How does X-Y-cut method for offline handwriting recognition work?

Question

I would like to get the idea of X-Y-cut algorithm used for offline handwriting recognition.

X-Y-cut is a histogram based approach, used for the reason of automatic handwriting recognition.

First steps of the algorithm are the following:

1. We use a binarized picture, which contains typed text.
2. We count the number of black pixels for each line and project the value on the Y-axis.
3. We do the same for the X-axis.

At this stage, before applying the classification techniques, I would like to know more details that we do after getting X and Y axes with histograms, which show a number of black pixels.

Answer 1

So, the idea is the following.

After we projected each line of black pixels on the Y axis, we can identify black pixels responsible for letters cutting each line of that histogram according to the X projection of white pixels.