Question Comments Original Image Modified image
1a. The approach to eliminate the high frequencies involved converting the image to discrete time using fft. The discretized image was then multiplied by a mask that portrayed the low frequency areas to just be at the corners of the image. The low-frequency image was then converted back to analog using the inverse fft. However, there are lossless issues from the FFT that causes the image to not be filtered appropriately. Mask

Image

1b. These images show how convolution was used with a gaussian filter to eliminate the high frequencies of the image. Although it eliminated more than just the high frequencies, this technique kept the frequencies of the corners better than the previous method.
2 The image of the flower was passed through a filter that sharpened the edges. This is noticeable on the edges of the flower.
3 The ghost of the dog was inserted into the grill page. A mask was created to crop only the dog's face. Then the cropped image was reduced to .35 of its original intensity. The two images were then added together to create the ghost dog in the smoke.
4 The image I changed was taken outdoors with an SLR. The ISO on the camera was set to 3200. As a result, applying histeq on the image caused more noise than an image that used less ISO.