This preview shows page 1. Sign up to view the full content.
Unformatted text preview: through Edge 4 compute a similar type of one dimensional edge property.
The operators as shown above also incorporate some image smoothing, since they accumulate
two measurements of edge contrast and then add them together. This type of operator will
diminish the effect of an errant intensity point.
The Sobel operator incorporates edge information in two directions as follows: Edge = (<A+2B + C - G – 2H – I >2 + < A + 2D + G – C – 2F – I>2) ½
This operator computes a weighted average intensity function along the borders of the subimage, and then forms two edge measurements perpendicular to each other. The two edge
properties are then combined in a “quadrature” measurement. The Sobel operator generally
enhances edges in an acceptable fashion. The squaring and square root operations will be very
time consuming, so one normally would modify this operator by replacing the squares with
absolute values and eliminating the square root. This is referred to as the modified Sobel
Contrast Operator Another approach to edge detection is the use of contrast differences. Consider the following:
Edge = E – (A + B + C + D + F + G + H + I )/ 8
This operator compares the intensity of the central pixel to the pixel’s surrounding. Here, only t
he difference in contrast between the central pixel and its neighbors is considered, regardless of
the distribution of the neighboring pixel intensities.
Edge-based segmentation represents a large group of methods based on information about edges
in the image, it is one of the earliest segmentation approaches. Edge-based segmentations rely
on edges found in an image by edge detecting operators -- these edges mark image locations of
discontinuities in gray level, color, texture, etc. Image resulting from edge detection cannot be
used as a segmentation result. Supplementary processing steps must follow to combine edges
into edge chains that correspond better with borders in the image. The final aim is to reach at
least a partial segmentation -- that is, to group local edges into an image where only edge chains
with a correspondence to existing objects or image...
View Full Document
- Spring '14