The forum just ate my reply :(

Second time lucky:

You need to be a bit more specific about your needs: e.g. does the 'canvas' represent a specific geographical location.  If so you can use the corner coordinates to calculate the ratio of pixel to physical coordinates and, assuming your map isn't rotated, simply subtract the top left map coordinate from the current GPS coordinate, scale the result with the calculated ratio and, assuming it's within the map bounds, display the user's location on screen.

If the screen represents an arbitrary space around the user's current location you'll need to capture a starting point (e.g. require a button press when the user is in a suitable location) and then use the current GPS coordinates to map the screen to the current location using principles similar to above.

I haven't done any of this stuff in Processing yet, but have worked with HP's now defunct mscape toolkit; am just getting to grips with the Calvium toolkit (for creating locative media on iPhone and hopefully in the not too distant future Android) and have also coded stuff in Flash on Windoze mobile...  Actually the Flash code should be easy enough to translate to Processing so I could perhaps dig that out at some point ;)