Convert Image to Text using Google Drive Free OCR Software

In my previous study post I've mentioned about how you can turn your commuting time into a productive one by using a text to speech online reader.
And some have asked me whether if it is possible to read an image file or pdf document.

The answer is Yes~
But! In order for the reader to perform text-to-speech translation you'll first have to convert your pdf/image file into a flat text file using an Optical Character Recognition (OCR) software~

---

So what is Optical Character Recognition and its uses?

OCR is a type of character recognition technology that is used for translating images or pdf files into pure text document. The algorithm will scan, detect, differentiate and convert characters to plain text at an efficient speed.

The extracted text can then be used for various applications:

• Data Mining and text mining 
• Text to Speech 
• Language translation (who needs an interpretor, I want my カレーライス) 
• Assistive tool for Blind and visually impaired 
• Files keeping, saving of storage space, compression. Typically an image file will take up more bytes than a text file.

---

How to extract text using Google Drive OCR software 

1. Login to Google drive
2. Click on the gear icon on the right hand side of your window 
3. Go to upload settings and check these two settings
   ☑Convert text from uploaded PDF and image files
   ☑Confirm settings before each upload 



Google Drive OCR settings 

4. Now when you upload a file a prompt window will appear and ask you to confirm your settings 
5. Leave the setting as it is if you are converting English Document 



I am going to convert a short Japanese passage

6. Once the upload is done, open up your new file
7. The extracted plain text will be shown below your image 

Raw Image Top, Words Below

See it works for Japanese Characters as well~ I am able to get a 100% accuracy from this Japanese characters image.

Things to note:

Google Drive file size limitation for OCRed is 2MB. (.jpg, .gif, .png and PDF)

It will only look at the first 10 pages of PDF when searching for your text to extract.

Always do your checks OCRed document may not always be accurate. 

For best accuracy:
Use high resolution well lighted image, good contrast between characters and background
Orientate your text in the correct direction. Typically the text will be scanned from left to right.
Multiple columns document dont work well with this OCR
Select the correct language (there are 38)
Use a top down view image like what I did.