no1nincompoop
asked on
Any idea in imaging...
Can any one tell me how does the imaging software works..
i.e. For a tif file which contains a barcode in it...
How the softwares recognize the barcode in it, and
How the softwares reads a particular area of the image and gives the text is that area..
i.e. For a tif file which contains a barcode in it...
How the softwares recognize the barcode in it, and
How the softwares reads a particular area of the image and gives the text is that area..
ASKER
thanks lunaskye...
i got some idea... but i still need some more information..
i just want to know what the OCR does...that is my requirement...
can any one provide me some solution or idea that... how to read text... how to read the pixel configuration....
i got some idea... but i still need some more information..
i just want to know what the OCR does...that is my requirement...
can any one provide me some solution or idea that... how to read text... how to read the pixel configuration....
Hmm.. thats a pretty in-depth question..
From what i hear its pretty complex. Their algorythms for identifiying text from blobs of pixels are most likely proprietary. I dont expect that they woudl ever share them willingly with anyone.
From my general understanding of it, though, i believe they try to form vectors based on the pixel configuration. General Meaning: Find and Isolate a certain blob of connected pixels, (i.e. a letter), then find extremes of that blob. From the extremes, create lines, a certain configuration of lines would be recognized as a letter.. cross reference a line configuration against a database of all available characters, and spit out the ascii code. There would have to be some tolerance involved of course, if osmething is CLOSE to the character, maybe they would taek the closest match.
Another idea is maybe they have on file a saved bitmap of characters, if by overlaying the "blob" from the scan over the recognized character, and using boolean logic, you could see how closely your scan resembles a saved character picture, THe one that is the closest match (the most pixels line up), that is the scanned character..
(Not an expert on this stuff)
-Andrew
From what i hear its pretty complex. Their algorythms for identifiying text from blobs of pixels are most likely proprietary. I dont expect that they woudl ever share them willingly with anyone.
From my general understanding of it, though, i believe they try to form vectors based on the pixel configuration. General Meaning: Find and Isolate a certain blob of connected pixels, (i.e. a letter), then find extremes of that blob. From the extremes, create lines, a certain configuration of lines would be recognized as a letter.. cross reference a line configuration against a database of all available characters, and spit out the ascii code. There would have to be some tolerance involved of course, if osmething is CLOSE to the character, maybe they would taek the closest match.
Another idea is maybe they have on file a saved bitmap of characters, if by overlaying the "blob" from the scan over the recognized character, and using boolean logic, you could see how closely your scan resembles a saved character picture, THe one that is the closest match (the most pixels line up), that is the scanned character..
(Not an expert on this stuff)
-Andrew
ASKER
hello moderators...
I haven't got the required answer.. so I cannot give points to any one...
Wht can be done...
Regards
no1nincompoop
I haven't got the required answer.. so I cannot give points to any one...
Wht can be done...
Regards
no1nincompoop
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
As far as the bar code goes, it is just a series of vertical lines. In a graphics file all pixels are saved in an array of colors (basically). By simply checking a certain area of that array, you can read to see where exactly the vertical lines of the bar code are. These lines can then be interpreted to mean whatever they should to the bar code reader.
As far as text.. I dont believe a tiff file saves things as separate entities, just as a big raster map. So for text, there are programs out there that examine the pixel configuration to try to "interpret" text out of it. Thats what OCR software does with an image that has been scanned... It can translate raster pixels into actual font text values.
I hope this helps..
-Andrew