Overview

This project aims to align images taken from different channels (e.g., blue, green, red) to create a composite RGB image. The alignment process corrects for any misalignment between the channels caused by various factors such as camera movement or lens distortion.

Approach

My approach involves using a combination of image processing techniques, including pyramid image alignment and cross-correlation, to find the optimal alignment parameters for each channel. Here's a brief overview of our approach:

1. Reading and Preprocessing: Read images and convert them to floating-point format.
2. Shrinking: If images are large, shrink them to improve processing speed.
3. Cropping: Crop each channel to remove any unwanted borders.
4. Alignment: Use pyramid image alignment to align each channel with the reference channel (usually green).
5. Offset Calculation: Calculate the offset values (shifts) for each channel.
6. Composite Image: Combine the aligned channels to create the final RGB image.
7. Save Results: Save the aligned image and offset values to files for further analysis.

Challenges

During the project, I encountered several challenges, such as:

• Large Image Sizes: Processing large images efficiently required implementing techniques like image shrinking.
• Alignment Accuracy: Achieving precise alignment was challenging, especially with images containing complex features.
• Edge Cases: Handling edge cases where images have minimal features or are heavily distorted required careful consideration.

Results

Blue shift: -27, -10

Red shift: 33, 3

Blue shift: -20, -6

Red shift: 22, -1

Blue shift: -13, -6

RRed shift: 16, 2

Blue shift: -17, -3

Red shift: 21, 1

Blue shift: -17, -5

Red shift: 20, -1

Blue shift: -14, -2

Red shift: 14, 9

Blue shift: -18, -7

Red shift: 19, 2

Blue shift: -21, -4

Red shift: 24, 3

Blue shift: -15, -8

Red shift: 18, 6

Example Images

Blue shift: -8, -4

Red shift: 12, 2

Blue shift: -14, -1

Red shift: 14, -1

---

---