ImageProcessing with NumPy

• Project from HCMUS's 2023 Applied Mathematics and Statistics course.
• Disclaimer: This program only supports RGB color model. Others such as RGBA, CMYK, or HSL, etc. might produce questionable results or not work at all.

Content

• lab03_project02.ipynb contains all the tasks and requirements of the project.
• Report.pdf includes my documentation for all the algorithms and required tasks.
  • First full-fledged report written in LaTeX, quite a pain but also very hype.
• sample folder holds the pictures that were used during implementation and documentation.
• figures folder includes every figure featured in Report.pdf.

ImageProcessing.ipynb

• The Miscellaneous functions exist solely for handling inputs and outputs, they definitely are NOT the main focus of the notebook.
  • I recommend skipping them entirely and just reading the descriptions in the Main function section.
• Past implementations at the bottom of the notebook are not entirely unusable, they are only inferior to the current implementations.
  • I think they are worth keeping to reflect the growth of the project.

Getting Started

Prerequisites

• Any IDE, preferably VSCode.
• Python 3.11 and Jupyter package.
  • Alternatively, the .ipynb file can be run on Google Colab or Anaconda.
• Everything from the Import libraries section of the notebook.

Installation

• Clone the repo.

Usage

• Make sure Import libraries is run at least once.
• For saving, 2 supported filetypes are .png and .pdf.
  • To opt out of saving, either leave the input blank or mess it up.
• Inputs must follow the format of <img_file> <function> <parameters>. Inputs do NOT need to be fully complete. As long as the <img_file> is correct, the <function> and especially <parameters> are optional. The <parameters> varies across functions, which is noted below.
  • Every input concerning tuples must be in the format of <Y>x<X>. E.g., 420x690 means y=420,x=690 or height=420,width=690.
  • E.g., img.jpg 7 0.5 600x400, img.png 7 100, img.jpg 7, img.png 4, etc.

Input notes

0. All - Inputting only the <img_file> will default to this function. <parameters> is NOT required.
1. Brightness - <factor>, float denoting how much the brightness will be adjusted.
   • E.g., -0.5 darkens by 50%, 0.35 brightens by 35%.
2. Contrast - <factor>, float denoting how much the contrast will be adjusted.
   • E.g., -0.5 adjusts by -127=int(-0.5*255), 0.35 adjusts by 89=int(0.35*255).
3. Flipping - <axis>, int representing the direction to flip the image.
   • 0 for vertical, 1 for horizontal, <blank> for both.
4. Grayscale and Sepia - 0 for grayscale, 1 for sepia, <blank> for both.
   • After 0 for grayscale, 0 for average, 1 for luminosity, <blank> for both.
5. Blurring and Sharpening - 0 for blurring, 1 for sharpening, <blank> for one execution of each.
   • After 0 or 1, input an int denoting how many times to blur or sharpen.
6. Center cropping - Two methods of inputting.
   • tuple of (height, width) denoting the dimension of the cropped image. E.g., 420x690, 727x135.
   • float denoting the ratio between the cropped and the original image. E.g., 0.5 means 50% of the original image.
7. Circle cropping - Two parameters.
   1. <radius> - The domains of this value produce different outcomes.
      • int (rad > 1) denoting the radius of the cropped image.
      • <float> (0 < rad <= 1) denoting ratio between the cropped and the original image.
      • Leave <blank> or 0 to use the smallest distance between the center and image walls.
   2. <center>, tuple (y, x) denoting the coordinate of the center of the cropped image.
      • Leave <blank> to use the center of the original image.
8. Cross Ellipses cropping - Two parameters.
   • To get the best result, everything should be left <blank>.
   1. <radius> - The datatypes of this value produces different outcomes.
      • tuple (rad_y, rad_x) denoting the radii of the cropped image.
      • int denoting the radius of the cropped image.
      • Leave <blank> or 0 to use the optimal dimension.
   2. <center> - tuple (y, x) representing the coordinate of the center of the cropped image.
      • Leave <blank> to use the center of the original image.

Built With

1.80.1	3.11.4	VSCode	11

(back to top)