I just copy and paste "clean" rectangle in image editor as quick solution ;-)
This can be done also by programming. e.g. you can create box for each symbol (see my code in previous email or Michael Young code[1] where he use opencv for this) and eliminate boxes that do not fit to some expected criteria (e.g. minimum height, position). For result boxes create rectangle that you can use in tesseract (with SetRectangle[2]) [1] http://ayoungprogrammer.blogspot.ca/2013/01/equation-ocr-part-1-using-contours-to.html [2] http://code.google.com/p/tesseract-ocr/source/browse/trunk/api/baseapi.h?r=814#337 Zdenko On Tue, Jan 15, 2013 at 3:36 AM, sunitha raghurajan < [email protected]> wrote: > Can you please tell me steps you did to remove noise?? Thanks for your > time.... greatly appreciated. > > > On Monday, January 14, 2013 4:33:06 PM UTC-5, zdenop wrote: > >> It looks like I was fast in my conclusion and I forget "first rule" for >> tesseract - remove all noise ;-)... Have a look at attached image - it >> works for me: >> tesseract image4.png output digits >> >> in output there is >> 263 5819 >> >> Zdenko >> >> On Sun, Jan 13, 2013 at 8:02 PM, zdenko podobny <[email protected]> wrote: >> >>> Hi, >>> >>> I think you will need to run training for this. I tried simple c++ code >>> that show confidence values (see attachment) and for your digit 6 it >>> produced: >>> >>> symbol 5, conf: 78.5236 ---- 5 conf: 78.523613 >>> ---- s conf: 77.376984 >>> ---- a conf: 71.858353 >>> ---- B conf: 66.046341 >>> >>> It produces recognized symbol ("5") with its confidence value + there >>> are results from Choice iterator with confidence values. "6" is not there... >>> If I interpret it correctly you can not expect that current English >>> language data file will recognize "your" "6" as "6". >>> >>> Zdenko >>> >>> >>> On Fri, Jan 11, 2013 at 12:09 AM, sunitha raghurajan < >>> [email protected]> wrote: >>> >>>> Yes, this is NH license plate. The first image is with out pre >>>> processing and the second one is after processing through opencv. >>>> >>>> >>>> >>>> On Tuesday, January 8, 2013 12:58:19 PM UTC-5, zdenop wrote: >>>>> >>>>> On 08.01.2013 17:13, sunitha raghurajan wrote: >>>>> > I am using Tesseract to read license plate. The tesseract is giving >>>>> wrong >>>>> > output for digit six. My question is, Can I train the tesseract for >>>>> single >>>>> > digit 'six'. Any help truly appreciated. >>>>> > >>>>> Can you post a example of image (with digit 6) that you try to >>>>> recognize? >>>>> >>>> >>> >> -- > You received this message because you are subscribed to the Google > Groups "tesseract-ocr" group. > To post to this group, send email to [email protected] > To unsubscribe from this group, send email to > [email protected] > For more options, visit this group at > http://groups.google.com/group/tesseract-ocr?hl=en > -- You received this message because you are subscribed to the Google Groups "tesseract-ocr" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/tesseract-ocr?hl=en
