How to convert a pixel of an image into a matrix form?

After canny edge detection and attaining the pixels of each character in the image, how to get the matrix of each and every alphabet?

Answer 1

[Image Analyst]

I assume you want to convert the text that you see in the image into ASCII or Unicode text. This is called Optical Character Recognition, or OCR. Thousands of people have been working on OCR for ages. No one here is an expert on OCR and besides, it's so complicated that it's way too much to give you a tutorial or code for an OCR MATLAB package here in the Answers forum. The best we can say is to check the File Exchange for OCR, or check Vision Bib for the papers that have been written on it, or to buy a package if you just need to get the job done. For example Adobe Acrobat has OCR built in (and I'm sure they've worked many years on it) - that could be the most cost- and time-efficient way for you to convert your images.

[Mohammad]

Okay, But I want it for my project that I am doing for my master's degree. I have been following the procedure followed in the link below http://www.mathworks.com/matlabcentral/fileexchange/8676-character-recognition-example-iiitraining-a-simple-nn-for-classification but in this code they have used 50 characters where as I am using 26. When I replace 50 with 26 and with the understaind of neural network it gives me an error.

[Image Analyst]

Sorry, I'm not familiar with that program or with Neural Networks. If you're doing your Master's degree you'll need to understand, and perhaps replicate, other people's work before you improve upon it, so I understand why you want to do it, but that's not work I can help you with.

[Mohammad]

Yes but I am not really very much familiar with image processing.. Although canny edge detection is done, I am having difficulty in the recognition of letters, as I don't know how to extract the different pixels from the image and to put it in matrices form for recognition. I read and tried applying entropy, mean, no. center white pixels, no. of white pixels in middle row, no. white cells in center column, corner pixels, boundary pixels, and the two diagonal pixels but none has shown me any sort of result as I know that I am making some mistake in it.

[Walter Roberson]

There is no "Yes but". Your not being familiar with image processing does not somehow make Image Analyst any more familiar with Neural Networks. Image Analyst told you where to find papers on the topic, and told you where you can find some of the OCR programs that have already been implemented.

If you are having difficulty implementing an algorithm you already understand, then first consult http://www.mathworks.com/matlabcentral/answers/8026-best-way-s-to-master-matlab . Once you have some code, if you are having difficulty understanding a particular error you are encountering, then start a new Question in which you post that code and describe the error message (and which line it is occurring on.)