Hello San,
The information, that you have, is some text string which is already rendered on the image, isn't it? If so, it will be very hard to implement your requirements without OCR. You can estimate the difficulty by yourself:
1) You have to find the _rendered_text_. Even if you know size, font and color of the text it is very complicated task related to visual/character recognition algorithms.
2) You have to replace alredy rendered text by yours one. But what do you plan to do with original text background? Just fill with some new color? In such case you should find the borders of the area to fill and guarantee that the result will look fine.