Hi there XnView
I had just installed and tested XnView 1.19.5, release 11 Jun 2010, on a Windows 7 box,
and I am impressed by the quality of the results of three different implementations
of the Edge_detect filter algorithm.
I understand that the source code is not disclosed but, can you share the ideas behind the three different implementation of this filter?
Thanks in advance and congratulations for a good job.
César
The three Edge_detect implementations
Moderators: XnTriq, helmut, xnview
Re: The three Edge_detect implementations
It's a filter, i use:
/* Edge Detect Light */
0, 1, 0,
1, -4, 1,
0, 1, 0,
/* Edge Detect Medium */
-1, -1, -1,
-1, 8, -1,
-1, -1, -1,
/* Edge Detect Heavy */
1, -2, 1,
-2, 4, -2,
1, -2, 1,
/* Edge Detect Light */
0, 1, 0,
1, -4, 1,
0, 1, 0,
/* Edge Detect Medium */
-1, -1, -1,
-1, 8, -1,
-1, -1, -1,
/* Edge Detect Heavy */
1, -2, 1,
-2, 4, -2,
1, -2, 1,
Pierre.
Re: The three Edge_detect implementations
Thanks for posting the convolution kernels for the edge filters. There must be a normalization step in addition to make the negative values produced by this kernel in the correct pixel range. I've tried to see if a simple linear transform using the max and min pixel to 255 and 0 in the grey scale, but this doesn't produce the correct normalization.
Would it be possible to provide more information? Thanks.
Would it be possible to provide more information? Thanks.