has advanced line tools. They use trace combs to find edges and contours.
Trace combs provide a very selective edge tracing. You decide yourself the
way and direction to search. The tool is thus tuned to find only the edges
of interest. The disadvantage is that you need knowledge of orientation and
image position of the measured object. A trace comb with a number of
parallel trace lines as shown below, is the most common one.
specify the size of the comb by its length, position and base line
angle. The base line is the “spine” of the comb making a right angle
with the trace lines.
you specify the start and stop position of the trace lines, e.g. start
tracing 3mm from the base line and stop at 12mm from the baseline. The sign
of the start and stop values decides on which side of the base line you are
tracing. A right hand rule is valid: Hold your right hand with the palm
towards you and the thumb in the direction of the base line, then your
fingers point in the positive direction. Observe that this direction is not
the same as the trace direction. The trace direction is positive if the
start value is less than the stop value, else it follows the negative
combs can either be rectangular as shown below, circular or radial.
lines, intensity profiles and edge detection
a single trace line, the image intensity (gray tone) is read step by step,
thus you get an intensity profile that you can present as a curve:
this case we see to distinctive jumps where the intensity changes from light
to dark and from dark to light again. We want to find the position of these
two jumps on the curve. An obvious method is to find the position where the
intensity curve crosses a threshold. In the example we then get the
the background intensity often varies, it is difficult to find a threshold
value for the various light conditions. The point to the left in the above
picture is may be not of any interest. An alternative is to search for jumps
in the intensity independent of the general intensity level. Differentiation
of the curve can do this, which is calculating the gradient of
the curve in all steps. Then you get a new curve:
jumps have turned to tops and bottoms clearly differing from the rest of the
curve, which however is close to zero. Finding the top and bottom, the curve
extreme values will now localize the edges. We also here use a
threshold value to avoid that noise is perceived as extreme values. Then
ripples on the curve are filtered away. Only extreme values with an absolute
value above the threshold are included.
conclude, we have two main principles to find edges along a trace line,
either as threshold crossings or as extreme values.