OCR Development for Numeric Data Extraction

I am in need of an OCR system, preferably designed in Python or C#, that will capably extract numbers from PNG images. The unique challenge lies within the font of these numbers which is small, and might seem strange to typical OCR programs. You're permitted to modify images, but bear in mind that the resulting numbers must maintain their accuracy post-modification.

Here are some key aspects:

*The Main Objective*
- The primary goal is to reliably extract numbers from my images.

Ideal candidates for this project are those with deep experience in Python or C#, as well as an extensive background in Optical Character Recognition systems. A thorough understanding of manipulating image data will be considered a major advantage. I am hopeful for a solution that is effective, efficient, and consistent. Can't wait to see your proposals.