Thank you for your assistance!
Regarding your first solution because it is not so complicated, let me explain what I understood, please:
1. Take the image and convert it in pixmap. The value of each pixel is presented in a hexadecimal value (3 bytes).
2. Take one of the bytes (green color). They are with equal values because of the grayscale image.
3. Take the sine wave as a filter because it is a symmetric function. The square of the input tapers more nicely.
4. Each element of the square of the input sine wave array is divided by the sum of all elements of the square of the input sine wave. Really the sum of all elements of the kernel is exactly 1. I tried to make it 2 and the image containing the background became white and the filtered image was completely black.
5. Make convolution using the byte (green color) from line 2 and result array from line 4.
6. Make subtraction between the byte (green color) (line 2) and the result from line 5. Subtraction is made dividing the byte (green color) from the background image.
7. Complete a second convolution deleting small spikes (1 - 2 pixels) from the image. The filter is an array with elements equal to 1.
8. The background image is created without using the second convolution.
As a conclusion, you create a background image and from the initial image you subtract the background image, right?
