SOI Image Processing 3

Filtered Image

To perform the filtering operation, the value of each pixel of the filter must be multiplied by the value of each pixel of the transform of the original image, and then the inverse Fourier transform performed on the product. Note that the filter must have complex values, so that the Gaussian surface must be duplicated in both the real and the imaginary parts of the filter.

Here is the image after:

It has been Fourier transformed.
Its transform has been multiplied by the filter.
The result of the filter has been inverse Fourier transformed.
The reconstructed image has been contrast (scale) expanded.

Image Mask

To make a movie of all images, it was desired to eliminate the variation in all pixel values which were not part of the solar disk. To do this, a simple circular mask was created around the image:

Filtered and Masked Image

This mask was then applied to the filtered image. All pixels which fall inside the mask retain their values, while all those which fall outside of the mask are set to the average value of a 25x25 pixel area in the upper left corner of the image.

Filtration in this way effectively removes the low-frequency noise, while preserving the data content of the image.

For more information

For more info about the SOHO spacecraft, see http://sohowww.nascom.nasa.gov/.

For more info about the SOI project, see http://soi.stanford.edu/.

To see the software used in this analysis, see http://www.intergalact.com/g1k/g1k.html.

Go to:

Page 1: Raw image and FFT.
Page 2: FFT Filter.
Page 3: Filtered and masked image.