Visual learning and recognition of 3-d objects from appearance View Full Text


Ontology type: schema:ScholarlyArticle     


Article Info

DATE

1995-01

AUTHORS

Hiroshi Murase, Shree K. Nayar

ABSTRACT

The problem of automatically learning object models for recognition and pose estimation is addressed. In contrast to the traditional approach, the recognition problem is formulated as one of matching appearance rather than shape. The appearance of an object in a two-dimensional image depends on its shape, reflectance properties, pose in the scene, and the illumination conditions. While shape and reflectance are intrinsic properties and constant for a rigid object, pose and illumination vary from scene to scene. A compact representation of object appearance is proposed that is parametrized by pose and illumination. For each object of interest, a large set of images is obtained by automatically varying pose and illumination. This image set is compressed to obtain a low-dimensional subspace, called the eigenspace, in which the object is represented as a manifold. Given an unknown input image, the recognition system projects the image to eigenspace. The object is recognized based on the manifold it lies on. The exact position of the projection on the manifold determines the object's pose in the image.A variety of experiments are conducted using objects with complex appearance characteristics. The performance of the recognition and pose estimation algorithms is studied using over a thousand input images of sample objects. Sensitivity of recognition to the number of eigenspace dimensions and the number of learning samples is analyzed. For the objects used, appearance representation in eigenspaces with less than 20 dimensions produces accurate recognition results with an average pose estimation error of about 1.0 degree. A near real-time recognition system with 20 complex objects in the database has been developed. The paper is concluded with a discussion on various issues related to the proposed learning and recognition methodology. More... »

PAGES

5-24

References to SciGraph publications

Journal

Related Patents

  • Methods And Apparatus To Facilitate Operations In Image Based Systems
  • System And Method Of Three-Dimensional Pose Estimation
  • Method For Image Recognition Of Variably Illuminated Solid Objects Using Fast Optimal Linear Approximation
  • Camera Pose Estimation Method For Augmented Reality Manual For Cars
  • Device, Method And Record Medium For Image Comparison
  • Low Latency Stabilization For Head-Worn Displays
  • Image Labeling For Cleaning Robot Deep Learning System
  • Device, Method And Record Medium For Image Comparison
  • Apparatus And Method For Identifying And Tracking Objects With View-Based Representations
  • Method Of Object Recognition, Apparatus Of The Same And Recording Medium Therefor
  • Object-Learning Robot And Method
  • Mobile Cleaning Robot Artificial Intelligence For Situational Awareness
  • Video-Based Face Recognition Using Probabilistic Appearance Manifolds
  • Device For Creating Image Feature From Image Having Any Shape
  • System And Method Of Determining Object Pose
  • Device, Method And Record Medium For Image Comparison
  • System And Method For Detecting And Tracking Features In Images
  • Creating Image Features From An Image Having Any Shape
  • Information Processing Method, Information Processing Apparatus, And Recording Medium
  • Object Recognition System And Process For Identifying People And Objects In An Image Of A Scene
  • Change Information Recognition Apparatus And Change Information Recognition Method
  • System, Method And Computer Program Product For Training A Three Dimensional Object Indentification System And Identifying Three Dimensional Objects Using Semantic Segments
  • System For Spot Cleaning By A Mobile Robot
  • Method For Image Recognition Of Variably Illuminated Solid Objects Using Fast Optimal Linear Approximation
  • Methods And Apparatus For Object Recognition Using Textons
  • System And Method For Detecting And Tracking Features In Images
  • Quotient Appearance Manifold Mapping For Image Classification
  • Object Recognition With Occurrence Histograms
  • Method And Apparatus For Object Recognition Using Probability Models
  • Clustering Appearances Of Objects Under Varying Illumination Conditions
  • Lambertian Reflectance And Linear Subspaces
  • Object Recognition With Co-Occurrence Histograms And False Alarm Probability Analysis For Choosing Optimal Object Recognition Process Parameters
  • Robot With Vision-Based 3d Shape Recognition
  • Methods, Systems, And Computer-Readable Media For Utilizing Radial Distortion To Estimate A Pose Configuration
  • Minimal-Latency Tracking And Display For Matching Real And Virtual Worlds In Head-Worn Displays
  • Quotient Appearance Manifold Mapping For Image Classification
  • Methods, Systems, And Computer Readable Media For Unified Scene Acquisition And Pose Tracking In A Wearable Display
  • Change Information Recognition Apparatus And Change Information Recognition Method
  • Method And Apparatus For Single Image 3d Vision Guided Robotics
  • Computational Imaging Of The Electric Grid
  • Information Processing Method, Information Processing Apparatus, And Recording Medium For Identifying A Class Of An Object By Using A Plurality Of Discriminators
  • System For Spot Cleaning By A Mobile Robot
  • Imaging Method And System
  • Identifiers

