Project 2 Webpage

Srikar Talluri

Part 1.1

Compute and show the gradient magnitude image

Include a brief description of gradient magnitude computation

Gradient magnitude, like its name, is just a measure of how fast the intensity of the image changes from pixel to pixel. This allows us to detect edges within the image and see an “imprint” of the contents of the image. The way this is done is through taking the derivative of the pixel values (grey-scaled picture) in both the x and y direction.

To then create the binaarized version, we apply a threshold to see where the gradient changes more than a threshold limit. My threshold was in the form THRESHOLD = 1/n. The best n that I found was 8, for a threshold of ⅛, which created the image above.

Part 1.2

Create a blurred version of the original image by convolving with a gaussian and repeat the procedure in the previous part

What differences do you see?

We see that there exists a lot of noise in the picture in part 1.1. This noise and the number of other artifacts is drastically reduced in the convolved image in this part. It can be seen that the imprint and lines that represent the edges of the photographer are much better and cleaner and smoother.

Now we can do the same thing with a single convolution instead of two by creating a derivative of gaussian filters. Convolve the gaussian with D_x and D_y and display the resulting DoG filters as images.