TY - JOUR
T1 - Region-based Saliency Estimation for 3D Shape Analysis and Understanding
AU - Zhao, Yitian
AU - Liu, Yonghuai
AU - Wang, Yongjun
AU - Wei, Baogang
AU - Yang, Jian
AU - Zhao, Yifan
PY - 2016/7/12
Y1 - 2016/7/12
N2 - The detection of salient regions is an important pre-processing step for many
3D shape analysis and understanding tasks. This paper proposes a novel
method for saliency detection in 3D free form shapes. Firstly, we smooth
the surfaces by a bilateral normal lter. Such a ltering method is capable
of smoothing the surfaces and retaining the local details. Secondly, a novel
method is proposed for the estimation of the saliency value of each vertex.
To this end, two new features are de ned: Retinex-based Importance Feature
(RIF) and Relative Normal Distance (RND). They are based on the human
visual perception characteristics and surface geometry respectively. Since
the vertex based method cannot guarantee that the detected salient regions
are semantically continuous and complete, we propose to re ne the vertex
based saliency values based on surface patches. The detected saliency is
nally used to guide the existing techniques for mesh simpli cation, interest
point detection, and overlapping point cloud registration. The comparative
studies based on real data from three publicly accessible databases show
that the proposed method outperforms ve selected state of the art ones for
saliency detection and 3D shape analysis and understanding
AB - The detection of salient regions is an important pre-processing step for many
3D shape analysis and understanding tasks. This paper proposes a novel
method for saliency detection in 3D free form shapes. Firstly, we smooth
the surfaces by a bilateral normal lter. Such a ltering method is capable
of smoothing the surfaces and retaining the local details. Secondly, a novel
method is proposed for the estimation of the saliency value of each vertex.
To this end, two new features are de ned: Retinex-based Importance Feature
(RIF) and Relative Normal Distance (RND). They are based on the human
visual perception characteristics and surface geometry respectively. Since
the vertex based method cannot guarantee that the detected salient regions
are semantically continuous and complete, we propose to re ne the vertex
based saliency values based on surface patches. The detected saliency is
nally used to guide the existing techniques for mesh simpli cation, interest
point detection, and overlapping point cloud registration. The comparative
studies based on real data from three publicly accessible databases show
that the proposed method outperforms ve selected state of the art ones for
saliency detection and 3D shape analysis and understanding
KW - Saliency
KW - 3D surface
KW - Retinex
KW - local detail
KW - global geometry.Preprint
KW - Global geometry
KW - Local detail
UR - http://www.scopus.com/inward/record.url?scp=84957900702&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84957900702&partnerID=8YFLogxK
UR - https://www.mendeley.com/catalogue/f7d33d5d-b0bc-3d6e-aebf-a9221f2f543d/
U2 - 10.1016/j.neucom.2016.01.012
DO - 10.1016/j.neucom.2016.01.012
M3 - Article (journal)
SN - 0925-2312
VL - 197
SP - 1
EP - 13
JO - Neurocomputing
JF - Neurocomputing
IS - 197
ER -