Browse Prior Art Database

A REPRESENTATION FOR SHAPE BASED ON PEAKS AND RIDGES IN THE DIFFERENCE OF LOW-PASS TRANSFORM

IP.com Disclosure Number: IPCOM000148154D
Original Publication Date: 1983-May-31
Included in the Prior Art Database: 2007-Mar-29
Document File: 42 page(s) / 3M

Publishing Venue

Software Patent Institute

Related People

Crowley, James L.: AUTHOR [+3]

Abstract

JAMES L. CROWLEY AND ALICE C. PARKER READfMG 2QOM w*-R1-TR-g3-4 OMPUTEFI S@iEBCE DEF'ARWT$&i\;*&. YALE tJNlVEt781Tv I t -r May 1983 A Representation far Shape Based G I Peaks ag-=d Rids, as th Diftevsnce of Low-Pass Transfor .,a Jarnies L. Cro ley Robotics Institute Carcegie-Mellon University Dept. of Electrical Engineei-iag-S ystcms University of So~lthern California May 1983 Copyright 1983 J. I,. Crowley This research was partially supported by:The C-MU Robotics Institute,National Science Fwnd3tioo Grant No. APR75 05154, and byNaval Electronics System Command (NELC) Grant No. NO0039 79 Z 0169 Abstract: This paper defines a multiplc rcsolrltion represcrltatlon for the two-dimensional y-scale shapes in an image. 'This represenration is constrccted by detecting peaks and ridges in the of Low Pass (DOLP) transfo~m. Descriptions of shapes which are ct~cocled in dlisefficiently decpite changes in size, orientation or position. Motii~atioi?~ f r a mnltiple resalution representation arc presented first, followed d?e definition of the

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 5% of the total text.

Page 1 of 42

 A REPRESENTATION FOR SHAPE BASED ON PEAKS AND RIDGES IN THE DIFFERENCE OF LOW-PASS TRANSFORM

JAMES L. CROWLEY AND ALICE C. PARKER

READfMG 2QOM w*-R1-TR-g3-4

OMPUTEFI S@iEBCE DEF'ARWT$&i\;*&.

YALE tJNlVEt781Tv I t -r

May 1983

-"--

[This page contains 1 picture or other non-text object]

Page 2 of 42

[This page contains 1 picture or other non-text object]

Page 3 of 42

A Representation far Shape

Based G ~ I
Peaks ag-=d Rids,

.,a

Jarnies L. Cro ley

Robotics Institute

Carcegie-Mellon University

Dept. of Electrical Engineei-iag-S ystcms

University of So~lthern California

May 1983

Copyright @ 1983 J. I,. Crowley

This research was partially supported by:
The C-MU Robotics Institute,
National Science Fwnd3tioo Grant No. APR75 - 05154, and by
Naval Electronics System Command (NELC) Grant No. NO0039 - 79 - Z

                  as th Diftevsnce of Low-Pass Transfor

- 0169

[This page contains 1 picture or other non-text object]

Page 4 of 42

[This page contains 1 picture or other non-text object]

Page 5 of 42

Abstract:

 This paper defines a multiplc rcsolrltion represcrltatlon for the two-dimensional y-scale shapes in an image. 'This represenration is constrccted by detecting peaks and ridges in the of Low Pass (DOLP) transfo~m. Descriptions of shapes which are ct~cocled in dlis
efficiently decpite changes in size, orientation or position.

Motii~atioi?~ f ~ r
a mnltiple resalution representation arc presented first, followed d?e definition of the

DOLP l'ransfarm. T'cchniques arc then presented for cncoding a syrnbofic desa~ipcion of fonns from the UOI.,P transform. This proccss i~lvolves detecting local peaks 2nd band-pass image and in Ole entire ttxrce-dimerisio!ld space defincd by the I>OI..P peaks in different band-pass images gi,ves a multiple nsohltion tree which
maxina in 'this tree providc ~~~~~~~~~ks for- aligning,
linking Ihe ridges in each DOLP band-pass itnage
band-pass image and describes tix positio~~s
of tile

and linkine the ridges in the. D,OLP three space tree.

 The principles for determining the corresporldence between symbols in pairs of descriptions we then described. Such corrcspondence matcl~ing is shown to be simplified by using at lower resolutions to constrain the possible correspondence at higher resolutions.

1.

[This page contains 1 picture or other non-text object]

Page 6 of 42

[This page contains 1 picture or other non-text object]

Page 7 of 42

1 Introduction
1.1 Motitation:A Mulci-li(:solution Stn~ctural DcscriptiorĂ´f Images
1.2 Properties of tl~c
I)\=prcscntation
1.3 Correspondence hilatching
1.4 Contcnts of this Papcr

2 The Diffcrctlcc of I . ~ i r
-Pass l'sansfoqn
2.1 Tllc P~lrposc
ot' the DOL.1) Transform

2.2 Defitiitioll of tl~c
DOl,P transform

2.3 Fast Computation l'cckr?iq:;es: Rcscvnpling and Cascadc Convolution
2.4 An Example: the DO1 .P Trailsfo~m of a Teapot Image 3 Constniction of tl~c
licprescntation from a DOLP Trans...