Re: [tesseract-ocr] Re: tessdata/eng.traineddata question

In opencv binarisation is 1 line of code, it's called threshold and you can choose various types. If I remember tomorrow I'll post some android demo code. Sent from my iPhone > On 22 Jan 2015, at 21:08, newbie wrote: > > Ok my question should have been phrased better, I aplogize.

Re: [tesseract-ocr] Re: tessdata/eng.traineddata question

At what point will you use Google to answer these simple questions? OpenCV has already been mentioned many times. Sent from my iPhone > On 22 Jan 2015, at 18:39, newbie wrote: > > Any idea of what free source is available for bininrizing in java ? > > Thanks > >> On Thursday, January 22, 201

Re: [tesseract-ocr] Re: tessdata/eng.traineddata question

Any idea of what free source is available for bininrizing in java ? Thanks On Thursday, January 22, 2015 at 12:32:53 PM UTC-5, rkomar wrote: > > On Tue, 20 Jan 2015, newbie wrote: > > > I found that vip1200.jpg works at scale Width(8654px) and > > height(5748px), but most of the time I either

Re: [tesseract-ocr] Re: tessdata/eng.traineddata question

On Tue, 20 Jan 2015, newbie wrote: I found that vip1200.jpg works at scale Width(8654px) and height(5748px), but most of the time I either get an "Invalid mem access" or out of mem(heap) error before I am able to rescale to the optimal scale.I need to come up with some other generic way to upsc

Re: [tesseract-ocr] Re: tessdata/eng.traineddata question

> Now with Olena, does it provide an api instead of a tool to preprocess(mark text regions) the image programatically ? Hi, Look at source for the "content_in_hdoc_hdlac" program in the distribution if it looks like Olena would work for you, it shows how to use Olena programmatically . Good lu

Re: [tesseract-ocr] Re: tessdata/eng.traineddata question

Not exactly an answer, but someone else with the same issue has gotten most of the way there. http://stackoverflow.com/questions/24385714/detect-text-region-in-image-using-opencv On 22 January 2015 at 15:35, newbie wrote: > ShreeDevi, > ImageMagick, seems like a manual tool, but I think the pro

Re: [tesseract-ocr] Re: tessdata/eng.traineddata question

ShreeDevi, ImageMagick, seems like a manual tool, but I think the problem I need to solve is - a generic way of image preprocessing for all images. Art, I have been looking for a text region segregation tool, had only one from matworks that looked promising. Now with Olena, does it provide a

Re: [tesseract-ocr] Re: tessdata/eng.traineddata question

Have you looked at imagemagick and related scripts for pre-processing the images? ShreeDevi भजन - कीर्तन - आरती @ http://bhajans.ramparivar.com On Wed, Jan 21, 2015 at 1:30 AM, newbie wrote: > I found that vip1200.jpg works at scale

Re: [tesseract-ocr] Re: tessdata/eng.traineddata question

I found that vip1200.jpg works at scale Width(8654px) and height(5748px), but most of the time I either get an "Invalid mem access" or out of mem(heap) error before I am able to rescale to the optimal scale. I need to come up with some other generic way to upscale and ocr images. Any ideas are

Re: [tesseract-ocr] Re: tessdata/eng.traineddata question

Oh, sorry for double post... wrong key. I have to say, that for example for captcha recognation, I do resize images to 200% or even 300%... same image not resized does not give any results. Not sure why. Probably, because font changes to more ... "oval". 2015-01-18 19:57 GMT+01:00 Marek FlashT Ruc

Re: [tesseract-ocr] Re: tessdata/eng.traineddata question

Don't use DPI metric, as it does not really count for Tesseract. The best results (that is from my experience) are obtained when font size is 70-90px (so it is a bit large for normal usage). 2015-01-15 1:58 GMT+01:00 Quan Nguyen : > You can use the command combine_tessdata >

[tesseract-ocr] Re: tessdata/eng.traineddata question

You can use the command combine_tessdata to unpack a traineddata file to examine its components. The eng.traineddata bundled with Tess4J is of 3.01 version. You may want to try 3.02 and see if it can produce be

Re: [tesseract-ocr] Re: tessdata/eng.traineddata question

On Wed, 14 Jan 2015, newbie wrote: Flash Thunder, I think I went ahead of myself in the email below. The upscaled image has the same dpi as the original image( 96dpi). I ahve upscaled pixels for which the ocr works without doing step 2 and 3(by trail and error). But I dont a

[tesseract-ocr] Re: tessdata/eng.traineddata question

Flash Thunder, I think I went ahead of myself in the email below. The upscaled image has the same dpi as the original image( 96dpi). I ahve upscaled pixels for which the ocr works without doing step 2 and 3(by trail and error). But I dont ahve a generic formula to upscale

[tesseract-ocr] Re: tessdata/eng.traineddata question

Thanks, I have it working by doing simple things. 1. I need to get the resolution upscaled to 300 dpi(including sharpening of the image) and it did the trick. On Monday, January 12, 2015 at 5:39:38 PM UTC-5, Flash Thunder wrote: > > It should identify those images without any problems, you just

[tesseract-ocr] Re: tessdata/eng.traineddata question

It should identify those images without any problems, you just need to prepare image right. 3 steps for you: 1. Tesseract likes when letters are about 70-100px height, so you need to resize your images. 2. Invert colors - as I noticed, it doesn't like it this way at all - letters must have to