How to extract text without using OCR in sikuli?

I do not really understand your question??

--- How to extract text without using OCR?
--- I want to extract test from an image

The latter is called OCR in Sikuli and is the feature Region.text() (http://sikuli.org/docx/region.html#extracting-text-from-a-region).

But be aware: bug 710586

As OCR feature is not reliable so is there any to extract text from an image?

No, the only solution available for free is Tesseract and this is used in Sikuli.

The only other possibility is to try to select the text using mouse actions and ctrl/cmd-c to get it to the clipboard and then use Env.getClipboard(), but ..
- this cannot be mixed with paste()
- only works, if the text is selectable via mouse actions

If we are talking about webpages, you might mix Sikuli with Selenium.

Does Sikuli has xml or html parser? If it has a feature like Watir, sikuli is both good at visual finds and finds which are located in xml/html between tags. As Niklikesh wrote, sometimes we really need to capture an area which is not easy to capture with visually. We have limited features at this point. These are OCR, find and doubleClick and Ctrl-C, and sth similar methods.