For example:
a1 a2 a3 a4 a5 a6 3x3 filter window a7 a8 a9You enter a1 - a9. The filter is placed over the image in such a way that the pixel being processed is covered by a5. The neighborhood pixels are then multiplied by the corresponding values in the filter, and then summed. The pixel at a5 is replaced by this sum.
There are four buttons below the table:
The Import button reads a kernel from a file you specify. The number of entries in the first line of the file determines the number of columns in the kernel. All lines of data following that should have the same number of entries. If there are more, they are ignored. If there are less, then zeros are used to make up the difference. Data entries should be separated by whitespace (i.e. spaces, tabs, or newlines).
The Export button writes a kernel to a file you specify, in the format specified above.
The Normalize button sums the entries in the kernel and then divides all entries by this sum. Usually this is desirable, since all the other filtering algorithms normalize their respective kernels before performing the filtering operation.
The Reset button resets all the entries in the kernel to zero.
Note: In order to filter pixels located near the edges of the image, edge pixel values are replicated to give sufficient data.