Faith Quist

16-311 Homework 3

1.       Digital Images

a.       My 3x3 mask that implements the average operator

b.      I used this mask on a picture of a bagpiper. The results are shown below.

Figure 1: Piper original†††††††††††††††††††††††††††††††††††††††††††† Figure 2: Piper blurred

c.       This operation loses information in two ways. First, unless the picture is extended, the mask would cut off the pixels around the border, creating a smaller image. Furthermore, since every pixel becomes the average of the pixels around it, it is impossible to deduce what the previous values were. Although you know what the pixels in the mask would have to add up to, you do not know how that sum was divided among those pixels.

d.      Weighted mask. I chose this weight because it more than doubles the amount that the central pixel contributes, but the neighbors are not made totally insignificant.

2.       Convolution

a.       Convolution is commutative, as shown by the following pictures. The horizontal mask is . The vertical mask is

piper_hvert.jpgpiper_vhorz.jpg

Figure 3: From left to right: image*horizontal, image*vertical, image*horizontal*vertical, image*vertical*horizontal

 

 

3.       Martian

 

 

 

 

 

 

Vision Lab

         Thresholding was done using a value that minimized the variance of the two resulting groups.

         Contrasting was achieved by giving pixels with an original value (z) a new value corresponding to the range maximum times the cumulative distribution function of the histogram at z.

         The histograms show values ranging from 0 to 255. The maximum bin of each histogram has a column height equal to the histogramís height.

Original

Thresholding

Contrasting

Histogram

mona.jpg

mona_threshed.jpg

mona_contrasted.jpg

mona_histogram.jpg

letterC.jpg

letterC_threshed.jpg

letterC_contrasted.jpg

letterC_histogram.jpg

tshirt.jpg

tshirt_threshed.jpg

tshirt_contrasted.jpg

tshirt_histogram.jpg