HOW TO: Preprocess and OCR low resolution images

I have updated the small image OCR demo originally posted on our blog to work with Leadtools version 20. Once you have installed our version 20 toolkit, all that is necessary for this to run is to extract and compile.

Low-resolution images can come from a variety of sources, but the most common source is probably screen capture. Screen capture images are usually 96 DPI on Windows. (This can vary depending on the user settings). Additionally, image representations of faxes may also be considered low-res and fall under the resolution threshold acceptable for OCR. Typically, OCR engines require images of 200 or 300 DPI in order to achieve acceptable results.

One easy solution for images that do not have noise or complex structures is to change the resolution of the image before performing OCR. This C# sample shows you how to change the resolution, then gets the text from the image and outputs it to the console.