Im using GOCR for reading the textual content in a photo/frame captured by a webcam.. But most f the time teh GOCR cannot read the content and gives erroneous outputs..I believe that the error maybe due to the lighting conditions present in the photo..
can some one please tell me why this photos are not read by the GOCR? Is there a problem with the lighting conditions? Are the letters too small? etc....
Thank you sooo much...
(btw, how can i add photos to this thread??? :/)
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Your samples are very clean, however I would recommend a bit of pre-processing:
1) Try turning your image to 2-color black and white. Doing this will eliminte some grey-dark pixels in the background.
2) Try Blurring/sharpening the image. The text has a "shadow" around, and may influence the results.
3) Maybe resizing the image? It may recognize your simbols as large-long bars instead of chars if the image is too big...
4) You have to build a database of chars to be recognized. The various -m options are documented in the code (easy to find), play with them. You will have to create an image of each possible chars, and add them to the DB for them to be recognized.
Hope it help,
-Jay
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
GOCR is not the better framework for it.
I have developped a program which used it. the result is good only if letters is correctly aligned. You can use "unpaper" to correct rotation.
GOCR must know letters with their size (resolution). If your picture change of resolution, GOCR don't find the good letter
You must use black and with picture, and have a picture clean (no pixel alone)
Im using GOCR for reading the textual content in a photo/frame captured by a webcam.. But most f the time teh GOCR cannot read the content and gives erroneous outputs..I believe that the error maybe due to the lighting conditions present in the photo..
can some one please tell me why this photos are not read by the GOCR? Is there a problem with the lighting conditions? Are the letters too small? etc....
Thank you sooo much...
(btw, how can i add photos to this thread??? :/)
I have uploaded the sample images to the flickr.. None of these images are read properly... I hope u can help me.... :(
http://www.flickr.com/photos/29294415@N04/2737855738/
http://www.flickr.com/photos/29294415@N04/2737855598/
http://www.flickr.com/photos/29294415@N04/2737019499/
thanks
Your samples are very clean, however I would recommend a bit of pre-processing:
1) Try turning your image to 2-color black and white. Doing this will eliminte some grey-dark pixels in the background.
2) Try Blurring/sharpening the image. The text has a "shadow" around, and may influence the results.
3) Maybe resizing the image? It may recognize your simbols as large-long bars instead of chars if the image is too big...
4) You have to build a database of chars to be recognized. The various -m options are documented in the code (easy to find), play with them. You will have to create an image of each possible chars, and add them to the DB for them to be recognized.
Hope it help,
-Jay
GOCR is not the better framework for it.
I have developped a program which used it. the result is good only if letters is correctly aligned. You can use "unpaper" to correct rotation.
GOCR must know letters with their size (resolution). If your picture change of resolution, GOCR don't find the good letter
You must use black and with picture, and have a picture clean (no pixel alone)
i advise you to try to implement an algorithm with skeleton letter.
http://www.developpez.net/forums/showthread.php?t=20716&highlight=traitement&page=5
it is a french link (i am french) lol
Good Luck!
Cédric