Conclusion
Scaling is necessary.
Image and vision systems often require scale transformation of images to different sizes to meet various display requirements. For this project we are particularly interested in up-scaling of images.
Different ways of Scaling.
Simple Scaling:
Introduction:
There are various interpolation methods, generally trading-off complexity with performance.
Simplest interpolation - Nearest-neighbor algorithm - (Nearest pixel value is replicated).
Bilinear interpolation - More Complex - The blank pixels values are populated by a linear function
between the closest existing pixels. More complex methods are bi-quadratic and bi-cubic interpolation.
Problems with Simple Scaling:
Nearest-neighbor algorithm - Results in a blocky-image artifact.
Bilinear Interpolation, Bicubic, Biquadratic - Results in Blurring.
Scaling after Edge Detection:
Introduction:
To preserve edges when scaling, we need to first detect where edges occur, then enhance them before scaling up. Edge detection is achieved with a high-pass filter which are all directional, meaning they detect edges in specific directions. Operators that are able to detect all edges regardless of direction are called "isotropic". This is achieved by using filters that operate on both image axes simultaneously. The most common isotropic edge detection filter is the laplacian mask which is what we have chosen for this project
Tradeoffs Associated with Edge Detection.
Smoothing operation is linear across the entire image, hence the edges that need to be preserved are also washed out by the Gaussian filter. So there is a trade off between edge-preservation and noise reduction and this limits the effectiveness of this method while scaling images.
Solution offered by Bilateral Filtering.
We saw that it was desirable to be able to preserve edges while washing out noise. This can be achieved by a non-linear filter that distinguishes between noise (filter out) and edges (enhance). We explore the bilateral filter proposed in [1] as a solution.
Our Results.
Comparing the performance with vs. without bilateral filtering, the improvement both in noise reduction and edge-preservation is apparent. We were able to effectively enhance the edges without amplifying noise.
Future Scope of this Project.
The future for this kind of system is the ability to compress and store images and then expand them when needed or at the receiving end with almost zero delay. There are lots of areas where this kind of system can be used. We have identified remote management as one of the most potential applications are eager to see how far this can go.