Please start any new threads on our new site at https://forums.sqlteam.com. We've got lots of great SQL Server experts to answer whatever question you can come up with.

 All Forums
 Site Related Forums
 The Yak Corral
 My current fuss

Author  Topic 

Stoad again
Posting Yak Master

240 Posts

Posted - 2009-03-17 : 12:14:39
My 12 samples What do you think on this?

Cheers!

tosscrosby
Aged Yak Warrior

676 Posts

Posted - 2009-03-17 : 12:41:10
Not much....

Terry

-- Procrastinate now!
Go to Top of Page

spirit1
Cybernetic Yak Master

11752 Posts

Posted - 2009-03-17 : 12:48:17
hey stoad!
that's not bad if it's done by a computer of course
what are you using?

___________________________________________________________________________
Causing trouble since 1980
Blog: http://weblogs.sqlteam.com/mladenp
Speed up SSMS development: www.ssmstoolspack.com <- version 1.5 out!
Go to Top of Page

mfemenel
Professor Frink

1421 Posts

Posted - 2009-03-17 : 13:20:06
Now if you could get this to fill in a text box for me when I try to buy concert tickets you'd make a fortune.

Mike
"oh, that monkey is going to pay"
Go to Top of Page

TG
Master Smack Fu Yak Hacker

6065 Posts

Posted - 2009-03-17 : 14:09:18
stoad!
I love the fact that you used an image from "bad santa"

Be One with the Optimizer
TG
Go to Top of Page

pootle_flump

1064 Posts

Posted - 2009-03-17 : 14:51:15
quote:
Originally posted by TG

stoad!
I love the fact that you used an image from "bad santa"

heh - I recognised the kid but not the scene!
Go to Top of Page

Stoad again
Posting Yak Master

240 Posts

Posted - 2009-03-17 : 15:41:32
Hi guys!

mladen, of course it's the computer:) it's my python script
(nothing python specific; the code can be easily rewritten in C++).
Reading an image file -> Converting it to the greyscale mode -> Populating a 2D array of bytes.
Then analyzing the matrix, pixel by pixel, and so on.
All is based (or rather inspired by) on the idea of so called Stroke Filter Response,
but I do it in much my own way.
Note: the code d-o-e-s n-o-t r-e-c-o-g-n-i-z-e texts.
I don't even know what those Chinese words on my pics mean (seems the very first one means "http").
As an intro to the subject a quote from an article of engineers from Samsung Advanced Institute of Technology:
quote:

... For example, from the Canny edge image shown
in Fig.1 (a), it is so difficult to find the text even for human
eyes. Besides the high complexity of text localization, we
think that the fundamental reason of these problems is that
no researcher finds the distinctive features of text. That is to
say, no one answers the basic question: what on earth is
text? Intensity-similarity CCA, edge, corner or texture
features are only the necessary conditions of text, as shown
in Fig. 1(b). Although recently someone uses Adaboost to ...



Teeg, flumpie, I like you like it :)
Go to Top of Page

spirit1
Cybernetic Yak Master

11752 Posts

Posted - 2009-03-17 : 16:54:00
yeah canny filter can be quite helpful. try bluring an image before you apply it. i found it gave better results.

___________________________________________________________________________
Causing trouble since 1980
Blog: http://weblogs.sqlteam.com/mladenp
Speed up SSMS development: www.ssmstoolspack.com <- version 1.5 out!
Go to Top of Page

Stoad again
Posting Yak Master

240 Posts

Posted - 2009-03-17 : 17:14:42
blurring... I had tried it, with unstable results

Go to Top of Page

Stoad again
Posting Yak Master

240 Posts

Posted - 2009-03-17 : 17:32:09
What I need is a good test set of images.
They say there is e.g. Microsoft test set of 46 images.
Mladen, if you were (are) in the subject, where can I see your results?
I can't find anywhere someone's else after-detecting-pics, to see and compare
PS
I don't know what Canny filter is :)
Go to Top of Page

spirit1
Cybernetic Yak Master

11752 Posts

Posted - 2009-03-17 : 17:34:33
no, i was never in the text finding field.
Canny filter is an edge detection method
http://www.pages.drexel.edu/~weg22/can_tut.html

i've used it with great success in the past.


___________________________________________________________________________
Causing trouble since 1980
Blog: http://weblogs.sqlteam.com/mladenp
Speed up SSMS development: www.ssmstoolspack.com <- version 1.5 out!
Go to Top of Page

spirit1
Cybernetic Yak Master

11752 Posts

Posted - 2009-03-17 : 17:45:34
i'm guessing you're using this?
http://svn.assembla.com/svn/MartinRais/Papers/2008%20-%20A%20new%20approeach%20for%20text%20segmentation%20using%20a%20stroke%20filter.pdf

___________________________________________________________________________
Causing trouble since 1980
Blog: http://weblogs.sqlteam.com/mladenp
Speed up SSMS development: www.ssmstoolspack.com <- version 1.5 out!
Go to Top of Page

Stoad again
Posting Yak Master

240 Posts

Posted - 2009-03-17 : 18:36:42
No! I've never seen this article (now I'm downloading it).
I have 3 or 4 other pdf-s thru which I just browsed and catched the main/basic idea/KEYWORDS.
Then I applied my own grey matter :) Forgot: at first I tried to implement a chinese algorithm,
based on stroke filter, without much success (due to my poor understanding of its (algo) steps)


Go to Top of Page

Stoad again
Posting Yak Master

240 Posts

Posted - 2009-03-17 : 19:32:16
Yes, it's those korean folk that I mentioned but another their article.
Their samples: http://img-fotki.yandex.ru/get/3110/funkybee.0/0_6f02_ad30bf5f_orig
See the 3rd column and false text captures on its 1st and the 4th frames.
Go to Top of Page
   

- Advertisement -