Computer Vision I at NTU 2017 Fall.

This course has 10 homeworks. The 10 homeworks are as follows:

1. Basic Image Manipulation
2. Basic Image Manipulation
3. Histogram Equalization
4. Mathematical Morphology - Binary Morphology
5. Mathematical Morphology - Gray Scaled Morphology
6. Yokoi Connectivity Number
7. Thinning
8. Noise Removal
9. General Edge Detection
10. Zero Crossing Edge Detection

0. Environment
1. Basic Image Manipulation
2. Basic Image Manipulation
3. Histogram Equalization
4. Mathematical Morphology - Binary Morphology
5. Mathematical Morphology - Gray Scaled Morphology
6. Yokoi Connectivity Number
7. Thinning
8. Noise Removal
9. General Edge Detection
10. Zero Crossing Edge Detection

• Programming Language: Python 3
• Programming IDE: Visual Studio Code
• Operating System: Windows 10 x64

• Part 1 of this homework is writing a program to generate the following images from lena.bmp.

  • Up-side-down lena.bmp.
  • Right-side-left lena.bmp.
  • Diagonally mirrored lena.bmp.
  • Code: HW1.1

• Part 2 of this homework is using any kind of software to do the following things:

  • Rotate lena.bmp 45 degrees clockwise.
  • Shrink lena.bmp in half.
  • Binarize lena.bmp at 128 to get a binary image.
  • Code: HW1.2

• Report

• Part 1 of this homework is to binarize lena.bmp with threshold 128 (0-127, 128-255).

  • Code: HW2.1

• Part 2 of this homework is to draw the histogram of lena.bmp.

  • Code: HW2.2

• Part 3 of this homework is to find connected components with following rules:

  • Draw bounding box of regions.
  • Draw cross at centroid of regions.
  • Omit regions that have a pixel count less than 500.
  • Code: HW2.3

• Report

• This homework is to do histogram equalization with following rules:

  • Adjust the brightness of lena.bmp to one-third.
  • Do histogram equalization on dark image.
  • Show the histogram of the final image.
  • Code: HW3.1

• Report

• This homework is to do binary morphology with following rules:

  • Please use the octagonal 3-5-5-5-3 kernel.
  • Please use the “L” shaped kernel to detect the upper-right corner for hit-and-miss transform.
  • Please process the white pixels (operating on white pixels).
  • Five images should be included in your report: Dilation, Erosion, Opening, Closing, and Hit-and-Miss.
  • Code: HW4.1

• Report

• This homework is to do gray scaled morphology with following rules:

  • Please use the octagonal 3-5-5-5-3 kernel.
  • Please take the local maxima or local minima respectively.
  • Four images should be included in your report: Dilation, Erosion, Opening, and Closing.
  • Code: HW5

• Report

• This homework is to do Yokoi connectivity number with following rules:

  • Please binarize leba.bmp with threshold 128.
  • Please down sampling binary.bmp from 512x512 to 64x64, using 8x8 blocks as unit and take the topmost-left pixel as the down sampling data.
  • Print Yokoi connectivity number to text file.
  • Code: HW6

• Report

• This homework is to do thinning operation with following rules:

  • Please binarize leba.bmp with threshold 128.
  • Do thinning operation on binary image.
  • Code: HW7

• Report

• This homework is to do noise removal with following rules:

  • Generate Gaussian noise with amplitude of 10 and 30.
  • Generate salt-and-pepper noise with probability of 0.1 and 0.05.
  • Use the 3x3 and 5x5 box filter on noise images.
  • Use the 3x3 and 5x5 median filter on noise images.
  • Use opening-then-closing and closing-then-opening filter on noise images.
  • Calculate the signal-to-noise-ratio (SNR) of noise images.
  • Code: HW8

• Report

• This homework is to do general edge detection with following rules:

  • Robert’s operator with threshold of 12.
  • Prewitt’s edge detector with threshold of 24.
  • Sobel’s edge detector with threshold of 38.
  • Frei and Chen’s gradient operator with threshold of 30.
  • Kirsch’s compass operator with threshold of 135.
  • Robinson’s compass operator with threshold of 43.
  • Nevatia-Babu 5x5 operator with threshold of 12500.
  • Code: HW9

• Report

• This homework is to do zero crossing edge detection with following rules:

  • Laplacian mask 1 with threshold of 15.
  • Laplacian mask 2 with threshold of 15.
  • Minimum variance Laplacian with threshold of 20.
  • Laplacian of Gaussian with threshold of 3000.
  • Difference of Gaussian with threshold of 1.
  • Code: HW10

• Report