COMPUTATIONAL NEUROSCIENCE FINAL PROJECT – DEPTH VISION Omri Perez 2013.

Slides:



Advertisements
Similar presentations
Seeing 3D from 2D Images. How to make a 2D image appear as 3D! ► Output and input is typically 2D Images ► Yet we want to show a 3D world! ► How can we.
Advertisements

November 12, 2013Computer Vision Lecture 12: Texture 1Signature Another popular method of representing shape is called the signature. In order to compute.
EXAM 2 !!! Monday, Tuesday, Wednesday, Thursday of NEXT WEEK.
CS 376b Introduction to Computer Vision 04 / 21 / 2008 Instructor: Michael Eckmann.
Space Perception and Binocular Vision
Depth Cues Pictorial Depth Cues: aspects of 2D images that imply depth
Extra Credit for Participating in Experiments Go to to sign upwww.tatalab.ca We are recruiting for a variety of experiments including basic.
Chapter 8: Vision in three dimensions Basic issue: How do we construct a three-dimension visual experience from two- dimensional visual input? Important.
Midterm 2 is March 11th and 12th Read Land article for March 5th.
Fundraiser Concert for Haiti U of L Students and Faculty Performing –9pm Blues Sen-sa-shun –10pm Shawna Romolliwa –11pm Dave Renter At The Slice, Friday.
Chapter 6 Opener. Figure 6.1 The Euclidean geometry of the three-dimensional world turns into something quite different on the curved, two-dimensional.
Binocular Disparity points (C) nearer than fixation (P) have crossed disparity points (F) farther than fixation have uncrossed disparity.
The Apparatus. Seeing in Stereo It’s very hard to read words if there are multiple images on your retina.
Midterm 1 Mean 71%. Detection and Loudness For example, tones of different frequencies that are judged to be equally loud have different SPLs (dB)
Psyc2320 Midterm II Review. Physiological Depth Cues – Accommodation.
Exam next week Covers everything about all sensory modalities except hearing This includes: vision balance/touch/taste/smell/ proprioception/theroception.
Fundraiser Concert for Haiti U of L Students and Faculty Performing –9pm Blues Sen-sa-shun –10pm Shawna Romolliwa –11pm Dave Renter At The Slice, Friday.
Read Pinker article for Thurs.. Seeing in Stereo.
3/23/2005 © Dr. Zachary Wartell 1 Depth and Size Perception.
Imaging Science FundamentalsChester F. Carlson Center for Imaging Science Binocular Vision and The Perception of Depth.
Motion Depth Cues – Motion 1. Parallax. Motion Depth Cues – Parallax.
Introduction to Computer Vision 3D Vision Topic 9 Stereo Vision (I) CMPSCI 591A/691A CMPSCI 570/670.
Today: Finish Gregory Discussion, finish stereograms Tuesday: Start colour vision Thursday: Finish colour vision - Read LAND for Thursday.
Stereograms seeing depth requires “only” two different images on the retina.
Binocular Disparity points nearer than horopter have crossed disparity
Stereoscopic Depth Disparity between the two retinal images indicates an objects distance from the plane of fixation.
Midterm 1: Mean 73.68% To go over your test, send to Arif
Infinity of Interpretations There are an infinite number of interpretations of the 2D pattern of light on the retina.
Sensation and Perception - depth.ppt © 2001 Dr. Laura Snodgrass, Ph.D. Depth Perception Four theoretical approaches –Cue theory unconscious calculation.
