VISUAL DETECTION TECHNIQUES IN SAR

1
VISUAL DETECTION TECHNIQUES IN SAR

• Presented by
• Sgt. Tygh Thompson
• Washington County Sheriffs Office

2
THIS PROGRAM IS BASED ON RESEARCH COMPILED BY
DR. KENNETH HILL HALIFAX REGIONAL SAR NOVA
SCOTIA, CANADA
3
OVERVIEW

• Introduction to the problem
• Seeing Vs. detecting
• Why clues are missed
• Visual search
• The Visual Briefing

4
THE LOST PERSON LEAVES CLUES

• At least 1300 tracks for every kilometer walked.

5
INTRODUCTION TO THE PROBLEM

• Why are so few clues detected?
• What can be done to increase visual detection?

6
SEEING Vs DETECTION

• Insert picture here

7
SEEING Vs DETECTION
8
SIGNAL DETECTION THEORY

• Perceptual judgement is based on a combination of
  sensation and decision making processes
• Stimulus event
• The eyes detect an item
• Neural activity
• The image is analyzed by the brain

9
SIGNAL DETECTION THEORY

• Comparison with personal standard
• The image is compared to what you think the item
  should look like
• Action or no action - decision
• Did the item meet my standard of what I am
  looking for or is it important?

10
VISUAL PERSPECTIVE

• The clue must be in the center of focus or vision
• Edge detection occurs in the Fovea (center of the
  Retina)
• The eyes recognize the edges of items
• The sharper the image of the edge, the easier the
  item is to detect
• The most visual information is revealed in the
  canonical or perfect perspective

11
VISUAL PERSPECTIVE

• How we may expect an item to appear

12
VISUAL PERSPECTIVE

• How we may expect an item to appear

13
VISUAL PERSPECTIVE

• How we may expect an item to appear

14
VISUAL PERSPECTIVE
How the item may appear in in the field
15
VISUAL PERSPECTIVE

• How the item may appear in in the field

16
VISUAL PERSPECTIVE

• How the item may appear in in the field

17
BARRIERS TO CLUE DETECTION

• Searchers underestimate the difficulty of the
  task
• They may confuse seeing with detecting
• They may expect clues to visually pop-out.
• Visual clues wont pop out unless there is
  contrast such as edge definition

18
DETECTION IS DIFFICULT WHEN..

• The target is outside of central vision
• There is poor contrast in color and texture from
  the targets background
• Contours of the target are disguised or
  camouflaged
• The searcher does not have a good idea of what
  the target looks like

19
CAMOUFLAGE

• The purpose of disruptive coloration (camouflage)
  is not to prevent detection, but to delay it.
• Loss of contrast and edges

20
THE SEARCH IMAGE

• A mental picture of the target
• Learned through perceptual experience
• May guide visual search
• What you expect or hope to see

21
DWELL TIME

• Defined as the length of time a searcher looks at
  a specific spot
• Affected by the searchers expectations
• Terminated by a decision that a target is not
  there to be detected

22
THE VISUAL BRIEFING

• Face searchers away from the briefer
• Place assorted clues on the ground
• Have searchers detect them
• Then place them from different perspectives

23
SEARCHER EXPECTANCY

• The subjective probability of detecting a clue
• Affects scanning rate
• Affects follow-up inspections
• Affects dwell time

24
GOOD CLUES

• Are relatively large
• Have high contrast from their background
• Are familiar to the searcher

25
SEARCHING FOR MULTIPLE TARGETS

• Is more difficult than searching for a single
  target or clue
• Improves with practice
• Improves with printing

26
CONCLUSIONS

• Detecting a clue is much more difficult than just
  seeing
• The clue must be detected
• There are many more misses when people believe
  there is a low probability of detecting a target
• Form recognition occurs in central vision

27
CONCLUSIONS CONT.

• In a complex visual field, form is rarely
  detected on the first pass
• While scanning the searcher will have to return
  their gaze back to the object in order to
  perceive it.
• The threshold for form recognition will depend
  upon the match between the retinal image and what
  the perceiver expects to see

28
CONCLUSIONS CONT.

• Forms are more difficult to recognize when
• They are outside of their usual context
• They are positioned differently from their ideal
  perspective (canonical view)
• The searcher does not have a clear idea of what
  the object should look like (search image)