A short look to mobile 3-D. Mobile handheld devices will be measuring 3-D environment around them. Softwares that process the 3-D data will be more important than ever.
1. 1/10
Mobile 3D
Dr.-Ing. Sung Joon Ahn
CurvSurf, Inc.
joon.ahn@curvsurf.com
2. 2/10
Future of Consumer’s Life
• Provided with information everywhere, every time,
any through, but kindly
• Mobile device
• Home appliance
• Driverless car
• Working bench
• OS, the core
• Platform
• Ubiquitous
• IoT.
3. 3/10
OS & 3D Mobile
• Market share is everything
• OS will govern
• New products and services are following
• Broaden the application field
• Killer applications
• All in one
• Synergy
• 3D Mobile
• The device that will break the current market saturation
• Must have such OS which supports 3D Mobile applications
• The race is just beginning.
4. 4/10
Mobile 3D Scanners
Microsoft Kinect by
Primesense (Apple)
Matterport Mantis Vision
Occipital Google PMDTechnologies
6. 6/10
2D vs. 3D
• 2D
• Camera, scanner, printer, facsimile, …
• Image processing (x, y, color)
• Photo editor vs. object recognition
• Object features: Shape, size, position (x, y), rotation (휃)
• 3D
• Stereo camera, 3D scanner, 3D printer, …
• 3D data processing (x, y, z, color)
• Computer graphics vs. computer vision
• Features: Shape, size, position (x, y, z), rotation (휔, 휑, 휅).
7. 7/10
Hardware vs. Software for Mobile 3D
• HW
• Since 1980’s
• Various 3D measuring principles
• Ripe enough to be integrated into a 3D Mobile
• SW
• Since 1960’s
• Image processing is popular (E.g. license plate recognition)
• 3D visualization and printing are also popular
• But, 3D object recognition shows a long way to go.
It would be the Holy Grail, never possible …
8. 8/10
CurvSurf, Inc.
• Strategy
• Frees App developers from mathematics
• Develops MW for automatic feature extraction
• App developers are responsible for specific applications
• MW
• Extracts 2D/3D features from 2D/3D data
• Objects: 2D/3D curves and 3D surfaces
• Shape, size, position, and pose of objects
• OS
• Provides basics for 3D Mobile operation
• 3D capturing, graphics, database, UI, ...
9. 9/10
MW of CurvSurf, Inc.
• Automatic 3D Object Recognition and Extraction
• A point is picked manually or automatically
• The proper surface type for the small point cloud patch
around the picked point is automatically determined
• Type of surface: Plane, sphere, cylinder, cone, or torus
• The inlier points are segmented and the surface parameters
are estimated simultaneously
• Surface parameters: Shape, angle, size, position, orientation
• Future of MW
• Includes further curves and surfaces in 2D/3D space
• E.g., ellipse, parabola, hyperbola, sine curve, catenary, helix,
superellipse, superellipsoid, paraboloid, hyperboloid, …
10. 10/10
Closing
• Further discussions are welcomed
• joon.ahn@curvsurf.com
• http://www.youtube.com/user/finger6207
Thank you!