AI Project: Matching w Skewed Digicam Photographs

1
AI ProjectMatching w/ Skewed Digicam Photographs

• Chris Trevino

2
The Vision

• The original goal of this project was
• To find a way to assist in the
• Restoration of broken stalagmites/stalactites
• In natural caverns

3
The Vision
4
The Vision
5
The Vision

• Normally
• Images/Cameras are registered to known reference
  axes
• Lighting is controlled
• Lots of expense time

6
The Vision

• In this case
• Images might be orthogonal to a surface
• Images might be along a cylindrical axis
• Lighting is UNcontrolled
• Opportunistic cheap digicam shots 4Mp

7
Test Data

• Caves are closed for bats (no kidding)
• Slicing vegetables seems to be a good alternative
• Internal variegation
• Perspective skewing and rotation

8
Test Data

• Jalapenos are HARD to match
• These are pairs
• Seeds are trouble
• Fast-longitudinal change isdifficult

9
Evolution

• The current comparison algorithm
• uses image intensities
• looks for similarities
• Future feature recognition?
• The uniqueness of this approach involves
  histograms of annuli

10
Interface

• The user must select the region of interest in an
  image
• Then the system analyzes it

Selection is Broken Into Layers
annuli
11
Example Histograms
These intensity graphs provide the basis for
comparing images.
12
Business Prospects

• Tool for anthropologists/archaeologists
• Forensics
• Job shop assembly parts-matching tasks
• Patenting some variation of the methods?