You can either read the PNG specification in detail and parse the file, or use existing code such as libpng.
The image matrix is nothing to do with the file format, though. It is a simple transformation matrix used to position and scale the image from its default 1 x 1 unit size. You should read 32000-1 if these concepts are not clear.
The attributes would be similar but you must provide width and height, check bitspercomponent, and consider the best colour space (32000-1 again if not clear).
