Peter Bailey
UMN address
bail0253@umn.edu
U of M contact info page
Assignment pdf
Assignment code
Problem 1
Histogram equalization and matching
Problem 1a code
Problem 1b code
Original image
Original histogram
Part A: histogram-equalized image
Equalized histogram
Smoothed equalized histogram
Part B: histogram-matched image
Matched histogram
Smoothed matched histogram
Problem 2
Noise filtering
Problem 2 code
Note: This code uses an
implementation
of the non-local means filter by Jose Vicente Manjon-Herrera.
Original image
Noisy image 1
Noisy image 2
Part A: 3x3 average 1
Part A: 3x3 average 2
Part A: 7x7 average 1
Part A: 7x7 average 2
Part A: 9x9 gaussian 1
Part A: 9x9 gaussian 2
Part B: 3x3 median 1
Part B: 3x3 median 2
Part B: 7x7 median 1
Part B: 7x7 median 2
Part C: Wiener filter 1
Part C: Wiener filter 2
Part D: non-local means filter 1
Part D: non-local means filter 2
Part E: NLM median filter 1
Part E: NLM median filter 2
Problem 3
Problem 3 code
3x3 mean smoothed, Laplacian-sharpened 1
3x3 mean smoothed, Laplacian-sharpened 2
7x7 mean smoothed, Laplacian-sharpened 1
7x7 mean smoothed, Laplacian-sharpened 2
3x3 median smoothed, Laplacian-sharpened 1
3x3 median smoothed, Laplacian-sharpened 2
7x7 median smoothed, Laplacian-sharpened 1
7x7 median smoothed, Laplacian-sharpened 2
9x9 gaussian smoothed, Laplacian-sharpened 1
9x9 gaussian smoothed, Laplacian-sharpened 2
Non-local means smoothed, Laplacian-sharpened 1
Non-local means smoothed, Laplacian-sharpened 2
Problem 4
Problem 4 code
Original image
Blurred, noisy image
3x3 median-filtered image
Restored, sharpened image
Problem 5
Problem 5 code
Original image
Red edge magnitude
Blue edge magnitude
Green edge magnitude
Vector color edge magnitude
Normalized difference: vector edge magnitude - combined edge magnitude
Problem 6
Problem 6 code
Original image (8 bits per pixel)
Normalized, DM-encoded image 1 (predictor error; 1 bit per pixel)
Decoded image 1
Normalized, DPCM-encoded image 2 (predictor error; 2 bits per pixel)
Decoded image 2