Any good algorithms for text localization in images?

Question

I would like to ask you if you know any good text localization algorithms that would detect text candidates in an image (for my OCR project)

Essentially, after 'applying' this algorithm I would like to be able to get regions (bounding boxes) with character candidates, e.g.

I am trying to find something that I might use but even if I find something it's most likely in an extremely difficult paper with really high maths that needs to be applied. I have already encountered MSER (Maximally Stable Extremal Regions) or Gradient Vector Flow method but both of them are quite difficult for me (although I understand a lot in maths I still have hard time figuring these out)

Answer 1

These may give you some pointers:

1. Algorithm to detect presence of text on image
2. https://stackguides.com/questions/850717/what-are-some-popular-ocr-algorithms?rq=1
3. What are good algorithms for vehicle license plate detection?

Answer 2

You can use Opencv3.0 "Scene Text Detection" functions. It is based on 'Class-specific Extremal Regions for Scene Text Detection'. It also has sample code.

You can find them at:

[1] http://docs.opencv.org/3.0-beta/modules/text/doc/erfilter.html

[2] https://github.com/Itseez/opencv_contrib/blob/master/modules/text/samples/textdetection.cpp

Papers:

[Neumann12] Neumann L., Matas J.: Real-Time Scene Text Localization and Recognition, CVPR 2012. The paper is available online at http://cmp.felk.cvut.cz/~neumalu1/neumann-cvpr2012.pdf

[Neumann11] Neumann L., Matas J.: Text Localization in Real-world Images using Efficiently Pruned Exhaustive Search, ICDAR 2011. The paper is available online at http://cmp.felk.cvut.cz/~neumalu1/icdar2011_article.pdf

[Gomez13] Gomez L. and Karatzas D.: Multi-script Text Extraction from Natural Scenes, ICDAR 2013. The paper is available online at http://158.109.8.37/files/GoK2013.pdf

[Gomez14] Gomez L. and Karatzas D.: A Fast Hierarchical Method for Multi-script and Arbitrary Oriented Scene Text Extraction, arXiv:1407.7504 [cs.CV]. The paper is available online at http://arxiv.org/abs/1407.7504

Answer 3

I don't know any easy or simple algorithms to do that. However, we are using a third-party toolkit named leadtools to do a similar behavior. I know that it can find the text zones automatically and recognize them. For more information, you can see this help topic which includes a code snippet:

Auto Zone

Also, you can do some image processing to the images before OCR them, since colors in images decrease the accuracy and the performance of the OCR recognition process.