image search using deformable contours by : preeyakorn tipwai advisor : suthep madarasmi, ph.d...
TRANSCRIPT
Image Search Using Deformable Contours
By : Preeyakorn Tipwai
Advisor : Suthep Madarasmi, Ph.D
Computer Vision Laboratoy, Computer Engineering Department
King Mongkut’s University of Technology Thonburi
Methodology
InspirationJain et al [1] , “Object Matchin
g Using Deformable Templates”
Our Methodogy Finding Hypotheses : MGH
T Peak Clustering :
Watershed Method Contour Matching : Smo
oth Membrane Fitting
Preprocessing
Given a sketched template Find tangent direction
Given a target image Calculte edge map : Canny Edge Detection Find tangent direction
MGHTMGHT2
r
L,
xc,yc
= 0
A line at the contour edge is extended in the direction until it meets the other end of the contour
r1, 1, 1, l1 r2, 2, 2, l2 r3, 3, 3, l3 r4, 4, 4, l4
0...19 15,180,195,99 9,179,219,101 8,177,216,102 9,176,198,100
20...39 17,160,23,5 14,159,38,7 18,161,175,62 15,162,195,95
30…49 19,165,31,53 20,170,8,52 22,167,15,52 18,159,158,12
… … … … …
340...359
23,105,346,11 24,103,165,11 21,102,346,18 22,104,195,24
R-Table
MGHTMGHT
θθβ c
L
L S c
Rotation Factor:
Scaling Factor :
New ref. Point :
xc = x + S r cos (
yc = y + S r sin (
Watershed for Peak Clustering
1. Shed, by labeling, at the first level, calculate peaks of each label2. Increase to higher level, shed again
2.1 Meet an area of previous level, shed to that area2.2 Not meet any area of previous level, make a new area ,
calculate a new peak
Grid Matching : Data and Smoothness ConstraintsInter-grid Matching: Consistency between adjacent grids
Coarse and Fine Matching
Coarse and Fine Matching
Inter-grid Matching: Example
label at coarser level label at finer level
-1
0
1
-4, -3 or –2
-1, 0 or 1
2, 3 or 4
Matching Algorithm
Update (u,v) : Gibbs Sampler with simulated annealing to minimize energy function
Template
Target Edge
Conclusion
A deformation model
Contour Matching
A method for image search
Future work: large image database, efficient method for minimizing energy, coarse-and-fine approach to computer vison modules
Similarity Retrieval Effectiveness
0
20
40
60
80
100
120
#1 #2 #3 #4 #5
Agreement
Average
0
20
40
60
80
100
#1 #2 #3 #4
0
20
40
60
80
100
#1 #2 #3 #4
circle shapecircle shape heart shapeheart shape bulb shapebulb shape
max : 100, min : 96ave : 98
max : 100, min : 92ave : 95
max : 96, min : 8ave : 75
Experimental Result
Template Target Edge Hypotheses
1.293034 1.452130 3.364521 4.4185782
Threshold : 1.5-3.2