I am trying to implement the Double Gaussian filter, which is defined in ISO surface profiling standard. I have found the algorithm using FFT. However I'd like to use a convolution method. I have already done the Gaussian filter.
The algorithm using FFT is just a fast way of implementing the convolution. Why do you want to run it slower?
Thank you, Fernando.
The reasons why I prefer to the convolution method are:
1) I have used it for a long time.
2) Any sample size possible, and a window unnecessary.
3) Faster than FFT in case of small kernels.
4) Easy to implement in GPU.
Hello,
Can you please tell me how you applied the double Gaussian filter or better tell me from where i can learn applying this?
Thank you for your question. I have derived the convolution algorithm, but not yet published.
Hints: The Double Gaussian filter is equal to apply the single Gaussian filter two times.
- Badges
- Science topic
- Similar topics
- Papers