Ah, thank you Achille, I can see the
new images you posted in my browser. Thanks for the tip RavensFan, I didn't
think to do that.
As to your question:
As I understand your application, the final result from your Solution1 &
Solution2 processing will NOT be the same. Here is why:
Solution2: only the incoming 15 lines have the binary morphology operations
applied to them (dilation, erosion). As these 15 lines "move up" in
your image, the binary morphology operations you apply to the subsequent groups
of incoming lines will not affect the rest of the image.
Solution1: the whole image will have the binary morphology operations applied
to it each time a 15 line group is added to the bottom and the oldest group
deleted from the top. If you process in this manner, by the time the 1st 15
line group reaches the top of your image, it will have had the binary
morphology operations performed on it 10 successive times. In other words, your
image would effectively be something like the following. Let's assume that you
just perform an erosion for the sake of simplicity:
_________________________________________________
DELETE group
10th group: has had 10 erosions
9th group: has had 9 erosions
8th group: has had 8 erosions
.
.
.
3rd group: has had 3 erosions
2nd group: has had 2 erosions
1st group: has had 1 erosion
ADD group
__________________________________________________
If you do this it will distort your particles. I think you should go with
solution 2, as this will apply the same processing to each group and then
assemble them into your image where you can then detect particles and perform
other analysis. This still leaves you with the problem, however, of the case
that you mentioned where an object spans 2 or more 15 line groups. I've thought
a little bit about this, but haven't come up with a good answer yet. One line
of thinking was that you could determine the largest size (in terms of lines of
your image) that you will expect for an object in your image. If the largest is
say, 34 lines tall, then it would be completely represented in the image within
3 groups of 15. What if you waited to perform the binary morphology
operations until these 3 groups reach positions 4, 5, & 6 and then perform
the operation on only these three groups? This wouldn't work too well, however,
because the same groups that contain the end of that object could contain the
beginning of another object. I think there are ways around this problem;
I just haven't come up with a good one yet.
I hope the above comments help. Thanks for being patient in re-posting your
example. Post back and we can continue the discussion. Perhaps someone else
will have an insight on this problem. Ravens Fan?
~Nate
