Recently I was working in photoshop, And photoshop is really great application. After using it for weeks, I am thinking about learning the logic behind the way it works. I tried few articles which was just about converting image to black n white, matching two exactly same images etc. My questions are

1) Can we develop application with features like photoshop?
2) Can I find any source, study material, books specially to work on pixels with photoshop? (I tried to search but I got only results about reading single pixel etc)
3) How can I match two images like the option in photoshop auto align - It aligns two layers by matching the pixels. So how can I match more than one pixel in the image?

You are talking about a lot of math. The data in an image file consists of Color, Red, Green, and Blue values. Some images also contain alpha. The bits are lined in an algorithm that varies slightly from image to image.

You can however look in open source projects and code snippit libraries (I found a few in MSDN's library). Mostly they use the image classes already built into .NET. However, you can look for the image data format standard; which usually consists of a set of bytes for each pixel. Here is a site that explains a little on how the image data works.