Unable to extract scanned pdf using TesseractOCRConfig Apache Tika

Question

My pdf contains scanned images and I want to extract text from it.

What I tried : I tried with AutoDetectParsers but no output.

I followed the solution provided in Apache Tika extract scanned PDF files and also Apache Tika Jira at https://issues.apache.org/jira/browse/TIKA-1729 but getting empty string without any error.

My configuration : Win 7 64-bit OS, JDK 1.8.0_45.

Any kind of help is welcome.

Geniegenii

Do you have Tesseract installed and at the location given in your config? Did you try following the Tika Troubleshooting Guide? – Popish

@Popish I am using maven to install all jars which includes Tesseract. I have taken a look at the Troubleshooting guide for No Content Extracted problem. I have used the most recent version(1.13) of the Apache-tika-app.jar and tried to use the GUI to check the extraction but no output. – Geniegenii

Tesseract is not a Java library, so Maven won't help you. You need to download and install the native program for your operating system – Popish

@Popish I don't want to use any software for this. I want to use TesseractOCR java api which can be used inside my java application. Anyway just for fun, I installed tesseract desktop app and tried my pdf, its extracting some incorrect words. – Geniegenii

Tesseract is a native program you have to download and install separately. All Tika ships is the appropriate wrappers around Tesseract to enable it to be used if installed – Popish

@Popish Can I exctract text from a scanned pdf without installing any native program in my system? If No, then it will drag me into a dependency of a native program to run my java application which I want to avoid. – Geniegenii

Try softwarerecs.stackexchange.com – Popish

@Popish Thanks for your help. I have installed Tesseract and tried to run tesseract from tika using new TesseractOCRConfig().setTesseractPath(tesseractFolder);. I can easily extract text from images, pdf containing single image but not from pdfs where multiple images are present. I am not getting any error but no output. – Geniegenii

This helped me link in solving the issue. The issue was : Tika dropped support for extracting TIFF images from PDFs in 1.13 and for that we need to add one more dependency <dependency> <groupId>com.github.jai-imageio</groupId> <artifactId>jai-imageio-core</artifactId> <version>1.3.1</version> </dependency>. Thanks. – Geniegenii

Answer 1

Steps to follow to solve this :

1. Install Tesseract in your system using 'tesseract-ocr-setup-3.05.00dev.exe' for Windows from: https://sourceforge.net/projects/tesseract-ocr-alt/files/ and set its location in your config.

   Java code :

   Parser parser = new AutoDetectParser();
   BodyContentHandler handler = new BodyContentHandler(Integer.MAX_VALUE);
   TesseractOCRConfig config = new TesseractOCRConfig();
   config.setTesseractPath(tPath);
   PDFParserConfig pdfConfig = new PDFParserConfig();
   pdfConfig.setExtractInlineImages(true);
   pdfConfig.setExtractUniqueInlineImagesOnly(false); // set to false if pdf contains multiple images.
   ParseContext parseContext = new ParseContext();
   parseContext.set(TesseractOCRConfig.class, config);
   parseContext.set(PDFParserConfig.class, pdfConfig);
   //need to add this to make sure recursive parsing happens!
   parseContext.set(Parser.class, parser);
   

2. Maven dependencies :

<dependencies> <dependency> <groupId>org.apache.tika</groupId> <artifactId>tika-parsers</artifactId> <version>1.13</version> </dependency> <dependency> <groupId>com.levigo.jbig2</groupId> <artifactId>levigo-jbig2-imageio</artifactId> <version>1.6.5</version> </dependency> <dependency> <groupId>com.github.jai-imageio</groupId> <artifactId>jai-imageio-core</artifactId> <version>1.3.1</version> </dependency> </dependencies>

I think it may be helpful. Thanks.

Geniegenii

Thank you for this. Beware of the licensing implications of using levigo and jai. If they were Apache 2.0 compatible, we would have embedded them. – Thrill