Description
In this project each student must implement three versions of a 7×7 mean filter. The first
version should use basic 2D convolution. The second version should use separable filters
(1×7 and 7×1). The third version should use separable filters and a sliding window.
Any pixel for which the convolution extends outside the image boundary (sometimes
called “edge cases”) should be given an output value of zero.
All three versions of the filter should produce the exact same output. This must be
verified by comparing the images using “diff” or a similar program and showing the
method used and result.
Each version of the filter should be timed, and the typical amount of time reported (for
example, the average amount of time over ten runs).
C-code for smoothing a 512 x 512 image is posted at the class website. The program
contains code demonstrating a 3×3 mean filter using basic 2D convolution, and how to
time a piece of the program. You can use this code to derive a solution to this lab.
You must write a brief report that verifies result similarity between the 3 versions, and
summarizes and compares the amounts of time each version of the filter takes.
Submit your C-code (as an attachment) and report (as an attachment) to
ece_assign@clemson.edu. Use as subject header ECE4310-1,#1 or ECE6310-1,#1. This
email is due by midnight of the due date.
