3D Shape Statistics

Dr Hamid Laga

The purpose of this project is to develop tools for the statistical analysis of 3D shapes and surfaces. The need for shape statistics arises in many areas. In medical image analysis, statistics of anatomical parts is important for automatic detection of their abnormal growth, and in the diagnosis of many diseases. In biometrics, methods for human recognition using faces, fingerprints and the morphology of body-parts rely in some way on a statistical analysis of shapes. In digital entertainment, statistical models of shapes enable automatic generation of 3D models, which is of major importance particularly in crowd simulation and background actor representation.

The focus of this project is on the development of algorithms for computing shape statistics. A typical scenario can be as follows; given a collection of 3D models of human bodies that represent some population, we would like to compute the average shape of each category of the population (e.g., persons aged between 12 and 18 years old), fit probabilistic models ideally from a parametric family (e.g., a Gaussian or Mixture of Gaussians), automatically detect abnormal, and automatically generate random instances (i.e., random sampling from the probabilistic model). Potential applications are numerous. The application of this method to plant biology is being undertaken.


H. Tabia,  H. Laga (2015) Covariance-based Descriptors for efficient 3D shape matching, retrieval and classification, IEEE Transactions on Multimedia, 17(9), 1591-1603.

H. Laga (2014) 3D Shape Classification and Retrieval Using Heterogeneous Features and Supervised Learning, in Machine Learning, Chapter 15, pp. 305-324. ISBN 978-953-7619-56-1, Hard cover,450 pages, Edited by: Abdelhamid Mellouk and Abdennacer Chebira, Publisher: IN-TECH, Jan 2009.

H. Tabia, H. Laga, D. Picard, P.-H. Gosselin (2014) Covariance Descriptors for 3D Shape Matching and Retrieval. IEEE International Conference on Computer Vision and Pattern Recognition (CVPR), 2014.

H. Laga, M. Mortara, M. Spagnuolo (2013) Geometry and Context for Semantic Correspondences and Functionality Recognition in Manmade 3D Shapes. ACM Transactions on Graphics (presented at Siggraph 2014), 32(5).

Hamid Laga, Hiroki Takahashi, Masayuki Nakajima, Spherical Wavelet Descriptors for Content-based 3D Model Retrieval. IEEE International Conference on Shape Modeling and Applications (SMI2006), Sendai, Japan, pp75-85, June 2006.

Areas of study and research

+ Click to minimise