Monocular vs. Binocular View Monocular view: one eye only! Many optical instruments are designed from one eye view. Binocular view: two eyes with each.
Reading Gregory 24 th Pinker 26 th. Seeing Depth What’s the big problem with seeing depth ?
1B50 – Percepts and Concepts Daniel J Hulme. Outline Cognitive Vision –Why do we want computers to see? –Why can’t computers see? –Introducing percepts.
PSYC 330: Perception Depth Perception. The Puzzle The “Real” World and Euclidean Geometry The Retinal World and Projective Geometry Anamorphic art.
Careers for Psychology and Neuroscience Majors Oct. 19th5-7pm in SU 300 Ballroom B.
CAP4730: Computational Structures in Computer Graphics 3D Concepts.
Binocular Vision, Fusion, and Accommodation
Chapter 5 Human Stereopsis, Fusion, and Stereoscopic Virtual Environments.
December 4, 2014Computer Vision Lecture 22: Depth 1 Stereo Vision Comparing the similar triangles PMC l and p l LC l, we get: Similarly, for PNC r and.
1 Computational Vision CSCI 363, Fall 2012 Lecture 20 Stereo, Motion.
1 Perception, Illusion and VR HNRS 299, Spring 2008 Lecture 8 Seeing Depth.
Physiological Depth Cues – Convergence. Physiological Depth Cues – Convergence – small angle of convergence = far away – large angle of convergence =
CS332 Visual Processing Department of Computer Science Wellesley College Binocular Stereo Vision Region-based stereo matching algorithms Properties of.
CSE 185 Introduction to Computer Vision Stereo. Taken at the same time or sequential in time stereo vision structure from motion optical flow Multiple.
Autostereograms Convergence is on a point at the same distance as the images Boxes and faces are on the horopter How many boxes would you see? boxes and.
Depth Perception Kimberley A. Clow
Chapter 10: Perceiving Depth and Size
October 16, 2014Computer Vision Lecture 12: Image Segmentation II 1 Hough Transform The Hough transform is a very general technique for feature detection.
How Far Away Is It? Depth Perception
Outline Of Today’s Discussion 1.Monocular & Binocular Depth Cues: Understanding Retinal Disparity.
Perception and VR MONT 104S, Fall 2008 Lecture 8 Seeing Depth
Outline Of Today’s Discussion 1.Some Disparities are Not Retinal: Pulfrich Effect 2.Random-Dot Stereograms 3.Binocular Rivalry 4.Motion Parallax.
1 Computational Vision CSCI 363, Fall 2012 Lecture 16 Stereopsis.
How do we see in 3 dimensions?
Binocular Disparity points nearer than horopter have crossed disparity points farther than horopter have uncrossed disparity.
1 Computational Vision CSCI 363, Fall 2012 Lecture 18 Stereopsis III.
Computational Vision CSCI 363, Fall 2012 Lecture 17 Stereopsis II
Perception of Depth. Cues to depth: 1. Oculomotor 2. Monocular 3. Binocular.
Independent Component Analysis features of Color & Stereo images Authors: Patrik O. Hoyer Aapo Hyvarinen CIS 526: Neural Computation Presented by: Ajay.
Careers for Psychology and Neuroscience Majors Oct. 19th5-7pm in SU 300 Ballroom B.
Summarized by Geb Thomas
Depth Perception, with Emphasis on Stereoscopic Vision
Computational Vision CSCI 363, Fall 2016 Lecture 15 Stereopsis
Space Perception and Binocular Vision
Binocular Stereo Vision
Binocular Stereo Vision
Binocular Stereo Vision
Stereopsis Current Biology
Binocular Disparity and the Perception of Depth
Receptive Fields of Disparity-Tuned Simple Cells in Macaque V1
Depth Perception.
Presentation transcript:

