Add padding transformation for vision.

[tnat410]

Adding padding for vision in the transformation pipeline of the data reader. Padding the given image on all sides with 0 (black) value.