This device is not compatible.

PROJECT

# Image Compression Using DCT with Graceful Quality Degradation

In this project, we’ll learn to perform lossy compression of a grayscale image using the Discrete Cosine Transform (DCT) with a graceful compromise on visual quality.

You will learn to:

Implement an image compression algorithm from scratch.

Handle files for reading and writing image data.

Analyze the effects of lossy compression on image quality.

Visualize image data using Python libraries.

Skills

Image Compression

Image Manipulation

Image Visualisation

Prerequisites

Understanding data types in Python

Hands-on experience with Python

Technologies

NumPy

Python

Matplotlib

Project Description

There are two kinds of image compression: lossless and lossy. In both kinds, the objective is to reduce the required space to represent the image. In lossless image compression, there is no loss of information, so the image quality is not altered. However, in lossy image compression methods, some of the information is lost, and the quality of the image is affected. The challenge with lossy image compression is to select the most irrelevant information for our visual system and compromise it to gain compression. The lost information affects the image quality, but the visual quality does not suffer much. This is called graceful degradation of the image quality.

The image compression algorithm implemented in this project is lossy. We’ll take a block of size $8\times8$ from a grayscale image and transform it into the frequency domain to select the information we want to discard or compromise. Every possible block is processed in the same way—in a serial manner. In this project, we’ll also see the extent to which the visual quality degrades as we compromise more and more information to achieve a higher compression ratio.

The task of producing compressed image data from a given image is achieved by an encoder. The decoder is responsible for reconstructing the image from the compressed data. The efficacy and compromise on quality is measured by comparing the original and reconstructed images.

The following diagram illustrates how the encoder and decoder interact:

Project Tasks

1

Getting Started

Task 1: Import Libraries

Task 2: Import the Image

Task 3: Pad the Image

2

Create the Image Encoder

Task 4: Implement the DCT Function

Task 5: Visualize the Image Blocks

Task 6: Apply DCT to the Image Blocks

Task 7: Quantize the Image Blocks

Task 8: Scan and Encode the Image Data

Task 9: Store Image Data in a File

3

Create the Image Decoder

Task 10: Read the Image Data from the File

Task 11: Decode the Image Data

Task 12: Implement the Inverse DCT Function

Task 13: Apply the Inverse DCT on the Image Blocks

Task 14: Obtain and Display the Final Image

4

Modulerization and Quality Comparison

Task 15: Modularize the Encoding Algorithm

Task 16: Modularize the Decoding Algorithm

Task 17: Test the Encoding and Decoding Modules

Task 18: Compute the Compression Ratio of the Compressed Images

Task 19: Plot the Compressed Images

Congratulations!