## Asymptotically Optimal and Liner-time Algorithm for Polygonal Curve Simplification (2005)

### BibTeX

@MISC{Chen05asymptoticallyoptimal,

author = {Long Chen and James Z. Wang and Jinchao Xu and Long Chen and James Z. Wang and Jinchao Xu},

title = {Asymptotically Optimal and Liner-time Algorithm for Polygonal Curve Simplification},

year = {2005}

}

### OpenURL

### Abstract

In many application domains involving shapes and curves, polygonal curve simplification is an important part of the computer analysis processes. In this work, we have developed asymptotically optimal and linear-time algorithms to approximate a polygonal curve by another polygonal curve whose vertices are a subset of the vertices of the original one. The algorithm developed in this paper can be applied to a vector map data reduction in geographical information system especially for large-scale data. The error of the approximation is measured by the area of the domain bounded by the two polygonal curves. Based on the equidistribution principle and local refinement/coarsening strategy, an efficient

### Citations

1250 | Shape matching and object recognition using shape contexts
- Belongie, Malik, et al.
- 2002
(Show Context)
Citation Context ...l results are shown. Index Terms Curve simplification, optimal approximation, shape analysis. I. INTRODUCTION Discrete curves are commonly used in automated pattern analysis problems involving shapes =-=[1]-=-, [2], [3]. They are also used in large scale geo-spatial databases to represent land features. For instance, a large amount of storage used for digital maps consists of curves such as roads, rivers, ... |

828 |
Computer Vision
- Ballard, Brown
- 1982
(Show Context)
Citation Context ... which give acceptable quality in the literature [14], [15], [16], [17]. In practice, a widely used curve simplification algorithm is the Douglas-Peucker (D-P) algorithm [18], along with its variants =-=[19]-=-, [20], [21]. It is often used as a heuristic algorithm and can be implemented to run in time O(N log N) [21]. We summarize the quality and complexity of those algorithms in the following table. In th... |

628 | Constructive Approximation
- Devore, Lorentz
- 1993
(Show Context)
Citation Context ...on [0, l] and an h ∈ (0, l), we define the finite difference ∆h(f, x) = f(x + h) − f(x), for x ∈ Ah := [0, l − h]. November 18, 2005 DRAFTsDRAFT FOR IEEE 11 The moduli of smoothness of f ∈ L 1 (0, l) =-=[27]-=- is defined as ω(f, t)1 = sup �∆h(f, ·)�L 0≤h≤t 1 (Ah), t ≥ 0. Given a continuous curve C with end points v0, v1, let P = {v0, v1} be the line segment connecting v0 and v1. We assume the line segment ... |

443 |
Algorithms for the reduction of the number of points required to represent a digitized line or its caricature
- Douglas, Peucker
- 1973
(Show Context)
Citation Context ...e some other faster algorithms which give acceptable quality in the literature [14], [15], [16], [17]. In practice, a widely used curve simplification algorithm is the Douglas-Peucker (D-P) algorithm =-=[18]-=-, along with its variants [19], [20], [21]. It is often used as a heuristic algorithm and can be implemented to run in time O(N log N) [21]. We summarize the quality and complexity of those algorithms... |

394 | SIMPLIcity: Semantics-sensitive integrated matching for picture libraries - Wang, Li, et al. |

386 | Three-dimensional object recognition from single two-dimensional images
- Lowe
- 1987
(Show Context)
Citation Context ... give acceptable quality in the literature [14], [15], [16], [17]. In practice, a widely used curve simplification algorithm is the Douglas-Peucker (D-P) algorithm [18], along with its variants [19], =-=[20]-=-, [21]. It is often used as a heuristic algorithm and can be implemented to run in time O(N log N) [21]. We summarize the quality and complexity of those algorithms in the following table. In this pap... |

315 |
A Posteriori Error Estimation in Finite Element Analysis. Wiley-Interscience
- Ainsworth, Oden
- 2000
(Show Context)
Citation Context ...rithm Asymptotically optimal O(N) The main theoretical tool we are going to use is the grid adaptation method used in the numerical solution of partial differential equations (PDEs) [22], [23], [24], =-=[25]-=-, [26]. Basically there are two types of grid adaptation. One is to refine the parts where the error is large and to coarsen the parts where the error is small. For this method, it is crucial to get a... |

