Project 1: Image Filtering and Hybrid Images

I resize all my input images to have same number of columns and rows right after loading them. The aspect ratio of the first image is preserved as a reference. The second image is forced into the new dimensions. After that low frequency and high frequency images are retrived by kernal convolution. Finally, Hybrid image is simply the summation of these two componenets.

Sample results

Figure below shows the results of running my code on a sample image (bir/plane) provided by the assignment. I add another sample which is politically correct. You can see Hillary turns into Trump as scaled down!

Low-passed and High-passed images

The hyprid image has two componenets: High-passed and low-passed images. Below you can see these componenets for the Hillary/Trump example.
Original High-passed
Original low-passed