For this part, we took the camerman image and computed gradients in each direction. To compute the gradient magntiude, we computed sqrt(gradient_x^2 + gradient_y^2). This allows us to visualize the edge strength or the direction of greatest increase.
Original
Gradient in the x direction
Gradient in the y direction
Gradient magnitude
Binarized (threshold 0.25)
This time, we blur using a gaussian filter. This greatly reduced the noise as compared to 1.1. The edges are cleaner and more defined.
Blurred
Gradient in the x direction
Gradient in the y direction
Gradient magnitude
Binarized (threshold 0.05)
Derivative of gaussian dx
Derivative of gaussian dy
Gradient in x
Gradient in y
Binarized magnitude
Binarized (threshold 0.05)
In this part, we sharpened images by taking the high frequency components of an image and readding it to the image. We do this by removing blurry parts with a gaussian filter to get only the high frequencies. We combined this operation into 1 convolution by creating an unsharp mask filter.
Taj
Sharpened Taj
New york
Sharpened new york
Eiffel tower
Sharpened Eiffel tower
In this part, we overlayed and combined two images by taking the lower frequency components of 1 image and averaging it with the high frequency components of another. The overlay of hilfinger and denero did not work so well. The shoulders were misaligned and their hair style is different. Additionally, denero was smiling with his teeth while hilfinger was not.
Derek
Nutmeg
Derek and Nutmeg
Hilfinger
Denero
Hilfinger & denero
Drawn character (kakashi)
Drawn character (gojo)
Combined characters
Jennie
Jisoo
Jennie and Jisoo
Jisoo Fourier transform
Jennie Fourier transform
lowpass (jisoo) Fourier transform
highpass (jennie) Fourier transform
hybrid Fourier transform
In this part, we computed laplacian and gaussian stacks to do multiresolution blending
mask used for orapple and blended skyline
apple
orange
orapple
skyline day
skyline night
blended skyline
irregular mask
Star Wars character
Friend (name is Jedi)
Blended