will have a reasonable size based on the current zoom level.

spec and add asserts.

headers to decide how to load images instead of using their file extensions.
This makes load_image() work even when the files have the incorrect file
extensions.

other words appear in the file path to the source code.

blending for drawing diagonal lines.

Ahti Leppänen.

to each detector individually.  This way one detector doesn't stomp on the
output of another detector.

directly on already loaded data rather than needing to use an XML file as
input.

some effort to detect them and output an identity matrix in that case.

small window when the first image was bigger than the screen.

Also added an overload of operator() for pyramid_down that takes just
a single image and downsamples it.

extract_fhog_features().  Also fixed a minor bug where empty planar HOG feature
maps had 0 planes in them rather than 31 empty planes as the spec says they
should.

what scanner type it is using.

This is the dlib/image_processing/generic_image.h file.  Then I changed all the
image processing functions so that they use this interface.  All the changes
are very minor, but there are just a lot of them.

Any user code that was using array2d objects to represent images will still
work.  However, this change makes it so that users can use their own custom
image objects with dlib by simply implementing a few global functions for their
image object.