
Motorcycle and Bicycle
Hays and Djokovic
Marilyn and Einstein
