Is it still broken in version 5? The thread you posted is from 2017! One thing I noticed in the meantime: I stored my PNGs with paint.net and auto bit depth recognition. The image is grayscale. So the image wasn't stored in 32bit. Now, I forced to save the image 32bit color depth and now it works. So I assume it's a bug regarding the bit depth handling. This also explains the memory access violation althrough the pixel range doesn't violet the image pixel size. If the system internally assumes a bigger depth then it also wants to process bigger memory chunks.
Is this problem known? Otherwise I can create a bug report. Best regards zdenop schrieb am Donnerstag, 14. September 2023 um 16:52:58 UTC+2: > > https://github.com/tesseract-ocr/tesseract/issues/845 > > Zdenko > > > št 14. 9. 2023 o 16:49 Gilad Pellaeon <ld.pe...@gmail.com> napísal(a): > >> Hi, >> >> I am new to Tesseract. I searched for an OCR library, found Tesseract and >> now I want to use it for a specific measure protocol. >> >> I built Tesseract 5.3.2 from source and the dependencies leptonica-1.83, >> libpng and OpenJPEG for Windows with the Latex Visual C++ compiler for >> Windows, x64. >> >> Then I did some first tests based on the examples from the documentation >> ( *Basic_example *and *SetRectangle_example*). As data set I use >> *eng.traineddata* from the *testdata_best* repo. >> >> Now, I have a behaviour which I can't classify. I tried to recognize a >> float value in a given rectangle (with *SetRectangle * ). Tesseract >> didn't converted it (empty return). Then I manually copied the rectangle >> and saved it in a new file (see attached Single_Number.png). Then I tried >> this file without the *SetRectangle *call*. *Now it works. >> >> The attached* Protocol_table.png *is the original image, but I removed >> all other stuff in the picture. So it's empty except the number at the >> original position. Now I have the following behaviour: in DEBUG mode the >> conversion works, in RELEASE mode not. >> >> I also tried to slighty enlarge the rectangle area (see last SetRectangle >> call in the code below). But now I got a runtime exception. The resolution >> of the picture is 2625x1682. So there should be no buffer overflow?! >> >> Am I doning something wrong here? Or what's the problem for this >> behaviour? >> >> This is my basic code: >> >> //std includes >> #include <iostream> >> >> //tesseract includes >> #include "tesseract/baseapi.h" >> >> //Leptonica includes >> #include "allheaders.h" >> >> >> >> //! >> int main() >> { >> tesseract::TessBaseAPI api; >> // Initialize tesseract-ocr with English, without specifying tessdata >> path >> if (api.Init(nullptr, "eng")) >> { >> std::cout << "Could not initialize tesseract." << std::endl; >> return 1; >> } >> >> // >> Pix* image = >> pixRead("D:/projects/cpp/Tesseract-Test/Protocol_Table.png"); >> //Pix* image = >> pixRead("D:/projects/cpp/Tesseract-Test/Single_Number.png"); >> api.SetImage(image); >> // Restrict recognition to a sub-rectangle of the image >> // SetRectangle(left, top, width, height) >> api.SetRectangle(807, 1393, 93, 49); >> //api.SetRectangle(707, 1293, 193, 149); >> // Get OCR result >> char* outText = api.GetUTF8Text(); >> if (outText) >> printf("OCR output:\n%s", outText); >> >> // Destroy used object and release memory >> api.End(); >> } >> >> -- >> 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 view this discussion on the web visit >> https://groups.google.com/d/msgid/tesseract-ocr/626e06c6-ea15-45d2-86da-1bba6c069e1cn%40googlegroups.com >> >> <https://groups.google.com/d/msgid/tesseract-ocr/626e06c6-ea15-45d2-86da-1bba6c069e1cn%40googlegroups.com?utm_medium=email&utm_source=footer> >> . >> > -- 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 view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/d262537a-9065-4876-b435-071a8e596745n%40googlegroups.com.