    URI

    http://scigraph.springernature.com/pub.10.1007/bf01421486

    DOI

    http://dx.doi.org/10.1007/bf01421486

    DIMENSIONS

    https://app.dimensions.ai/details/publication/pub.1020275860


    Indexing Status Check whether this publication has been indexed by Scopus and Web Of Science using the SN Indexing Status Tool
    Incoming Citations Browse incoming citations for this publication using opencitations.net

    JSON-LD is the canonical representation for SciGraph data.

    TIP: You can open this SciGraph record using an external JSON-LD service: JSON-LD Playground Google SDTT

    [
      {
        "@context": "https://springernature.github.io/scigraph/jsonld/sgcontext.json", 
        "about": [
          {
            "id": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/08", 
            "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
            "name": "Information and Computing Sciences", 
            "type": "DefinedTerm"
          }, 
          {
            "id": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/0801", 
            "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
            "name": "Artificial Intelligence and Image Processing", 
            "type": "DefinedTerm"
          }
        ], 
        "author": [
          {
            "affiliation": {
              "alternateName": "NTT Basic Research Laboratory, Atsugi-Shi, 243-01, Kanagawa, Japan", 
              "id": "http://www.grid.ac/institutes/None", 
              "name": [
                "NTT Basic Research Laboratory, Atsugi-Shi, 243-01, Kanagawa, Japan"
              ], 
              "type": "Organization"
            }, 
            "familyName": "Murase", 
            "givenName": "Hiroshi", 
            "id": "sg:person.011047516675.56", 
            "sameAs": [
              "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.011047516675.56"
            ], 
            "type": "Person"
          }, 
          {
            "affiliation": {
              "alternateName": "Department of Computer Science, Columbia University, 10027, New York, N.Y.", 
              "id": "http://www.grid.ac/institutes/grid.21729.3f", 
              "name": [
                "Department of Computer Science, Columbia University, 10027, New York, N.Y."
              ], 
              "type": "Organization"
            }, 
            "familyName": "Nayar", 
            "givenName": "Shree K.", 
            "id": "sg:person.0640261263.22", 
            "sameAs": [
              "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.0640261263.22"
            ], 
            "type": "Person"
          }
        ], 
        "citation": [
          {
            "id": "sg:pub.10.1007/bf00054921", 
            "sameAs": [
              "https://app.dimensions.ai/details/publication/pub.1031178628", 
              "https://doi.org/10.1007/bf00054921"
            ], 
            "type": "CreativeWork"
          }, 
          {
            "id": "sg:pub.10.1007/bf00201981", 
            "sameAs": [
              "https://app.dimensions.ai/details/publication/pub.1049615829", 
              "https://doi.org/10.1007/bf00201981"
            ], 
            "type": "CreativeWork"
          }, 
          {
            "id": "sg:pub.10.1007/bf00337644", 
            "sameAs": [
              "https://app.dimensions.ai/details/publication/pub.1024974725", 
              "https://doi.org/10.1007/bf00337644"
            ], 
            "type": "CreativeWork"
          }, 
          {
            "id": "sg:pub.10.1038/343263a0", 
            "sameAs": [
              "https://app.dimensions.ai/details/publication/pub.1000677865", 
              "https://doi.org/10.1038/343263a0"
            ], 
            "type": "CreativeWork"
          }
        ], 
        "datePublished": "1995-01", 
        "datePublishedReg": "1995-01-01", 
        "description": "The problem of automatically learning object models for recognition and pose estimation is addressed. In contrast to the traditional approach, the recognition problem is formulated as one of matching appearance rather than shape. The appearance of an object in a two-dimensional image depends on its shape, reflectance properties, pose in the scene, and the illumination conditions. While shape and reflectance are intrinsic properties and constant for a rigid object, pose and illumination vary from scene to scene. A compact representation of object appearance is proposed that is parametrized by pose and illumination. For each object of interest, a large set of images is obtained by automatically varying pose and illumination. This image set is compressed to obtain a low-dimensional subspace, called the eigenspace, in which the object is represented as a manifold. Given an unknown input image, the recognition system projects the image to eigenspace. The object is recognized based on the manifold it lies on. The exact position of the projection on the manifold determines the object's pose in the image.A variety of experiments are conducted using objects with complex appearance characteristics. The performance of the recognition and pose estimation algorithms is studied using over a thousand input images of sample objects. Sensitivity of recognition to the number of eigenspace dimensions and the number of learning samples is analyzed. For the objects used, appearance representation in eigenspaces with less than 20 dimensions produces accurate recognition results with an average pose estimation error of about 1.0 degree. A near real-time recognition system with 20 complex objects in the database has been developed. The paper is concluded with a discussion on various issues related to the proposed learning and recognition methodology.", 
