Russell Kirmayer, HW 3. Questions #1,2,3:

Vision Lab Images directory Source CodeQuestion 1:

1a. A 3x3 mask that has 1/9th in each of the boxes will implement the average operator

1b. Convolution, which is used for image blurring and correction

1c. Set the mask as a 3x3 matrix M, which is convoluted with a pixel P. The data is only reclaimable if M is invertible, such that P*M*M_invert = P. However, our 1/9th mask as mentioned in 1a. is not invertible, so there will be a loss of data. This property is derived from the properties of invertible, square matrices

The center pixel will be weighted with a value of 1/2, and all the remaining, neighboring pixels will be weighted at 1/16th (8*(1/16) + 1/2 = 1).

Questions 2 and 3: