To see how this works, consider multiplying the original image by 3 and the unsharp mask by 2. When the 2x unsharp mask is subtracted from the 3x original, the small detail (contained in the high frequency parts of the original) is emphasized by a factor of 3. If we think of the original image consisting of (High + Low) and the unsharp mask consisting of only Low, we have, mathematically (3*High + 3*Low) - 2*Low = 3*High + Low.
| Original (A) | Unsharp Mask (B) | Merged (3*A - 2*B) | |
|---|---|---|---|
|
|
![]()
|
The Unsharp Mask Filter creates the unsharp mask. It works on the image in the A buffer and leaves the result in the B buffer. You are given choices as to how big to make the unsharp mask in both the row and column directions. The size of the filter is given by (2*rows + 1) x (2*cols + 1). The example above used a 3x3 filter.