This project deals with a different kind of interpolation used to resize digital images.
Known and widely used interpolations (Nearest neighbour,Bilinear) tend to "destroy" edges in the image.
Abstract
This project deals with a different kind of interpolation used to resize digital images.
Known and widely used interpolations (Nearest neighbour,Bilinear) tend to “destroy” edges in the image.
Directional interpolation works locally on a group of pixels and uses only pixels on one side of the edge in order to interpolate the new sample.
The algorithm
The figure below describes the full algorithm:
the 4 main functions that :
- Choose an area (2*2 or 4*4 pixels)
- Divide the area into 2 or 3 groups by their value using Max Lloyd quantization
- Decide to which group the new pixel belongs
- Interpolate using only the selected group’s pixels
Results
The result images preserve edges better than bilinear interpolation.
Acknowledgments
We would like to thank our supervisor Ari Shenhar and Mr. Johanan Erez for their support and guidance throughout this project.