If you butt them up against each other horizontally and vertically you would have a better chance of a fixed set of crops working.
Sent from my iPhone > On 8 Feb 2015, at 12:18, Josh Wolcott <jswolc...@gmail.com> wrote: > > The issue is I need to scan 9 cards at a time. I can crop the individual > cards out ok. But I can not accurately crop out the two text boxes without > some sort of automation. > > Ugh... opencv wont dl from soruceforge... what the... > > The attached input is my preferred input. The actual input image is a 20 meg > tiff. I'm putting the jpg up for times sake. > >> On Sunday, February 8, 2015 at 7:11:28 AM UTC-5, Allistair C wrote: >> Could you upload a scanned card at the resolution and angle that you tried >> without success? >> >> Sent from my iPhone >> >>> On 8 Feb 2015, at 12:05, Josh Wolcott <jswo...@gmail.com> wrote: >>> >>> I will look in to opencv. Thank you. >>> >>> I spent many hours on this last night with less than no success. I tried >>> simply placing the cards more carefully so that I had more easily cropped >>> data, however even then there is way way way too much variance. So even the >>> "simple" solution is unworkable. >>> >>> I'm actually rather amazed at how difficult this has turned out to be. I >>> went in to it assuming there would be an out of box solution to command >>> line OCR. The project was going swimmingly until I actually got to this. My >>> patience is beginning to wain =( >>> >>>> On Sunday, February 8, 2015 at 4:23:25 AM UTC-5, Allistair C wrote: >>>> I would personally use opencv rather than IM. It has more sophisticated >>>> routines to build on. >>>> >>>> http://stackoverflow.com/questions/16746473/opencv-find-bounding-box-of-largest-blob-in-binary-image >>>> >>>> Sent from my iPhone >>>> >>>>> On 8 Feb 2015, at 00:02, Josh Wolcott <jswo...@gmail.com> wrote: >>>>> >>>>> You know you really put it well there. I think a good question is "Do I >>>>> have the time for this". I really like the idea of being able to drop >>>>> them any old way and letting the application deal with it. I like the >>>>> resiliency of it. However, "bravely struggle" is well put. It's a battle. >>>>> >>>>> >>>>> If I just deal with the fact that I will have to put the images in >>>>> straight I can simply go crop crazy and I think I will get good output. >>>>> It's not sexy but.... It will get me past these projects. >>>>> >>>>> I am going to look in to this hough line detector though. >>>>> >>>>> Another thing I note... as in most things I'm not the first one to attack >>>>> this exact issue. There are a couple projects on github that do this with >>>>> a web cam. I don't know what they are doing though because I don't read >>>>> python. I need to use a scanner any way... would be nice to see how they >>>>> approached it though. >>>>> >>>>>> On Saturday, February 7, 2015 at 5:20:59 PM UTC-5, Dmitri Silaev wrote: >>>>>> Two approaches then. >>>>>> >>>>>> First, you drop the cards in the scanner any old way and then bravely >>>>>> struggle to rectify images. Or, just always place the cards evenly, >>>>>> possibly many at a time and crop fixed regions. >>>>>> >>>>>> If you prefer the first one, and I're not going to program, I may >>>>>> suggest ImageMagick, particularly read about Hough Line Detector. >>>>>> Essentially, then you'll need to employ a kind of programmatic thinking >>>>>> to make use of detector results for image deskewing (need to study more >>>>>> ImageMagick command line switches). Then crop and do OCR. >>>>>> >>>>>> But think twice, wouldn't it be an easier strategy to place the cards >>>>>> evenly? >>>>> >>>>> -- >>>>> You received this message because you are subscribed to the Google Groups >>>>> "tesseract-ocr" group. >>>>> To unsubscribe from this group and stop receiving emails from it, send an >>>>> email to tesseract-oc...@googlegroups.com. >>>>> To post to this group, send email to tesser...@googlegroups.com. >>>>> Visit this group at http://groups.google.com/group/tesseract-ocr. >>>>> To view this discussion on the web visit >>>>> https://groups.google.com/d/msgid/tesseract-ocr/9fcb7b78-7fc3-46d7-b791-8abfddae30f1%40googlegroups.com. >>>>> For more options, visit https://groups.google.com/d/optout. >>> >>> -- >>> You received this message because you are subscribed to the Google Groups >>> "tesseract-ocr" group. >>> To unsubscribe from this group and stop receiving emails from it, send an >>> email to tesseract-oc...@googlegroups.com. >>> To post to this group, send email to tesser...@googlegroups.com. >>> Visit this group at http://groups.google.com/group/tesseract-ocr. >>> To view this discussion on the web visit >>> https://groups.google.com/d/msgid/tesseract-ocr/589bf73c-f2eb-4a2f-a094-35c541638444%40googlegroups.com. >>> For more options, visit https://groups.google.com/d/optout. > > -- > You received this message because you are subscribed to the Google Groups > "tesseract-ocr" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to tesseract-ocr+unsubscr...@googlegroups.com. > To post to this group, send email to tesseract-ocr@googlegroups.com. > Visit this group at http://groups.google.com/group/tesseract-ocr. > To view this discussion on the web visit > https://groups.google.com/d/msgid/tesseract-ocr/6fd491bf-2e02-42c4-9c37-085a94e1489b%40googlegroups.com. > For more options, visit https://groups.google.com/d/optout. > <color.jpg> -- You received this message because you are subscribed to the Google Groups "tesseract-ocr" group. To unsubscribe from this group and stop receiving emails from it, send an email to tesseract-ocr+unsubscr...@googlegroups.com. To post to this group, send email to tesseract-ocr@googlegroups.com. Visit this group at http://groups.google.com/group/tesseract-ocr. To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/A0E0E3F9-916B-4E72-ADAB-291CD7D0068B%40gmail.com. For more options, visit https://groups.google.com/d/optout.