247 |
Interpolation of operators
- Bennett, Sharpley
- 1988
(Show Context)
Citation Context ...Θ H (N) , and FN is linear in (k, k + 1) for k = 0, .., N − 1. For another metric, let us list the error estimates below: δ H (C, P) ≤ ω(θ, l)1 and δ V (C, P) ≤ lω(θ, l)1. By the interpolation theory =-=[32]-=-, we can obtain estimate for other error metrics. For example, the bound l 1/2 ω(θ, l)1 is suitable for the L 2 -type error. The grid distribution function can be defined similarly. For digital curves... |

127 | R.: Shape similarity measure based on correspondence of visual parts
- Latecki, Lakmper
- 2000
(Show Context)
Citation Context ...are shown. Index Terms Curve simplification, optimal approximation, shape analysis. I. INTRODUCTION Discrete curves are commonly used in automated pattern analysis problems involving shapes [1], [2], =-=[3]-=-. They are also used in large scale geo-spatial databases to represent land features. For instance, a large amount of storage used for digital maps consists of curves such as roads, rivers, and highwa... |

70 | Convexity rule for shape decomposition based on discrete contour evolution
- Latecki, Lakämper
- 1999
(Show Context)
Citation Context ...ive line segments is replaced with a single line segment joining their end points. If no such point exists, the algorithm stops. It can also be thought of as a discrete version of the curve evolution =-=[28]-=-. In the local coarsening algorithm, the evaluation of the error is not a problem since we are using area metric. The error is the area of the triangle composed by vi−1, vi, and vi+1. Subroutine Local... |

63 |
F.: Approximation of polygonal curves with minimum number of line segments
- Chan, Chin
- 1992
(Show Context)
Citation Context ...the optimal curve simplification problem is to find the shortest path in the direction weighted graph. The algorithm presented in [8] requires O(N 2 log N) running time. It was improved to O(N 2 ) in =-=[9]-=-. A recent algorithm [10] achieves time O(N 4/3+δ ) for any δ > 0. Another approach, utilizing dynamic programming, finds the solution of the problem recursively using the results of problems in small... |

56 | Techniques for assessing polygonal approximation of curves
- Rosin
- 1997
(Show Context)
Citation Context ...tes of the two end points are given. Thus curve simplification algorithms are necessary to represent a curve by using less coordinate information, while preserving the perceptual characteristics [4], =-=[5]-=-. Polygonal curve simplification has the potential to significantly reduce not only the storage space, but also the plotting time and resource usage on battery-powered hand-held devices such as Global... |

44 | MPEG-4 and ratedistortion-based shape-coding techniques
- Katsaggelos, Kondi, et al.
- 1998
(Show Context)
Citation Context ...c information systems (GIS), computer graphics, digital cartography, pattern recognition, shape-based image database retrieval, and data compression November 18, 2005 DRAFTsDRAFT FOR IEEE 3 [4], [6], =-=[7]-=-. In this paper, we provide mathematical analysis of algorithms on the polygonal curve simplification and further develop more efficient algorithms. A. Problem Definition To simplify the discussions, ... |

44 |
Good approximations by splines with variable knots II. Springer Lecture note series
- Boor
- 1973
(Show Context)
Citation Context ...N 4/3+δ ) Our algorithm Asymptotically optimal O(N) The main theoretical tool we are going to use is the grid adaptation method used in the numerical solution of partial differential equations (PDEs) =-=[22]-=-, [23], [24], [25], [26]. Basically there are two types of grid adaptation. One is to refine the parts where the error is large and to coarsen the parts where the error is small. For this method, it i... |

42 | Near-linear time approximation algorithms for curve simplification
- Agarwal, Har-Peled, et al.
(Show Context)
Citation Context ... discarding the rest. This algorithm is very fast, but sometimes yields poor quality approximations. There are some other faster algorithms which give acceptable quality in the literature [14], [15], =-=[16]-=-, [17]. In practice, a widely used curve simplification algorithm is the Douglas-Peucker (D-P) algorithm [18], along with its variants [19], [20], [21]. It is often used as a heuristic algorithm and c... |