        "genre": "article", 
        "id": "sg:pub.10.1007/bf01421486", 
        "isAccessibleForFree": false, 
        "isPartOf": [
          {
            "id": "sg:journal.1032807", 
            "issn": [
              "0920-5691", 
              "1573-1405"
            ], 
            "name": "International Journal of Computer Vision", 
            "publisher": "Springer Nature", 
            "type": "Periodical"
          }, 
          {
            "issueNumber": "1", 
            "type": "PublicationIssue"
          }, 
          {
            "type": "PublicationVolume", 
            "volumeNumber": "14"
          }
        ], 
        "keywords": [
          "input image", 
          "recognition system", 
          "real-time recognition system", 
          "average pose estimation error", 
          "unknown input image", 
          "accurate recognition results", 
          "pose estimation algorithm", 
          "pose estimation error", 
          "object of interest", 
          "sensitivity of recognition", 
          "low-dimensional subspace", 
          "pose estimation", 
          "recognition problem", 
          "object poses", 
          "object model", 
          "appearance representation", 
          "recognition results", 
          "recognition methodology", 
          "object appearance", 
          "two-dimensional images", 
          "compact representation", 
          "complex objects", 
          "image sets", 
          "pose", 
          "rigid objects", 
          "illumination conditions", 
          "variety of experiments", 
          "estimation algorithm", 
          "scene", 
          "traditional approaches", 
          "large set", 
          "sample objects", 
          "appearance characteristics", 
          "images", 
          "objects", 
          "recognition", 
          "learning", 
          "eigenspace", 
          "estimation error", 
          "exact position", 
          "visual learning", 
          "representation", 
          "set", 
          "eigenspace dimension", 
          "algorithm", 
          "system", 
          "reflectance properties", 
          "database", 
          "illumination", 
          "subspace", 
          "performance", 
          "error", 
          "methodology", 
          "issues", 
          "estimation", 
          "number", 
          "model", 
          "experiments", 
          "dimensions", 
          "projections", 
          "shape", 
          "interest", 
          "variety", 
          "manifold", 
          "results", 
          "intrinsic properties", 
          "appearance", 
          "position", 
          "discussion", 
          "characteristics", 
          "degree", 
          "properties", 
          "conditions", 
          "contrast", 
          "reflectance", 
          "sensitivity", 
          "samples", 
          "problem", 
          "paper", 
          "approach"
        ], 
        "name": "Visual learning and recognition of 3-d objects from appearance", 
        "pagination": "5-24", 
        "productId": [
          {
            "name": "dimensions_id", 
            "type": "PropertyValue", 
            "value": [
              "pub.1020275860"
            ]
          }, 
          {
            "name": "doi", 
            "type": "PropertyValue", 
            "value": [
              "10.1007/bf01421486"
            ]
          }
        ], 
        "sameAs": [
          "https://doi.org/10.1007/bf01421486", 
          "https://app.dimensions.ai/details/publication/pub.1020275860"
        ], 
        "sdDataset": "articles", 
        "sdDatePublished": "2022-09-02T15:49", 
        "sdLicense": "https://scigraph.springernature.com/explorer/license/", 
        "sdPublisher": {
          "name": "Springer Nature - SN SciGraph project", 
          "type": "Organization"
        }, 
        "sdSource": "s3://com-springernature-scigraph/baseset/20220902/entities/gbq_results/article/article_279.jsonl", 
        "type": "ScholarlyArticle", 
        "url": "https://doi.org/10.1007/bf01421486"
      }
    ]
     

    Download the RDF metadata as:  json-ld nt turtle xml License info

    HOW TO GET THIS DATA PROGRAMMATICALLY:

    JSON-LD is a popular format for linked data which is fully compatible with JSON.

    curl -H 'Accept: application/ld+json' 'https://scigraph.springernature.com/pub.10.1007/bf01421486'

    N-Triples is a line-based linked data format ideal for batch operations.

    curl -H 'Accept: application/n-triples' 'https://scigraph.springernature.com/pub.10.1007/bf01421486'

    Turtle is a human-readable linked data format.

    curl -H 'Accept: text/turtle' 'https://scigraph.springernature.com/pub.10.1007/bf01421486'

    RDF/XML is a standard XML format for linked data.

    curl -H 'Accept: application/rdf+xml' 'https://scigraph.springernature.com/pub.10.1007/bf01421486'


     

    This table displays all metadata directly associated to this object as RDF triples.

