SIMULATING THE GRASSFIRE TRANSFORM USING AN ACTIVE CONTOUR MODEL

被引:169
作者
LEYMARIE, F [1 ]
LEVINE, MD [1 ]
机构
[1] MCGILL UNIV,DEPT ELECT ENGN,MONTREAL H3A 2A7,QUEBEC,CANADA
关键词
ACTIVE CONTOUR MODEL; CONTOUR AND REGION SHAPE FEATURES; DEFORMABLE SKELETON; DISTANCE SURFACE; EUCLIDEAN SKELETON; MULTISCALE REPRESENTATION; RIDGE SUPPORT; 2-D GRASSFIRE SIMULATION;
D O I
10.1109/34.107013
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
In this paper, we present a new method for shape description of planar objects that integrates both region and boundary features. Our method is an implementation of a 2-D dynamic grassfire that relies on a distance surface on which elastic contours minimize an energy function. A Euclidean distance transform combined with an active contour model, such as the snake, is used for this minimization process. Boundary information is integrated into the model by the extraction of curvature extrema and arcs of constant curvature. The use of an active contour on a field of grass, represented as a distance surface, combined with the curvature features of the boundary permits us to extract a Euclidean skeleton representation of the shape while bypassing many of the discretization problems found in other skeletonization algorithms. We propose a new concept for skeletal branch significance based on the notion of the local deformation introduced by symmetry points on the distance surface. We call this the ridge support. Furthermore, we show how the ridge support can be evaluated in terms of the velocity of formation of a symmetric axis or in terms of the slope amplitude of a tangent to the symmetric axis. We propose a new concept (the deformable skeleton), which is useful for tracking deformable shapes. We refer to this as the dynamic skeleton.
引用
收藏
页码:56 / 75
页数:20
相关论文
共 86 条
[1]   A NEW ALGORITHM FOR EXTRACTING THE INTERIOR OF BOUNDED REGIONS BASED ON CHAIN CODING [J].
ALI, SM ;
BURGE, RE .
COMPUTER VISION GRAPHICS AND IMAGE PROCESSING, 1988, 43 (02) :256-264
[2]  
[Anonymous], 1982, IMAGE ANAL MATH MORP
[3]   A WIDTH-INDEPENDENT FAST THINNING ALGORITHM [J].
ARCELLI, C ;
DIBAJA, GS .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1985, 7 (04) :463-474
[4]   FROM LOCAL MAXIMA TO CONNECTED SKELETONS [J].
ARCELLI, C ;
CORDELLA, LP ;
LEVIALDI, S .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1981, 3 (02) :134-143
[5]   A CONTOUR CHARACTERIZATION FOR MULTIPLY CONNECTED FIGURES [J].
ARCELLI, C ;
DIBAJA, GS .
PATTERN RECOGNITION LETTERS, 1987, 6 (04) :245-249
[6]  
ARCELLI C, 1978, IEEE T SYST MAN CYB, V8, P139
[7]  
Arcelli C., 1986, Eighth International Conference on Pattern Recognition. Proceedings (Cat. No.86CH2342-4), P283
[8]  
Arcelli C., 1986, Image Analysis and Processing. Proceedings of the Third International Conference, P137
[9]   A ONE-PASS 2-OPERATION PROCESS TO DETECT THE SKELETAL PIXELS ON THE 4-DISTANCE TRANSFORM [J].
ARCELLI, C ;
DIBAJA, GS .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1989, 11 (04) :411-414
[10]   PATTERN THINNING BY CONTOUR TRACING [J].
ARCELLI, C .
COMPUTER GRAPHICS AND IMAGE PROCESSING, 1981, 17 (02) :130-144