Hi,

This is a short summary about my GSoC progress. The original plan is listed as below:

1. The conversion between spatial domain and frequency domain. 2. Implementation of filters.
3. Filtering processing.
4. Miscellaneous services.
5. Applications.

Part 1 looks like very easy to implement, but in actual it is the most challenging part. Pippin and me discussed for many times about how to storage a frequency "image" in a GeglBuffer. The data structure of the buffer was changed three times. The ops(at svn version r2358) which used 4-component buffer had been already usable, but after discussion we decide to migrate the code to a new babl_format, which has 8 components. Now the migration has completed.

Part 2 and Part 3 based on the same codes. Filtering processing is an complex-number matrix(the frequency "image") multiplying another complex-number matrix, and a filter is the matrix. The code for Part 3 has been written and in debugging. The Part 2 haven't start yet. In Part2, every op's job is calculating to generate the matrix with the parameters like bandwidth and cut-off frequency, and then use.Part3's codes to process the image.

Part 4 and Part 5 haven't start, they are not the most important parts after all. Once Part 1,2,3 are stable, Part 4 and Part 5 can be easily completed, and Part 5 maybe need to use definition language like XML.

What to do after mid-term: 1. test and fix bugs in exist codes. 2. implement kinds of filters(matrices) like Gaussian-lowpass etc. 3. some miscellaneous functions such as a homomorphic filter. (optional) 4. documents for this frequency processing kit.

Regards, Zhang