    163 TRIPLES      21 PREDICATES      109 URIs      97 LITERALS      6 BLANK NODES

    Subject Predicate Object
    1 sg:pub.10.1007/bf01421486 schema:about anzsrc-for:08
    2 anzsrc-for:0801
    3 schema:author N697147f5fdbe497b85dc2bf64eba7eec
    4 schema:citation sg:pub.10.1007/bf00054921
    5 sg:pub.10.1007/bf00201981
    6 sg:pub.10.1007/bf00337644
    7 sg:pub.10.1038/343263a0
    8 schema:datePublished 1995-01
    9 schema:datePublishedReg 1995-01-01
    10 schema:description The problem of automatically learning object models for recognition and pose estimation is addressed. In contrast to the traditional approach, the recognition problem is formulated as one of matching appearance rather than shape. The appearance of an object in a two-dimensional image depends on its shape, reflectance properties, pose in the scene, and the illumination conditions. While shape and reflectance are intrinsic properties and constant for a rigid object, pose and illumination vary from scene to scene. A compact representation of object appearance is proposed that is parametrized by pose and illumination. For each object of interest, a large set of images is obtained by automatically varying pose and illumination. This image set is compressed to obtain a low-dimensional subspace, called the eigenspace, in which the object is represented as a manifold. Given an unknown input image, the recognition system projects the image to eigenspace. The object is recognized based on the manifold it lies on. The exact position of the projection on the manifold determines the object's pose in the image.A variety of experiments are conducted using objects with complex appearance characteristics. The performance of the recognition and pose estimation algorithms is studied using over a thousand input images of sample objects. Sensitivity of recognition to the number of eigenspace dimensions and the number of learning samples is analyzed. For the objects used, appearance representation in eigenspaces with less than 20 dimensions produces accurate recognition results with an average pose estimation error of about 1.0 degree. A near real-time recognition system with 20 complex objects in the database has been developed. The paper is concluded with a discussion on various issues related to the proposed learning and recognition methodology.
    11 schema:genre article
    12 schema:isAccessibleForFree false
    13 schema:isPartOf N43e679b3464f4d46a3b0489b5f14bfab
    14 Nc67000ffcc0742bc894e8da1f0ec1645
    15 sg:journal.1032807
    16 schema:keywords accurate recognition results
    17 algorithm
    18 appearance
    19 appearance characteristics
    20 appearance representation
    21 approach
    22 average pose estimation error
    23 characteristics
    24 compact representation
    25 complex objects
    26 conditions
    27 contrast
    28 database
    29 degree
    30 dimensions
    31 discussion
    32 eigenspace
    33 eigenspace dimension
    34 error
    35 estimation
    36 estimation algorithm
    37 estimation error
    38 exact position
    39 experiments
    40 illumination
    41 illumination conditions
    42 image sets
    43 images
    44 input image
    45 interest
    46 intrinsic properties
    47 issues
    48 large set
    49 learning
    50 low-dimensional subspace
    51 manifold
    52 methodology
    53 model
    54 number
    55 object appearance
    56 object model
    57 object of interest
    58 object poses
    59 objects
    60 paper
    61 performance
    62 pose
    63 pose estimation
    64 pose estimation algorithm
    65 pose estimation error
    66 position
    67 problem
    68 projections
    69 properties
    70 real-time recognition system
    71 recognition
    72 recognition methodology
    73 recognition problem
    74 recognition results
    75 recognition system
    76 reflectance
    77 reflectance properties
    78 representation
    79 results
    80 rigid objects
    81 sample objects
    82 samples
    83 scene
    84 sensitivity
    85 sensitivity of recognition
    86 set
