This tool allows the user to use a touch screen device and finger gestures to dynamically create or modify a 3D probabiliy distribution.

Input Parameters:

• (Optional) Width, height, and matrix representing overall probability distribution map

Output Parameters:

• Mastrix representing newly created or modified probability distribution map

• Create test application with OpenTK (OpenGL for C#).
• Interface components:
  • Interface to specify width and height for new probability distribution map.
  • Networking component to communicate to central repository.
  • Button to contact central repository to load current probability distribution map.
  • Button to save distribution map to central repository.
  • Interface to load probability distribution map from a file.
  • Interface to save probability distribution map to a file.
  • Button to view probability distribution in 3D (read-only).
• Implement generating 3D surface based on probability distribution map.
• Implement colormap on 3D surface.
• (Optional) Overlay terrain map on top of probability distribution map.
• Implement 2D navigation gestures (two-finger gestures)
  • Zoom in (two fingers going opposite directions moving away from each other)
  • Zoom out (two fingers going opposite directions moving closer to each other)
  • Pan (two fingers going the same directions in parellal)
• Implement 3D navigation gestures (two-finger gestures)
  • Rotate left (one finger fixed, other finger moving circularly counter-clockwise)
  • Rotate right (one finger fixed, other finger moving circularly clockwise)
  • Rotate up (one finger fixed, other finger moving toward first finger vertically)
  • Rotate down (one finger fixed, other finger moving away from first finger vertically)
• Implement 2D modification gestures (one-finger gestures)
  • Raise hill (clockwise circular motion)
  • Flatten hill (counter-clockwise circular motion)
  • Delete (cross)
  • Un-do (straight down)
  • Re-do (straight up)
• Gesture Recognition:
  • First detect one-finger or two-finger gesture.
  • Certain gestures only work with certain views.
  • 2D navigation gestures just compute and classify.
  • 3D navigation gestures just compute and classify.
  • 2D modification gestures use k-nn to learn
    • Build training interface
    • Raise hill add Gaussian distribution. First circle defines Gaussian shape (3 standard deviation)
    • Flatten hill substract Gaussian distribution. First circle defines Gaussian shape (3 standard deviation)
    • Delete, first stroke in the cross defines bounding box.

• Play with sample touch screen test app
• Code up gesture recognition (training and classification)
• Code up a quick prototype DistMod GUI (with openGL)