Project 0

Project 0

Project 0

Question Comments Modified Image Original Image

First way to remove high frequencies, with ifft2 and fft2 It should be noted that low frequencies are stored towards the top left corner OF EACH QUADRANT in files like JPG's. Hence after taking the fft2 we should filter out every quadrant except the 2nd (for every quadrant). Also the smaller the quadrant is allowed to be, the blurrier the image.

fftshift below centers the low frequencies to the center while the higher frequencies are towards the edges of the picture.

http://www.flickr.com/photos/marhowie/4009398441/

Second way, use fspecial to get a gaussian mask and apply with conv2 There is some amount of cropping done on the image as the borders of the convovled image are very noisy
http://www.flickr.com/photos/gallry/3334852802/

Sharpen an image using fspecial('unsharp') and imfilter By using circular as the boundary option, the values outside the convolution array assume pixel coloration were periodic within the array
http://www.loc.gov/pictures/collection/prok/item/prk2000002140/

Insert a ghost into an image Convolve a potrait with a laplacain to get just the sharp lines and then line it up in another picture
http://www.flickr.com/photos/29570073@N03/2919239165/

http://www.flickr.com/photos/68307703@N00/2258381658/

Reveal hidden nightlife in a dark image Note the individual building details, like rooftop covers, plants behind the wall and license plate starting with the letter N
http://www.flickr.com/photos/16508730@N00/396253091/sizes/m/in/photostream/

Project 0
Question	Comments	Modified Image	Original Image
First way to remove high frequencies, with ifft2 and fft2	It should be noted that low frequencies are stored towards the top left corner OF EACH QUADRANT in files like JPG's. Hence after taking the fft2 we should filter out every quadrant except the 2nd (for every quadrant). Also the smaller the quadrant is allowed to be, the blurrier the image. fftshift below centers the low frequencies to the center while the higher frequencies are towards the edges of the picture.		http://www.flickr.com/photos/marhowie/4009398441/
Second way, use fspecial to get a gaussian mask and apply with conv2	There is some amount of cropping done on the image as the borders of the convovled image are very noisy		http://www.flickr.com/photos/gallry/3334852802/
Sharpen an image using fspecial('unsharp') and imfilter	By using circular as the boundary option, the values outside the convolution array assume pixel coloration were periodic within the array		http://www.loc.gov/pictures/collection/prok/item/prk2000002140/
Insert a ghost into an image	Convolve a potrait with a laplacain to get just the sharp lines and then line it up in another picture		http://www.flickr.com/photos/29570073@N03/2919239165/ http://www.flickr.com/photos/68307703@N00/2258381658/
Reveal hidden nightlife in a dark image	Note the individual building details, like rooftop covers, plants behind the wall and license plate starting with the letter N		http://www.flickr.com/photos/16508730@N00/396253091/sizes/m/in/photostream/