evaluation of the stability of sift keypoint correspondence across cameras
DESCRIPTION
Evaluation of the stability of SIFT keypoint correspondence across cameras. or .. “can we put a ‘C’ in SIFT?” max van kleek 6.869: learning and interfaces thursday may 11, 2005. ubiquitous computing: computers (and cameras) are everywhere!. little sister: follow-me-around user modeling. - PowerPoint PPT PresentationTRANSCRIPT
Evaluation of the stability of SIFT keypoint correspondence across cameras
or .. “can we put a ‘C’ in SIFT?”
max van kleek6.869: learning and interfacesthursday may 11, 2005
ubiquitous computing: computers (and cameras) are everywhere!
little sister: follow-me-around user modeling
object correspondence acrossvarying cameras/lighting/scenes is an important subgoal
applications: buliding a personal life log for yourself, interest profiling, social network mining, health care
identifying objects with local features: the SIFT transform
Orientation histogram = SIFT feature vector
recognizing objects
identified and orientedkeypoints used to votefor pose orientationsin a hough transform
pose-and-scale space
keypoint correspondence
1. % keypoints detected2. stability of orientation histogram
object orientation (away from frontal parallel)
object deformation
lighting direction, intensity, shading
lens distortion, sharpness, ccd “quality”, noise, capture artifacts
Mikolajczyk, K., C. Schmidt “A Performance Evaluation of Local Descriptors”, CVPR ‘03
Lowe, D.G. “Distinctive Image Features fromScale-Invariant Keypoints”, ICJV 2004.
Me! well, sort of..
???
cameras vary widely in sizes, configurations, capabilities, and prices
the experiment
Logitech QC Express
640x48016-bit colorYUV4:2:2AGC, Auto Exposuremanual focusUSB iface$15
Logitech QC Pro 3000
CCD by Phillips640x48016-bit colorYUV4:2:2, RGBAGC, Auto ExposureAuto WBmanual focusUSB iface$50
Sony EVI-D30Steerable NTSCcamera,DV capture card
720x480 luminance,less for colorRaw DV AGC, Auto ExposureAuto focus
~$300 + $200
Nikon Coolpix 990Digital still camera,
2048x1536RGBAuto Gain, Auto WB, Auto ExposureAuto focus
$1000 -> $500
experiment setup:5 incandenscent lights12 ft between camera and subject
acquiring image sets
background10 images stationary
2 front 2 face right 2 face left
for each camera:
= 16 images/cam * 4 cameras = 48
320x240 (or standard, and downsampled afterwards) RGB colorspace; jpeg quality 100;default camera settings except disabled AGC, disabled AE (locked to optimal settings)
algorithm for keypoint correspondence
source image contrast stretch over whole set
contrast-stretchedbackground images
background model(mean)
-
find(p(img) < epsilon)
compute_sift_points
dilate fg mask with a disc strel
intersectionfilter out bgkey points
keeping onlyrelevant keypointsby intersecting keypoints with foreground pts
image A w/ sift keypoints
orientation histograms for each keypoint in A
orientation histograms for each keypoint in B
image B w/ sift keypoints
orientation histograms for each keypoint in A
orientation histograms for each keypoint in B
match keypoints using nearest-neighbor in SIFT space
orientation histograms for each keypoint in A
orientation histograms for each keypoint in B
match keypoints using nearest-neighbor in SIFT space
1.1 -> 1.2 1.2 -> 1.1
sanity check: same (dv) camera, slightly different pose 1.1: 15 keypoints detected1.2: 11 keypoints detected
6 properly assigned8 in common
5 properly assigned8 in common
4.1 -> 1.1 1.1 -> 4.1
nikon coolpix versus sony steerable 4.1: 18 keypoints detected1.1: 15 keypoints detected
5 properly assigned10 in common
5 properly assigned10 in common
4.1 -> 3.1 3.1 -> 4.1
nikon coolpix versus qc pro 4.1: 18 keypoints detected3.1: 17 keypoints detected
2 properly assigned10 in common
5 properly assigned10 in common
4.1 -> 2.1 4.1 -> 2.1
nikon coolpix versus qc express 4.1: 18 keypoints detected2.1: 22 keypoints detected
2 properly assigned10 in common
0 properly assigned10 in common
other results:
qc pro vs qc exp (3.1 -> 2.1) 3.1->2.1 : 17 / 22 4 correct out of 6 in common2.1->3.1: 22 / 17 0 correct out of 6 in common
poor reproducibility with qcs?
qc exp test (is the qc exp just too noisy?)2.1->2.2 : 22/20 1 correct out of 8 in common2.2->2.1 : 20/22 2 correct out of 8 in common
yes.
qc pro reproducibility3.1->3.2: 17/24 6 correct out of 9 in common3.2->3.1 : 24/17 7 correct out of 9 in common
angle test using qc pro3.1->3.4 : 0 correct out of 0 in common
sensitive to out-of-plane rotation
experiment setup:5 incandenscent lights3 ft between camera and robot
320x240 (or standard, and downsampled afterwards) RGB colorspace; jpeg quality 100;default camera settings except disabled AGC, (locked to optimal settings)
• parameters: – bins / histogram
– pixels / quadrant
– quadrants / keypoint
– gaussian dropoff covariance
keypoint splitting / multiple primary gradient directions
source keypoint merging
histogram blurring