40 |
M.: Polygonal approximations of a curve-formulations and algorithms
- Imai, Iri
- 1988
(Show Context)
Citation Context ... One unified approach for the optimal curve simplification problems is to formulate polygonal November 18, 2005 DRAFT v3 � � v4 � v5sDRAFT FOR IEEE 5 curve simplification in terms of the graph theory =-=[8]-=-. Graph-based algorithms create a graph by connecting the vertices of PN with directional edges. Each edge is assigned a weight related to the error metric. Then the optimal curve simplification probl... |

38 | K.R.: Efficient algorithms for approximating polygonal chains
- Agarwal, Varadarajan
- 2000
(Show Context)
Citation Context ...fication problem is to find the shortest path in the direction weighted graph. The algorithm presented in [8] requires O(N 2 log N) running time. It was improved to O(N 2 ) in [9]. A recent algorithm =-=[10]-=- achieves time O(N 4/3+δ ) for any δ > 0. Another approach, utilizing dynamic programming, finds the solution of the problem recursively using the results of problems in smaller size [11]. It can be i... |

36 |
M.: An optimal algorithm for approximating a piecewise linear function
- Imai, Iri
- 1986
(Show Context)
Citation Context ...ger k, discarding the rest. This algorithm is very fast, but sometimes yields poor quality approximations. There are some other faster algorithms which give acceptable quality in the literature [14], =-=[15]-=-, [16], [17]. In practice, a widely used curve simplification algorithm is the Douglas-Peucker (D-P) algorithm [18], along with its variants [19], [20], [21]. It is often used as a heuristic algorithm... |

35 |
Optimum polygonal approximation of digitized curves
- Perez, Vidal
- 1994
(Show Context)
Citation Context ...nt algorithm [10] achieves time O(N 4/3+δ ) for any δ > 0. Another approach, utilizing dynamic programming, finds the solution of the problem recursively using the results of problems in smaller size =-=[11]-=-. It can be implemented with O(N 2 ) time [12]. Recently an iterative reduced-search dynamic programming approach, proposed in [13], reduces the complexity to O(N 2 /M). Due to the high computational ... |

25 |
Automated line generalization
- McMaster
(Show Context)
Citation Context ...rdinates of the two end points are given. Thus curve simplification algorithms are necessary to represent a curve by using less coordinate information, while preserving the perceptual characteristics =-=[4]-=-, [5]. Polygonal curve simplification has the potential to significantly reduce not only the storage space, but also the plotting time and resource usage on battery-powered hand-held devices such as G... |

23 |
Feedback and adaptive finite element solution of one dimensional boundary value problems, Numerische Mathematik 44
- Babu§ka, Vogelius
- 1984
(Show Context)
Citation Context ...r algorithm Asymptotically optimal O(N) The main theoretical tool we are going to use is the grid adaptation method used in the numerical solution of partial differential equations (PDEs) [22], [23], =-=[24]-=-, [25], [26]. Basically there are two types of grid adaptation. One is to refine the parts where the error is large and to coarsen the parts where the error is small. For this method, it is crucial to... |

20 |
An O(nlogn) implementation of the douglas-peucker algorithm for line simplification
- Hershberger, Snoeyink
- 1994
(Show Context)
Citation Context ...acceptable quality in the literature [14], [15], [16], [17]. In practice, a widely used curve simplification algorithm is the Douglas-Peucker (D-P) algorithm [18], along with its variants [19], [20], =-=[21]-=-. It is often used as a heuristic algorithm and can be implemented to run in time O(N log N) [21]. We summarize the quality and complexity of those algorithms in the following table. In this paper, we... |

19 |
On detecting dominant points
- Ansari, Delp
(Show Context)
Citation Context ...e integer k, discarding the rest. This algorithm is very fast, but sometimes yields poor quality approximations. There are some other faster algorithms which give acceptable quality in the literature =-=[14]-=-, [15], [16], [17]. In practice, a widely used curve simplification algorithm is the Douglas-Peucker (D-P) algorithm [18], along with its variants [19], [20], [21]. It is often used as a heuristic alg... |

