A Constrained Regularization Approach to Robust Corner Detection

Kwanghoon Sohn; Winser E. Alexander; Jung H. Kim; Wesley E. Snyder

doi:10.1109/21.293500

A Constrained Regularization Approach to Robust Corner Detection

Kwanghoon Sohn, Winser E. Alexander, Jung H. Kim, Wesley E. Snyder

Department of Electrical and Electronic Engineering

Research output: Contribution to journal › Article › peer-review

12 Citations (Scopus)

Abstract

This paper presents a method of optimal boundary smoothing for curvature estimation and a method of corner detection for consistent representation of objects for computer vision applications. The existing methods for curvature estimation have a common problem in determining a unique smoothing factor. We propose a constrained regularization (CR) approach to overcome that problem. The curvature function computed on the preprocessed boundary, which is obtained by the CR approach, gives consistent corner detection results. Ideal corners rarely exist for a real boundary. They are often rounded due to the smoothing effects of the preprocessing. In addition, a human recognizes both sharp corners and slightly rounded segments as corners. Hence, we establish a criterion, called “corner sharpness”, which is qualitatively similar to a human's capability to detect corners.

Original language	English
Pages (from-to)	820-828
Number of pages	9
Journal	IEEE Transactions on Systems, Man and Cybernetics
Volume	24
Issue number	5
DOIs	https://doi.org/10.1109/21.293500
Publication status	Published - 1994 May

Bibliographical note

Funding Information:
In the local methods, the boundaries need to be segmented in order to represent them analytically. In general, there are two possible approaches to the boundary segmentation problem. One is to detect comers through angle or comer detection schemes. It mainly depends on a curvature function. The other approach is to obtain a piecewise linear polygonal approximation of the digitized boundary subject to certain constraints on the goodness of fit. The vertices found in the linear polygonal approximation are usually called break points. However, this approach may result in different boundary segmentation Manuscript received August 5, 1992; revised June 23, 1993. This work was supported in part by the U.S. Army Research Office under Grant DAAL03-90-0913, in part by the National Science Foundation under Grant ECD-8212696, and in part by the Federal Aviation Administration under Contract DTF A01-87-C-00043. K. Sohn is with the Satellite Communications Division, Electronics and Telecommunications Research Institute, P.O. Box 8, Daeduk Science Town, Daejeon, 305-606, Korea. W. E. Alexander is with the Dept. of Electrical & Computer Engineering, North Carolina State University, Raleigh, NC 27695-791 1 USA. J. H. Kim is with the Dept. of Electrical Engineering, North Carolina A&T State University, Greensboro, NC 2741 1 USA. W. E. Snyder is with the Dept. of Radiology, Bowman Gray School of Medicine, Wake Forest University, Winston-Salem, NC 27157-1022, USA. IEEE Log Number 9400636.

All Science Journal Classification (ASJC) codes

Engineering(all)

Access to Document

10.1109/21.293500

Cite this

@article{794c0155f4634929929760c2124cbde5,

title = "A Constrained Regularization Approach to Robust Corner Detection",

abstract = "This paper presents a method of optimal boundary smoothing for curvature estimation and a method of corner detection for consistent representation of objects for computer vision applications. The existing methods for curvature estimation have a common problem in determining a unique smoothing factor. We propose a constrained regularization (CR) approach to overcome that problem. The curvature function computed on the preprocessed boundary, which is obtained by the CR approach, gives consistent corner detection results. Ideal corners rarely exist for a real boundary. They are often rounded due to the smoothing effects of the preprocessing. In addition, a human recognizes both sharp corners and slightly rounded segments as corners. Hence, we establish a criterion, called “corner sharpness”, which is qualitatively similar to a human's capability to detect corners.",

author = "Kwanghoon Sohn and Alexander, {Winser E.} and Kim, {Jung H.} and Snyder, {Wesley E.}",

