Programming Project #0 (proj0)15-463: Computational Photography
Natasha Kholgade
Programming Project #0 (proj0)
1a. Using FFT to remove high frequencies: When an image of the Golden Gate bridge was used, it is observed that some of the high frequency components at the right extreme of the bridge are removed (vertical bars), but there is a lot of aliasing due to the removal of the high frequencies in the other regions.
FFT:
1b. Using a gaussian to remove high frequencies: When a gaussian with a spread of 4 was used on the image, the high frequency components are again removed but now the mid-frequencies are blurred out quite a bit (the left bars of the bridge, the horizontal and diagonal bars).
2. Sharpening an image: A careful examination of the two images shown below will reveal that veins along the leaf insect's body and the detail around his eyes have been exaggerated using the 'unsharp' filter.
3. Ghost into an image: For this purpose, the image of a 'lonely street' (first image) and the silhouette of a man in a trenchcoat (second image) were obtained from the web. A point was chosen in the lonely street image in which to place the silhouette, and a shift mask was designed with those coordinates. The silhouette was resized, inverted and convolved with a gaussian of spread 2.5 (third image). The resulting ghost image was convolved with the shift mask to place it in the desired location (fourth image). The lonely street image was dot-multiplied with an inversion of the shift mask convolved with the resized silhoutte (fifth image) and added to the shifted ghost to get the final image of a ghost on a lonely street (sixth image).
4. Revealing night life: The image of a night scene is revealed to using a histogram equalization to contain windows and patches on the walls
