Its the mathematical term for estimation and constructing new data from existing data. In the context of what you are doing, it’s resizing images.
You are doing something called linear interpolation. This works great for shrinking an image. However, have you considered what happens when your ASCII resolution is greater than your image resolution? This is where bilinear and bicubic interpolation come in.
These algorithms are cool but are also massive overkill for your average use case. They only make a different in a very niche use case (when your ASCII resolution is greater than your image, such as pixelart)