Just like grayscale images, color images are also affected by noise.
However, color images have three channels (Red, Green, and Blue), so noise can affect each channel differently.
1. Causes of Noise in Color Images:
- Usually, noise affects all three channels (R, G, B) equally.
- But sometimes only one or two channels may have more noise, due to:
- Camera malfunction (e.g., red channel’s sensor problem)
- Uneven lighting or filters, e.g., using a red filter reduces light reaching the red sensor → causes more noise in red channel
📌 CCD sensors (used in cameras) are more noisy at low light levels.
2. Noise Filtering in Color Images:
There are two ways to reduce noise:
✅ A. Per-Channel Filtering (Simple Method):
- Process each R, G, and B channel independently, like grayscale images.
- Example: Apply a mean or median filter on each channel separately.
✅ B. Vector-Based Filtering (Advanced Method):
- Treat each pixel as a 3D RGB vector.
- Apply filters in vector space (e.g., vector averaging or vector median).
📌 Example:
- Averaging filter gives same result in both methods (per-channel or vector).
- But median filter in vector space is complex, because we need to sort RGB vectors, not just numbers.
3. Challenge: Vector Ordering for Median Filter
- For scalar values (like grayscale), median is simple: just sort and pick the middle.
- For RGB vectors, we must define a rule to compare and order vectors.
- This process is called vector ordering, and it is difficult and complex.