note = "Funding Information: In the local methods, the boundaries need to be segmented in order to represent them analytically. In general, there are two possible approaches to the boundary segmentation problem. One is to detect comers through angle or comer detection schemes. It mainly depends on a curvature function. The other approach is to obtain a piecewise linear polygonal approximation of the digitized boundary subject to certain constraints on the goodness of fit. The vertices found in the linear polygonal approximation are usually called break points. However, this approach may result in different boundary segmentation Manuscript received August 5, 1992; revised June 23, 1993. This work was supported in part by the U.S. Army Research Office under Grant DAAL03-90-0913, in part by the National Science Foundation under Grant ECD-8212696, and in part by the Federal Aviation Administration under Contract DTF A01-87-C-00043. K. Sohn is with the Satellite Communications Division, Electronics and Telecommunications Research Institute, P.O. Box 8, Daeduk Science Town, Daejeon, 305-606, Korea. W. E. Alexander is with the Dept. of Electrical & Computer Engineering, North Carolina State University, Raleigh, NC 27695-791 1 USA. J. H. Kim is with the Dept. of Electrical Engineering, North Carolina A&T State University, Greensboro, NC 2741 1 USA. W. E. Snyder is with the Dept. of Radiology, Bowman Gray School of Medicine, Wake Forest University, Winston-Salem, NC 27157-1022, USA. IEEE Log Number 9400636.",

year = "1994",

month = may,

doi = "10.1109/21.293500",

language = "English",

volume = "24",

pages = "820--828",

journal = "IEEE Transactions on Systems, Man and Cybernetics",

issn = "0018-9472",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

number = "5",

}

TY - JOUR

T1 - A Constrained Regularization Approach to Robust Corner Detection

AU - Sohn, Kwanghoon

AU - Alexander, Winser E.

AU - Kim, Jung H.

AU - Snyder, Wesley E.

N1 - Funding Information: In the local methods, the boundaries need to be segmented in order to represent them analytically. In general, there are two possible approaches to the boundary segmentation problem. One is to detect comers through angle or comer detection schemes. It mainly depends on a curvature function. The other approach is to obtain a piecewise linear polygonal approximation of the digitized boundary subject to certain constraints on the goodness of fit. The vertices found in the linear polygonal approximation are usually called break points. However, this approach may result in different boundary segmentation Manuscript received August 5, 1992; revised June 23, 1993. This work was supported in part by the U.S. Army Research Office under Grant DAAL03-90-0913, in part by the National Science Foundation under Grant ECD-8212696, and in part by the Federal Aviation Administration under Contract DTF A01-87-C-00043. K. Sohn is with the Satellite Communications Division, Electronics and Telecommunications Research Institute, P.O. Box 8, Daeduk Science Town, Daejeon, 305-606, Korea. W. E. Alexander is with the Dept. of Electrical & Computer Engineering, North Carolina State University, Raleigh, NC 27695-791 1 USA. J. H. Kim is with the Dept. of Electrical Engineering, North Carolina A&T State University, Greensboro, NC 2741 1 USA. W. E. Snyder is with the Dept. of Radiology, Bowman Gray School of Medicine, Wake Forest University, Winston-Salem, NC 27157-1022, USA. IEEE Log Number 9400636.

PY - 1994/5

Y1 - 1994/5

N2 - This paper presents a method of optimal boundary smoothing for curvature estimation and a method of corner detection for consistent representation of objects for computer vision applications. The existing methods for curvature estimation have a common problem in determining a unique smoothing factor. We propose a constrained regularization (CR) approach to overcome that problem. The curvature function computed on the preprocessed boundary, which is obtained by the CR approach, gives consistent corner detection results. Ideal corners rarely exist for a real boundary. They are often rounded due to the smoothing effects of the preprocessing. In addition, a human recognizes both sharp corners and slightly rounded segments as corners. Hence, we establish a criterion, called “corner sharpness”, which is qualitatively similar to a human's capability to detect corners.

AB - This paper presents a method of optimal boundary smoothing for curvature estimation and a method of corner detection for consistent representation of objects for computer vision applications. The existing methods for curvature estimation have a common problem in determining a unique smoothing factor. We propose a constrained regularization (CR) approach to overcome that problem. The curvature function computed on the preprocessed boundary, which is obtained by the CR approach, gives consistent corner detection results. Ideal corners rarely exist for a real boundary. They are often rounded due to the smoothing effects of the preprocessing. In addition, a human recognizes both sharp corners and slightly rounded segments as corners. Hence, we establish a criterion, called “corner sharpness”, which is qualitatively similar to a human's capability to detect corners.

UR - http://www.scopus.com/inward/record.url?scp=0028422887&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0028422887&partnerID=8YFLogxK

U2 - 10.1109/21.293500

DO - 10.1109/21.293500

M3 - Article

AN - SCOPUS:0028422887

SN - 0018-9472

VL - 24

SP - 820

EP - 828

JO - IEEE Transactions on Systems, Man and Cybernetics

JF - IEEE Transactions on Systems, Man and Cybernetics

IS - 5

ER -

A Constrained Regularization Approach to Robust Corner Detection

Abstract

Bibliographical note

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this