View selection for sketch-based 3D model retrieval using visual part shape description - Archive ouverte HAL Access content directly
Journal Articles The Visual Computer Year : 2017

View selection for sketch-based 3D model retrieval using visual part shape description

(1) , (1) , (2)
1
2

Abstract

Hand drawings are the imprints of shapes in human’s mind. How a human expresses a shape is a consequence of how he or she visualizes it. A query-by-sketch 3D object retrieval application is closely tied to this concept from two aspects. First, describing sketches must involve elements in a figure that matter most to a human. Second, the representative 2D projection of the target 3D objects should be limited to “the canonical views” from a human cognition perspective. We advocate for these two rules by presenting a new approach for sketch-based 3D object retrieval that describes a 2D shape by the visual protruding parts of its silhouette. Furthermore, we present a list of candidate 2D projections that represent the canonical views of a 3D object. The general rule is that humans would visually avoid part occlusion and symmetry. We quantify the extent of part occlusion of the projected silhouettes of 3D objects by skeletal length computations. Sorting the projected views in the decreasing order of skeletal lengths gives access to a subset of the best representative views. We experimentally show how views that cause misinterpretation and mismatching can be detected according to the part occlusion criteria. We also propose criteria for locating side, off axis, or asymmetric views.
Fichier principal
Vignette du fichier
Author_Manuscript_VisualComputer.pdf (2.1 Mo) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-01396333 , version 1 (19-12-2016)

Identifiers

Cite

Zahraa Yasseen, Anne Verroust-Blondet, Ahmad Nasri. View selection for sketch-based 3D model retrieval using visual part shape description. The Visual Computer, 2017, 33 (5), pp.565-583. ⟨10.1007/s00371-016-1328-7⟩. ⟨hal-01396333⟩

Collections

CNRS INRIA INRIA2
194 View
450 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More