COMP343 Fundamentals of Multimedia Computing

Image Processing Project
Image Foreground/ Background Separation and Processing

Examples of the Complete Process

For the following examples the input face image was:
The input background image was:

During processing, the original background is removed, and the input background image is automatically scaled to fit in to the new size and aspect ratio.
1. Foreground is 'aged' to look like an old photo
 2. Foreground is converted to gray scale
 3. Foreground is blurred
 4. Foreground is posterized
5. Foreground is sharpened
6. Foreground has East-West slopes emphasised 

Good Foreground/ Background Separation

Original Image
Image After Foreground/ 
Background Separation

Some minor problems near the coat, in part a 
consequence of compression artifacts

Problematic Background/ Foreground Separation

Original Image
Image After Foreground/ 
Background Separation
 Comments
This image caused great problems because of the low number of colours used, the lack of a defined edge between foreground and background, the varying background colours, and the poor contrast between foreground and background
Good separation, apart from the lost shoulder, in part a consequence of low contrast.
The background image is the map of the world shown above. The third image shows colour change as a result of too few colours in the GIF lookup table.
This image caused problems for some students typically because of the left/ right approach they took when separating foreground from background. A top-down approach would have worked in this case because of the gray gradient in the background.
Copyright David Rossiter 2000