# What is meant by binary images of handwritten digits?

I can understand what a handwritten digit is, but what is meant by "binary image"? Can someone explain it?

I am studying machine learning concepts. In doing so, I've come across many types of inputs. For example: binary, continuous, and discrete. The inputs are provided through MNIST handwritten image datasets. In that, I have read about datasets consisting of binary images.

I would also like to know about the difference between greyscale and binary images.

• Is this question about statistics, machine learning, data analysis, data mining, or data visualization? If so, can you please clarify the connection & what you are asking? Commented Jun 20, 2014 at 2:23
• This question appears to be off-topic because it is not about statistics as defined in the help Commented Jun 20, 2014 at 3:15
• @Glen_b: can you suggest a better stackexchange site for it? Commented Jun 20, 2014 at 8:23
• Maybe dsp, since that includes image processing (but their help about what's on topic isn't very informative). Commented Jun 20, 2014 at 9:06

It's a digital representation of the digit in raster (pixel) form, just like the pictures on the Internet you've seen. The additional difference is that the pixels be binary -- black or white, no color or shade of gray.

• Is bilevel image and binary image are different? Commented Jun 25, 2014 at 1:48
• Same.
– Emre
Commented Jun 25, 2014 at 1:49
• Thanks. Binary pixels means 0-1. If Grey level included then how pixels will be represented in digits? Commented Jun 25, 2014 at 1:54
• By binarization, typically mapping gray levels below a threshold to black, and to white otherwise.
– Emre
Commented Jun 25, 2014 at 2:02
• It seems you would be well served by reading a book on image processing. These are elementary questions and I'm sure you will have more as you learn.
– Emre
Commented Jun 25, 2014 at 2:34