| |
 |
|
|
Science Forum Index » Image Processing Forum » about object recognition
Page 1 of 1
|
| Author |
Message |
| Zeng Fucen |
 Posted: Thu Apr 24, 2008 9:47 pm |
|
|
|
Guest
|
hi, everyone:
my research is about a object recognition.This object is a big
character target. I want to calculate the exact position in the image.
I cannot use the template match method , because there are so many
different morphing shapes .First, I segment the processing image into
several regions by using some method. After that, I use the invariant
moment to determine which region is my target region. But the
invariant moment only fit to the non-perspective image and it's also
imprecision.
I wonder if any of you know is there any other good method or hint to
solve such a problem? thanks first !!! |
|
|
| Back to top |
|
| lomas |
| Posted: Tue Apr 29, 2008 5:20 pm |
|
|
|
Guest
|
On Apr 25, 3:47 pm, Zeng Fucen <zengfu...@gmail.com> wrote:
Quote: hi, everyone:
my research is about a object recognition.This object is a big
character target. I want to calculate the exact position in the image.
I cannot use the template match method , because there are so many
different morphing shapes .First, I segment the processing image into
several regions by using some method. After that, I use the invariant
moment to determine which region is my target region. But the
invariant moment only fit to the non-perspective image and it's also
imprecision.
I wonder if any of you know is there any other good method or hint to
solve such a problem? thanks first !!!
How perspective your images are? If it's too much, this will be a
challenge work. A car may be just a point in some perspective image.
So I think you should find what indeed you are handling.
a most kind of object recognition methods are based on training or
statistics, e.g. ANN SVM adaBoost.
Best Regards,
lomas |
|
|
| Back to top |
|
| |
|
Page 1 of 1
All times are GMT - 5 Hours
The time now is Thu Jul 24, 2008 10:33 pm
|
|