COMPUTATIONAL NEUROSCIENCE FINAL PROJECT – DEPTH VISION Omri Perez 2013

 Pictorial Depth Cues  Physiological Depth Cues  Motion Parallax  Stereoscopic Depth Cues

Two Physiological Depth Cues: 1. Accommodation 2. Convergence

– Accommodation

Accommodation  relaxed lens = far away  accommodating lens = near What must the visual system be able to compute unconsciously?

– Convergence

Convergence  small angle of convergence = far away  large angle of convergence = near – What two sensory systems is the brain integrating? – What happens to images closer or farther away from fixation point?

Parallax

– Parallax  Points at different locations in the visual field move at different speeds depending on their distance from fixation  y27Gk&NR=1 y27Gk&NR=1

Seeing in Stereo

It’s very hard to read words if there are multiple images on your retina

But how many images are there on your retinae?

 Your eyes have a different image on each retina  hold pen at arms length and fixate the spot  how many pens do you see?  which pen matches which eye?

 Your eyes have a different image on each retina  now fixate the pen  how many spots do you see?  which spot matches which eye?

 Binocular disparity is the difference between the two images

 Disparity depends on where the object is relative to the fixation point:  objects closer than fixation project images that “cross”  objects farther than fixation project images that do not “cross”

 Corresponding retinal points

 Points in space that have corresponding retinal points define a plane called the horopter or Panum’s fusional area The Horopter

 Points not on the horopter will be disparate on the retina (they project images onto non-corresponding points)

 The nature of the disparity depends on where they are relative to the horopter

 points nearer than horopter have crossed disparity  points farther than horopter have uncrossed disparity

 Why don’t we see double vision?

 Images with a small enough disparity are fused into a single image

 Why don’t we see double vision?  Images with a small enough disparity are fused into a single image  The region of space that contains images with close enough disparity to be fused is called Panum’s Area

 Panum’s Area extends just in front of and just behind the horopter

 Our brains interpret crossed and uncrossed disparity as depth  That process is called stereoscopic depth perception or simply stereopsis

 Stereopsis requires that the brain can encode the two retinal images independently

 Primary visual cortex (V1) has bands of neurons that keep input from the two eyes separate

 The basic processing unit of depth perception  The cortical column consists of a complete set of orientation columns over a cycle of 180º and of right and left dominance columns in the visual cortex. A hypercolumn may be about 1 mm wide.

 To compute the binocular depth of stereo images Left Right

 To emulate the receptive fields of V1 neurons we use the Gabor function.  Even symmetry  Odd symmetry Sinus.* 2D Gaussian  Gabor

 We filter by doing a 2D convolution of the filters with the image. The different results are averaged together. Left Right

2D cross correlation (xcorr2) Tip: In most cases, peak cross correlation results in the x axis (columns) between the left and right eye should only be positive!

 1. maximum of cross correlation  2. First neuron to fire in a 2D LIF array (winner take all). The input is the cross correlation result.  3. Population vector of 2D LIF array after X simulation steps. Same input. Notice the horizontal smearing in 2 and 3 is because of cross over activity when switching patches

1. Load a pair of stereo images. You can use the supplied function image_loader.m which makes sure the image is in grayscale and has a proper dynamic range. 2. Generate the filters. You can use the supplied function generate_filters.m to generate the array of filters. I urge you to try out different sizes (3 rd parameter) than the default ones in the function. 3. Filter each of the two images using the filters from the previous stage. You can use the function filter_with_all_filters.m to do this. 4. Now, using the two filtered images, iterate over patches, calculate the cross correlation matrix and determine the current depth using the methods 1-3 described in the previous slide. You can tweak the overlap of the patches to reduce computation time. Note for methods 2 and 3: a. you can use the supplied function LIF_one_step_multiple_neurons.m to simulate the LIF neurons b. For methods 2 and 3 it is wise to normalize the xcorr2 results, e.g. by dividing by the maximum value or some such normalization.

5. Incorporate all these into a function of the form: result = find_depth_with_LIF( Left_im_name,Right_im_name, method_num,patch_size, use_filters ) Where result is the matrix representing the estimated depth (pixel shifts), Left and Right_im_name are the names of the stereo images. method_num is the number of the method (1-3, see above). patch_size the ratio of the patch size to the image dimensions. E.g. in an image that is 640x480 a ratio of 1/15 will produce a patch of size ~43x32. Please note that in matlab the indexing convention is rowsxcols (and not x,y) so the image is actually 480x640. use_filters is a flag that determines whether to filter the images (step 3 in previous slide) before computing the depth map (useful for debugging, however should be set to true when generating the final results).  In addition to the supplied stereo image pairs, you should also generate a left and right random dot stereogram image pair using the supplied function RDS_generator.m together with a mask (I supplied you with an example mask, RDS_Pac-Man_mask.png)  You can find other stereo image pairs online, e.g.  Bonus: You can add a fourth depth estimation method of our choice. This can be something you read somewhere or an original idea. For example you can use one of the methods 1-3 but change the patch scan so it won’t be an orderly right to left then down one row and right to left. Instead it can be a random scan which, among other things, will cause several regions to be left uncalculated but other regions more tightly sampled.

1. Your code together with any images (regular and RDS) you used and the supplied images. 2. A document showing the depth results on the two supplied stereo image pairs and one RDS you generated, for each of the 3 methods. (If you chose to do the bonus then show the results for the bonus method as well). Don’t forget when showing results for the RDS to relate them to the mask used to generate it. The document should contain a concise explanation of what you did, your algorithms and interpretation of the results.

 The project should be submitted by mail to Omri.  Good Luck and a succesful test period (and vacation?) !!!