FFT-filtering

This is an example of the use of the FFT-filter. On the left processing without the FFT-filter, on the right with a 20 pixel FFT-filter. These graphs are presented when the option show FFT and show Graph are checked on the Align Frames page.

During alignment the reference image and the frame that will be aligned are combined and the resulting information is shown as and FFT-spectrum image. These images show colours and the red area of the colour-range represents the highest values. The middle of the image stands for 0 displacement in X or Y and when the red area is exactly in the middle the displacment between the images is 0,0. When the red zone moves away from the middle displacement is clear (see displaced). The two FFTs show nearly the same as they both show a red area in the middle. But the image on the left shows a clear red-cross with a chance for spurios peaks along the axis of the FFT. The image on the right (where the filter was used) shows a nearly round zone of red values. This makes this estimate of displacement more reliable. The size of the FFT-filter is depending on the images, the graph below the FFT-spectrum shows the position of the FFT-filter (blue line), it is wise to put the filter to the right of the bend in the graph and to aim for a small but round red-area in the FFT-spectrum image.

The two images below show the effects of filtering even clearer, again on the left no filter and on the right with FFT-filtering. The left image again shows signs of very high levels on the Y-axis and there is a great chance that this will lead to misalignment. The image on the righ shows a clear zone that is probably best for alignment, estimating the middle/peal of this red area is not too difficult.