3D landmark model discovery from a registered set of organic shapes

Creusot, Clement; Pears, Nick; Austin, Jim

doi:10.1109/cvprw.2012.6238915

Cited by 8 publications

(2 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Note that feature point saliency is an interesting topic in itself. Recent work has attempted to learn the most salient feature points from data [108].…”

Section: Methodsmentioning

confidence: 99%

Inverse Rendering of Faces with a 3D Morphable Model

Aldrian

Smith

2013

IEEE Trans. Pattern Anal. Mach. Intell.

179

160

View full text Add to dashboard Cite

In this thesis, we present a complete framework to inverse render faces with a 3D Morphable Model. By decomposing the image formation process into a geometric and photometric part, we are able to state the problem as a multilinear system which can be solved accurately and efficiently. As we treat each contribution as independent, the objective function is convex in the parameters and a globally optimal solution can be found. We start by recovering 3D shape using a novel algorithm which incorporates generalisation errors of the model obtained from empirical measurements. The algorithm is extended so it can efficiently deal with mixture distributions. We then describe three methods to recover facial texture, and for the second and third, diffuse lighting, specular reflectance and camera properties from a single image. These methods make increasingly weak assumptions and can all be solved in a linear fashion. We further modify our framework so it accounts for global illumination effects. This is achieved by incorporating statistical models for ambient occlusion and bent normals into the image formation model. We show that solving for ambient occlusion and bent normal parameters as part of the fitting process improves the accuracy of the estimated texture map and illumination environment. We present results on challenging data, rendered under complex natural illumination with both specular reflectance and occlusion of the illumination environment. We evaluate our findings on publicly available datasets, where we are able to obtain state-ofthe-art results. Finally, we present a practical method to synthesise a larger population from a small training-set and show how the new instances can be used to build a flexible PCA model.

show abstract

“…Note that feature point saliency is an interesting topic in itself. Recent work has attempted to learn the most salient feature points from data [108].…”

Section: Methodsmentioning

confidence: 99%

Inverse Rendering of Faces with a 3D Morphable Model

Aldrian

Smith

2013

IEEE Trans. Pattern Anal. Mach. Intell.

179

160

View full text Add to dashboard Cite

show abstract

“…Mean, Gaussian, principal curvatures, shape index, and curvedness, evaluated with varying neighbourhood size and bin size, were adopted as descriptors by Creusot et al to support keypoints detection on 3D faces. LDA was employed to define weights to combine matching score maps over a population of neighbouring and nonneighbouring vertices, relative to the relevant landmark, and experiments were carried out on FRGC v2 [22] and BFM database [23]. Histograms of shape index (HoS) with 8 bins, the shape index itself, and principal curvatures were used to develop a mesh-based method for 3D facial expression recognition to be tested on BU3D-FE [24] [25] [26] and Bosphorus [27] databases.…”

mentioning

confidence: 99%

Novel descriptors for geometrical 3D face analysis

Marcolin

Vezzetti

2016

Multimed Tools Appl

View full text Add to dashboard Cite

3D face was recently investigated for various applications, including biometrics and diagnosis. Describing facial surface, i.e. how it bends and which kinds of patches is composed by, is the aim of studies of Face Analysis, whose ultimate goal is to identify which features could be extracted from three-dimensional faces depending on the application. In this study, we propose 105 novel geometrical descriptors for Face Analysis. They are generated by composing primary geometrical descriptors such as mean, Gaussian, principal curvatures, shape index, curvedness, and the coefficients of the fundamental forms, and by applying standard functions such as sine, cosine, and logarithm to them. The new descriptors were mapped on 217 facial depth maps and analysed in terms of descriptiveness of facial shape and exploitability for localizing landmark points. Automatic landmark extraction stands as the final aim of this analysis. Results showed that some newly generated descriptors were sounder than the primary ones, meaning that their local behaviours in correspondence to a landmark position is thoroughly specific and can be registered with high similarity on every face of our dataset.

show abstract