If you just want to extract the characters from images you can use OCR-equiped softwares like as Foxit Phantom, infty reader and online OCR tools. 

If you want to develop character recognition algorithm using MATLAB you can use something like this code:

https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&cad=rja&uact=8&ved=0ahUKEwjExPbKud7RAhWJtBoKHbV3D4sQFggoMAM&url=https%3A%2F%2Fwww.mathworks.com%2Fmatlabcentral%2Ffileexchange%2F18169-optical-character-recognition--ocr-%2Fcontent%2FOCR%2FOCR.m&usg=AFQjCNEfhmNnq61IuVfvXLaxl68LX1FzSg&bvm=bv.145063293,d.d2s

https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&cad=rja&uact=8&ved=0ahUKEwjExPbKud7RAhWJtBoKHbV3D4sQFggoMAM&url=https%3A%2F%2Fwww.mathworks.com%2Fmatlabcentral%2Ffileexchange%2F18169-optical-character-recognition--ocr-%2Fcontent%2FOCR%2FOCR.m&usg=AFQjCNEfhmNnq61IuVfvXLaxl68LX1FzSg&bvm=bv.145063293,d.d2s

You can use the Google cloud vision api

Hey Satish,

as you asked for Matlab, I agree with Saed. The OCR module from FileExchange is pretty good, if you have a somehow standardized image acquisition procedure. Since your scans look quite different, you might have to do some image processing (cutting, increase contrast, etc.) to get the best results. I used it to retrieve patient data from x-ray scans (which is highly standardized) and it worked nearly without errors.

Greetings, David

http://www.mathworks.com/matlabcentral/fileexchange/18169

hi, David those images which i have attached is capture by camera not scanned images. so some algorithms are not working on them specially  word extraction i have applied so many algorithms on them for pre-processing of those images. i need some algorithms which are work on camera captured images.

I would be really surprised if it did not work with these images (as camera vs. scanner is not a big difference if you do it correctly, e.g. illumination, rotation etc.). Just for fun i edited your last image (admittedly the easiest) and put it into an online, free OCR converter and this worked perfectly fine:

"Digital cameras possess several unique advantages for document capture. Compact digital cameras, especially camera phones, are convenient to carry around. Professional digital cameras, with resolutions now comma* exceeding ten million pixels, have been widely used for various large-scale book digitization projects, showcasing the nor-destructive nature of digital camera, capture. However, document capture with digital cameras has many inherent limitations {1}. It is very difficult to project uniform lighting onto a document surface, and this often results in uneven illumination and color shift in the acquired images. For documents captured with handheld compact cameras, text blur is also commonplace. These degradations are of interest to this paper. For the purpose of correcting non-uniform illumination, an illumination-reflectance model "

Cheers, David

http://www.onlineocr.net/

Thanks David , ok I see that online OCR conversation website. But I need to do that thing by using MATLAB here I am having problems. I didn't had perfect MATLAB code to perform the uneven illumination and deblur. 

Obviously, the webtool was just a proof of concept. Matlab will be more powerful. So either you just need to scan these three pages (which already works with the simple webtool) or you want to implement it in MatlB for which the OCR MFEx contribution will be a very good start for further improvements. Just be creative and adapt it for your needs. 

ok sir, i look forward to it. thanks 

Hi , for line detection  you can use the Hough transform.