silvester / tesseract-ocr-for-php

A wrapper to work with TesseractOCR inside your PHP scripts.

TesseractOCR for PHP

A wrapper to work with TesseractOCR inside your PHP scripts.

Installation

Via composer (https://packagist.org/packages/thiagoalessio/tesseract_ocr)

{
    "require": {
        "thiagoalessio/tesseract_ocr": ">= 0.2.0"
    }
}

Or just clone and put somewhere inside your project folder.

$ cd myapp/vendor
$ git clone git://github.com/thiagoalessio/tesseract-ocr-for-php.git

Dependencies

• TesseractOCR

IMPORTANT: Make sure that the tesseract binary is on your $PATH. If you're running PHP on a webserver, the user may be not you, but _www or similar. If you need, there is always the possibility of modify your $PATH:

$path = getenv('PATH');
putenv("PATH=$path:/usr/local/bin");

Windows users

I received several messages from people trying to get this library running under Windows, so I decided to write a short tutorial that can be found here.

Usage

Basic usage

<?php
require_once '/path/to/TesseractOCR/TesseractOCR.php';
//or require_once 'vendor/autoload.php' if you are using composer

$tesseract = new TesseractOCR('images/some-words.jpg');
echo $tesseract->recognize();

Defining language

Tesseract has training data for several languages, which certainly improve the accuracy of the recognition.

<?php
require_once '/path/to/TesseractOCR/TesseractOCR.php';
//or require_once 'vendor/autoload.php' if you are using composer

$tesseract = new TesseractOCR('images/sind-sie-deutsch.jpg');
$tesseract->setLanguage('deu'); //same 3-letters code as tesseract training data packages
echo $tesseract->recognize();

Inducing recognition

Sometimes tesseract misunderstand some chars, such as:

0 - O
1 - l
j - ,
etc ...

But you can improve recognition accuracy by specifing what kind of chars you're sending, for example:

<?php
$tesseract = new TesseractOCR('my-image.jpg');
$tesseract->setWhitelist(range('a','z')); //tesseract will threat everything as downcase letters
echo $tesseract->recognize();

$tesseract = new TesseractOCR('my-image.jpg');
$tesseract->setWhitelist(range('A','Z'), range(0,9), '_-@.'); //you can pass as many ranges as you need

You can even do cool stuff like this one:

<?php
$tesseract = new TesseractOCR('617.jpg');
$tesseract->setWhitelist(range('A','Z'));
echo $tesseract->recognize(); //will return "GIT"

Troubleshooting

Warnings like Permission denied or No such file or directory

To solve this issue you can specify a custom directory for temp files:

<?php
$tesseract = new TesseractOCR('my-image.jpg');
$tesseract->setTempDir('./my-temp-dir');