15 |
An efficient algorithm for the optimal polygonal approximation of digitized curves
- Salotti
(Show Context)
Citation Context ...r any δ > 0. Another approach, utilizing dynamic programming, finds the solution of the problem recursively using the results of problems in smaller size [11]. It can be implemented with O(N 2 ) time =-=[12]-=-. Recently an iterative reduced-search dynamic programming approach, proposed in [13], reduces the complexity to O(N 2 /M). Due to the high computational complexity, these methods are limited to relat... |

12 |
Optimal anisotropic simplicial meshes for minimizing interpolation errors in L p -norm
- Chen, Sun, et al.
- 2007
(Show Context)
Citation Context ...ree knots piecewise linear approximation in the approximation theory, which is well studied in the literature [22], [29], [27]. We shall present an asymptotic analysis based on our recent approach in =-=[30]-=-. The following theorem shows that an asymptotically optimal approximation is obtained by equidistributing the error. Lemma 2. �f − fI�L 1 (0,1) ≥ 1 12M 2 � � 1 0 |f ′′ (x)| 1/3 dx � 3 + O(h 3 ), wher... |

8 | Reduced-search dynamic programming for approximation of polygonal curves
- Kolesnikov, Fränti
- 2003
(Show Context)
Citation Context ...he problem recursively using the results of problems in smaller size [11]. It can be implemented with O(N 2 ) time [12]. Recently an iterative reduced-search dynamic programming approach, proposed in =-=[13]-=-, reduces the complexity to O(N 2 /M). Due to the high computational complexity, these methods are limited to relatively small values of N when less powerful machines are used. In applications such as... |

8 |
Fast polygonal approximation of digital curves
- Debled-Rennesson, Tabbone, et al.
- 2004
(Show Context)
Citation Context ...rding the rest. This algorithm is very fast, but sometimes yields poor quality approximations. There are some other faster algorithms which give acceptable quality in the literature [14], [15], [16], =-=[17]-=-. In practice, a widely used curve simplification algorithm is the Douglas-Peucker (D-P) algorithm [18], along with its variants [19], [20], [21]. It is often used as a heuristic algorithm and can be ... |

5 |
A posteriori error error estimates for the finite element method
- Babuˇska, Rheinboldt
- 1978
(Show Context)
Citation Context ...δ ) Our algorithm Asymptotically optimal O(N) The main theoretical tool we are going to use is the grid adaptation method used in the numerical solution of partial differential equations (PDEs) [22], =-=[23]-=-, [24], [25], [26]. Basically there are two types of grid adaptation. One is to refine the parts where the error is large and to coarsen the parts where the error is small. For this method, it is cruc... |

4 |
Nonlinear segmented function approximation and analysis of line patterns
- McClure
- 1975
(Show Context)
Citation Context ...proximation based on this grid minimizes the error �f − fI�L 1 (0,1). It is called free knots piecewise linear approximation in the approximation theory, which is well studied in the literature [22], =-=[29]-=-, [27]. We shall present an asymptotic analysis based on our recent approach in [30]. The following theorem shows that an asymptotically optimal approximation is obtained by equidistributing the error... |

3 |
Adaptive approximation of curves. Approximation theory: a volume dedicated to Borislav Bojanov, 43–57
- Binev, Dahmen, et al.
- 2004
(Show Context)
Citation Context ...Asymptotically optimal O(N) The main theoretical tool we are going to use is the grid adaptation method used in the numerical solution of partial differential equations (PDEs) [22], [23], [24], [25], =-=[26]-=-. Basically there are two types of grid adaptation. One is to refine the parts where the error is large and to coarsen the parts where the error is small. For this method, it is crucial to get an effi... |

3 |
Rational approximation of real valued functions
- Petrushev, Popov
- 1987
(Show Context)
Citation Context ...as the Hausdorff metric δ H (PN, PM) = max{max y∈PN min x∈PM where �y − x� is the Euclidean distance in R 2 . �y − x�, max min �y − x�} x∈PM y∈PN November 18, 2005 DRAFTsDRAFT FOR IEEE 20 In the book =-=[31]-=- (page 288), Petrushev and Popov obtained an error indicator for the Hausdorff error (see also [26]). Theorem 3. [Petrushev and Popov 1987] Let C be a curve such that θ function is well defined. Let P... |