JoaoESmoreira

Image Compression Project

The purpose of this project is to explore and understand the fundamental concepts of Image Compression, with a particular focus on the JPEG codec. The work consists of implementing and analyzing the main mechanisms used in image compression using Python.

The full repository and documentation can be found here: Image Compression Project Repository.

Tools and Technologies

Programming Language: Python
Libraries: NumPy, SciPy, Matplotlib, Pillow (PIL)
Image Formats: BMP, JPEG

Main Tasks

Basic JPEG Compression Compress BMP images into JPEG format with different quality levels (high, medium, low) using an image editor. Compare and analyze the results.
Encoder and Decoder Functions Create encoder and decoder functions to encapsulate all the developed sub-functions.
RGB Visualization Implement functions to:
- Read and visualize BMP images.
- Apply a user-defined colormap.
- Separate and reconstruct RGB channels.
Pre-processing (Padding) Implement image padding to ensure dimensions are multiples of 32×32, replicating the last row and column as needed.
Color Model Conversion Implement conversion between RGB and YCbCr color models and verify correctness by comparing pixel values and channels.
Sub-sampling Implement sub-sampling and up-sampling functions (e.g., 4:2:2 and 4:2:0 schemes). Analyze the compression ratio and visual quality differences between methods.
Discrete Cosine Transform (DCT) Apply the DCT to image channels:
- On the full channel.
- On 8×8 and 64×64 blocks.
Visualize and discuss compression potential and differences between block sizes.
Quantization Implement quantization and inverse quantization of DCT coefficients. Test different quality factors (10, 25, 50, 75, 100) and analyze compression versus distortion.
DPCM Coding Implement Differential Pulse Code Modulation (DPCM) for DC coefficients and analyze the compression effects.
End-to-End Encoding and Decoding Build a full JPEG-like pipeline integrating all components. Decode images for various quality factors and compute metrics such as:
- MSE (Mean Squared Error)
- RMSE (Root Mean Squared Error)
- SNR (Signal-to-Noise Ratio)
- PSNR (Peak Signal-to-Noise Ratio)
Compare the decoded results with the original image and analyze trade-offs.

Footer

The contents of this website are licensed under the Creative Commons Attribution-NoDerivatives 4.0 International License (CC-BY-ND 4.0).

The source code of this website is licensed under the MIT license, and available in GitHub repositor. User-submitted contributions to the site are welcome, as long as the contributor agrees to license their submission with the CC-BY-ND 4.0 license.