问题:

Is it possible that read text from an image without using any third party APIs?

回答1:

Yes, you can definitely write your own optical character recognition (OCR) software. It's robustness depends entirely on the effort you put into it.

For example, recognizing characters of an exact size and font could be as simple as region matching against known character shapes; more general strategies might require some sort of statistical matching or even machine learning techniques.

OCR is still an active area of research at top-tier computer science universities and institutions (e.g. Google), so don't expect to find any quick-and-easy solutions.

回答2:

There is no built-in OCR support in the standard Java libraries, so you would have to implement it yourself.

If you do opt to go for a third party library, the following page lists several suggestions: https://stackoverflow.com/questions/971344/java-based-ocr-sdk-api

回答3:

As those libraries were written, it is clear that someone with the right knowledge can write such libraries. Therefore it's possible to extract the text from an image without 3rd party libraries, just write your own.

Whether you are able to do so, it is impossible for me to know.