BML OCR:camera: (Experimental)

BML Transaction Receipt OCR based on Tesseract OCR. At the moment this returns the result as an array. We will be switching to an Receipt object with regex matching for a later release.

Would not recommend to use in production.

Installation

Via Composer:

composer require baraveli/bml-ocr

‼️ This library depends on Tesseract OCR, version 3.02 or later.

Note for Linux users

It is pretty simple to install tesseract, run the following commands:

sudo apt install tesseract-ocr

If the default installation doesn't include english language pack. You have to pull that one seperately, run the following command:

sudo apt-get install tesseract-ocr-eng

Note for Windows users

There are many ways to install Tesseract OCR on your system, but if you just want something quick to get up and running, I recommend installing the Capture2Text package with Chocolatey.

choco install capture2text --version 3.9

⚠️ Recent versions of Capture2Text stopped shipping the tesseract binary.

Note for macOS users

With MacPorts you can install support for individual languages, like so:

$ sudo port install tesseract-<langcode>

But that is not possible with Homebrew. It comes only with English support by default, so if you intend to use it for other language, the quickest solution is to install them all:

$ brew install tesseract --with-all-languages

Usage

use Baraveli\BmlOcr\BmlOcr;

BmlOcr::make("/home/image/imagename.jpg", __DIR__)
        ->detect();

The first argument of the make method is the path to the receipt you want to do OCR on. and the second argument is the temporary directory path. When the image path is first given it creates a temporary image with higher sharpness because the original BML receipt uses grey color for typography. It is really hard to detect by using just the default image. It deletes after finishing the detection.

When you call the detect method it will return the result as an array.

$receipt = BmlOcr::make("/home/image/imagename.jpg", __DIR__)
        ->detect();

var_dump($receipt);

Result:

array:9 [
  0 => "Transaction Receipt"
  1 => "Status SUCCESS"
  2 => "Message Transfer transaction is successful."
  3 => "Ref # BLAZ418696504822"
  4 => "Date 30/07/2020 21:46"
  5 => "From SHIHAM A.RAHMAN"
  6 => "To Mohamed Jinas"
  7 => "7730000151614"
  8 => "Amount MVR 750"
]