masters mini presentation
-
Upload
kartik-t-padmanabhan -
Category
Documents
-
view
125 -
download
0
Transcript of masters mini presentation
1
How Much Reality? Comfortable 3-D viaSingle Lens Stereo
Kartik Thumbavanam PadmanabhanAdvisor : Prof. Mel Siegel
2
How do we humans perceive depth?
How do we see the world in 3-D ?Principle of stereoscopy
3
Depth Cues
Binocular Parallax Convergence and AccommodationPrimary Cue – Binocular DisparityOther Cues
OcclusionRelative SizeSuperposition
4
Binocular Parallax"apparent motion" or a “perspective shift”
5
ConvergenceSimultaneous inward movement of both eyes toward each other
6
Accommodation
Expansion and contraction of ciliarymuscles
7
Primary Cue-Binocular Disparity
8
Is More Disparity Good ?
Conventional Binocular StereoscopyHigh Intraocular separation – High disparity
IssuesTime to fuseEye strain
Disparity α 1/(Ease Of Fusion)
9
How much disparity is just enough ?
How much intraocular separations?Big enough to stimulate stereopsisAllow strain free viewing
Intraocular separations of 1-3mm of human 60-65mm
10
Solution :Single Lens Stereoscopy
Light striking different sub-regions of aperture are differentDifferentiate two views in a single optic path
11
Techniques Implemented
Single Optic PathSpatial Domain
Anaglyphic StereoscopyTemporal and Spatial Domain
Light Valve Technique
12
Anaglyphically Encoded Single Lens
StereoscopyColor Filters
Subtractive FilteringRed/Blue color encoded viewsDecoded by Anaglyph glasses
13
Relative Disparity
14
Why is this technique not preferred ?
Increased Eye-channel crosstalkLack of brightnessRetinal RivalryColor Bombardment
15
Single Lens Stereo via Light ValvesLCD Shutters
http://electronics.howstuffworks.com/lcd3.htm
16
Experimental Setup12V DC
120V AC
Controller Box
Camera
TV
17
Advantages
More LightLess Cross-talkNo retinal rivalryNo color bombardment
18
Experiments
Goal of experimentsExperiments
Quantitave Comparison of the study of Binocular Stereo vs. Single Lens
Eye StrainDepth PerceptionTime of Fusion
19
Nature Of Experiments
Involving less learningLess human brain-eye co-ordinationHelpful in predicting the depth cuesTo validate the single-lens stereo in more quantitave mannerExperiments performed using brainteasers, loop-trapping puzzles, Rubix cube and Pick and Place experiments
20
Experimented Conditions:Each of the above 4 experiments are performed under 5 different conditions with total of 18 subjects:
• Direct Vision Two Eyes • Direct Vision One Eye • Direct Vision Monoscopic • Binocular Stereo • Single-Lens Stereo
21
ScreenshotsNo Stereo
Binocular Stereo Single-Lens Stereo
22
Observations:Eye Strain: Less Stressful Direct Vision Two Eyes <Direct Vision One Eye < Single Lens Stereo < Direct Vision Monoscopic < Binocular Stereo More StressfulPerspective distortion was noticed frequently when the experiments were done near camera.Motion parallax was noticed in pick and place operation when the part of background was made moving. (noticed very recently)
23
Observations…contd.The standard deviations obtained in any case was less whenever there was a ‘trick’ associated with the experiment irrespective of conditions. There was a little learning associated to Pick and place and Rubix cube experiment and more learning in case of brainteasers.Overall spread of standard deviations were in the order: Direct Vision Two Eyes < Direct Vision One Eye < Single Lens Stereo < Direct Vision Monoscopic < Binocular Stereo
24
Conclusions from experiments
Very apparent viewing comfort with Single Lens Stereo with no compromise on depth perceptionTime to fuse is less for Single Lens StereoFocusing closer to the object enhances depth perceptionDepth cues are exploited in a quantitative way.
25
Thank you