Results with EdgeDetector 1.0
After Smoothing the Original Image
The original head image was smoothed with
Gaussian filters with three different sigmas (SG). When
EdgeDetector 1.0 was run, the value of
parameter ES was fixed
at 40.0. Images in each section correspond to the LL value of
100.0, 90.0, 80.0 and 70.0 in the order of from left to righ in
each row, and from top to bottom row.
Due to the foot-print of the Gaussian filter, the pixels on the
four sides of the image were set to a fix value of zero after
smoothing. This led to the false horizontal edges detected close
to the bottom of the result images shown below.
Though the edges detected from the smoothed image look more
connected, the location of these edges can not be determined as accurately as when no smoothing was applied to the original image.
For projects such as
Surface Reconstruction, if this type of smoothing is applied, an extra step
should be taken to locate the edges in the original image.
The smoothed images: SG = 1, 2 and 3
SG=1; LL=100.0, 90.0, 80.0 and 70.0; ES=40.0
SG=2; LL=100.0, 90.0, 80.0 and 70.0; ES=40.0
SG=3; LL=100.0, 90.0, 80.0 and 70.0; ES=40.0
Edge Detection Home Page
Results from the head image || Results from the Lena image
EdgeDetector || Download Synthetic Test Images