    87 shape
    88 subspace
    89 system
    90 traditional approaches
    91 two-dimensional images
    92 unknown input image
    93 variety
    94 variety of experiments
    95 visual learning
    96 schema:name Visual learning and recognition of 3-d objects from appearance
    97 schema:pagination 5-24
    98 schema:productId N3bb3f5d82be94a55a47fc5a0d5f872a2
    99 N9aec38ee261643fe8ad51f8e1b87a454
    100 schema:sameAs https://app.dimensions.ai/details/publication/pub.1020275860
    101 https://doi.org/10.1007/bf01421486
    102 schema:sdDatePublished 2022-09-02T15:49
    103 schema:sdLicense https://scigraph.springernature.com/explorer/license/
    104 schema:sdPublisher Na3bddaa7775c42318a85b71342b88c45
    105 schema:url https://doi.org/10.1007/bf01421486
    106 sgo:license sg:explorer/license/
    107 sgo:sdDataset articles
    108 rdf:type schema:ScholarlyArticle
    109 N3bb3f5d82be94a55a47fc5a0d5f872a2 schema:name dimensions_id
    110 schema:value pub.1020275860
    111 rdf:type schema:PropertyValue
    112 N43e679b3464f4d46a3b0489b5f14bfab schema:volumeNumber 14
    113 rdf:type schema:PublicationVolume
    114 N697147f5fdbe497b85dc2bf64eba7eec rdf:first sg:person.011047516675.56
    115 rdf:rest Nc0e88310d8db4d339a581c541e4db1bf
    116 N9aec38ee261643fe8ad51f8e1b87a454 schema:name doi
    117 schema:value 10.1007/bf01421486
    118 rdf:type schema:PropertyValue
    119 Na3bddaa7775c42318a85b71342b88c45 schema:name Springer Nature - SN SciGraph project
    120 rdf:type schema:Organization
    121 Nc0e88310d8db4d339a581c541e4db1bf rdf:first sg:person.0640261263.22
    122 rdf:rest rdf:nil
    123 Nc67000ffcc0742bc894e8da1f0ec1645 schema:issueNumber 1
    124 rdf:type schema:PublicationIssue
    125 anzsrc-for:08 schema:inDefinedTermSet anzsrc-for:
    126 schema:name Information and Computing Sciences
    127 rdf:type schema:DefinedTerm
    128 anzsrc-for:0801 schema:inDefinedTermSet anzsrc-for:
    129 schema:name Artificial Intelligence and Image Processing
    130 rdf:type schema:DefinedTerm
    131 sg:journal.1032807 schema:issn 0920-5691
    132 1573-1405
    133 schema:name International Journal of Computer Vision
    134 schema:publisher Springer Nature
    135 rdf:type schema:Periodical
    136 sg:person.011047516675.56 schema:affiliation grid-institutes:None
    137 schema:familyName Murase
    138 schema:givenName Hiroshi
    139 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.011047516675.56
    140 rdf:type schema:Person
    141 sg:person.0640261263.22 schema:affiliation grid-institutes:grid.21729.3f
    142 schema:familyName Nayar
    143 schema:givenName Shree K.
    144 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.0640261263.22
    145 rdf:type schema:Person
    146 sg:pub.10.1007/bf00054921 schema:sameAs https://app.dimensions.ai/details/publication/pub.1031178628
    147 https://doi.org/10.1007/bf00054921
    148 rdf:type schema:CreativeWork
    149 sg:pub.10.1007/bf00201981 schema:sameAs https://app.dimensions.ai/details/publication/pub.1049615829
    150 https://doi.org/10.1007/bf00201981
    151 rdf:type schema:CreativeWork
    152 sg:pub.10.1007/bf00337644 schema:sameAs https://app.dimensions.ai/details/publication/pub.1024974725
    153 https://doi.org/10.1007/bf00337644
    154 rdf:type schema:CreativeWork
    155 sg:pub.10.1038/343263a0 schema:sameAs https://app.dimensions.ai/details/publication/pub.1000677865
    156 https://doi.org/10.1038/343263a0
    157 rdf:type schema:CreativeWork
    158 grid-institutes:None schema:alternateName NTT Basic Research Laboratory, Atsugi-Shi, 243-01, Kanagawa, Japan
    159 schema:name NTT Basic Research Laboratory, Atsugi-Shi, 243-01, Kanagawa, Japan
    160 rdf:type schema:Organization
    161 grid-institutes:grid.21729.3f schema:alternateName Department of Computer Science, Columbia University, 10027, New York, N.Y.
    162 schema:name Department of Computer Science, Columbia University, 10027, New York, N.Y.
    163 rdf:type schema:Organization
     




    Preview window. Press ESC to close (or click here)


    ...