Estimating 3D shape and texture using pixel intensity, edges, specular highlights, texture constraints and a prior

By Sami Romdhani, Thomas Vetter. 2005

331 citations recorded by the source corpus.

Subfields: Computational Mechanics, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Mechanics

In Computer Vision and Pattern Recognition

Citations in the corpus

A Hybrid Model for Identity Obfuscation by Face Replacement (2018, background): “Such parametric models can act as strong regularizers for 3D face reconstruction problems, and have been widely used in optimization based [5,14,12,15,16,17] and learningbased [18,19,20,21,22,23] settings.”
I M Avatar: Implicit Morphable Head Avatars from Videos (2021, methodology): “Conventional methods [15, 17, 19, 20, 41, 43, 45, 47] based on morphable mesh models [3, 27, 36] can fit the shape and texture parameters to images for a given subject.”
Self-Supervised Multi-level Face Model Learning for Monocular Reconstruction at Over 250 Hz (2017, background): “Optimization-based Approaches: Many approaches for monocular face reconstruction [50], reconstruction based on image collections [51], and the estimation of highquality 3D face rigs [26] are based on energy optimization.”
FaceScape: A Large-Scale High Quality 3D Face Dataset and Detailed Riggable 3D Face Prediction (2020, methodology): “The emergence of 3DMM has sim-pliﬁed the single-view reconstruction of face to a model ﬁtting problem, which could be well solved by ﬁtting facial landmarks and other features[38, 46] or regressing the parameter of 3DMM with a deep neural network[20, 63].”
Large Pose 3D Face Reconstruction from a Single Image via Direct Volumetric CNN Regression (2017, methodology): “The work of [20] describes a multi-feature based approach to 3DMM fitting using non-linear least-squares optimization (Levenberg-Marquardt), which given appropriate initialisation produces results of good accuracy.”
EMOCA: Emotion Driven Monocular Face Capture and Animation (2022, methodology): “Existing methods can be categorized into optimization-based [4,5,9,10,31,45,63,69,84,90] and learning-based.”
Extreme 3D Face Reconstruction: Seeing Through Occlusions (2017, methodology): “Baseline methods include the analysis by synthesis 3DMM fitting of [40], the flow based approach of [14], the recent 3DDFA method for facial landmark detection and shape estimation [58], the deep 3DMM method used to produce our foundation shapes [51], and the deep system for detailed reconstructions”
Recognizing Rotated Faces From Frontal and Side Views: An Approach Toward Effective Use of Mugshot Databases (2008, background): “Police mugshot databases1 usually consist of a frontal view and a side view per person (Fig.”
DenseReg: Fully Convolutional Dense Shape Regression In-the-Wild (2016, methodology): “ “in the wild”, to ﬁt the template face to the image by obtaining a coordinate transformation for each vertex xj. We use the ﬁttings provided by [55] which were ﬁt using a modiﬁed 3DMM implementation [39]. However, for the purpose of this paper, we require a per-pixel estimate of the location in UV ”
Extreme 3D Face Reconstruction: Looking Past Occlusions (2017, methodology): “Baseline methods include the analysis by synthesis 3DMM ﬁtting of [38], the ﬂow based approach of [15], the recent 3DDFA method for facial landmark detection and shape estimation [53], the deep 3DMM method used to produce our foundation shapes [46], and the deep system for detailed reconstructions…”
3D Morphable Face Models—Past, Present, and Future (2019, unknown): “s approaches evaluate the photometric consistency between the observations and the estimates. Some approaches have explored the use of other image features, such as edges, or SIFT [Booth et al. 2017; Romdhani and Vetter 2005], in order to obtain higher fidelity reconstructions. Occlusion robust reco”
Accurate 3D Face Reconstruction With Weakly-Supervised Learning: From Single Image to Image Set (2019, methodology): “With a 3DMM, reconstruction can be performed by an analysis-by-synthesis scheme using image intensity [5] and other features such as edges [39].”
Single View Reconstruction of Curved Surfaces (2006, background): “Romdhani and Vetter [11] use a linear basis for the 3D shape and then MAP estimation to instantiate shape from a single image, while Grauman et al. [3] learn to generate virtual views from a single silhouette from which a visual hull can be computed.”
Face Alignment in Full Pose Range: A 3D Total Solution (2015, methodology): “Since only the visible model vertices are fitted, it is the first method to cover arbitrary poses [23], [44], but it suffers from the one-minute-per-image computational cost.”
Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Network (2018, methodology): “Most of earlier methods are to establish the correspondences of the special points between input images and the 3D mean geometry including landmarks[21,6,22,7,23,24,25] and local features[26,27,25], then solve the non-linear optimization function to regress the 3DMM coeﬃcients.”
Photorealistic Facial Texture Inference Using Deep Neural Networks (2016, background): “Most state-of-the-art monocular facial capture frameworks [49, 42] rely on linear PCA models [5] and important appearance details for photorealistic rendering such as complex skin pigmentation variations and mesoscopic-level texture details (freckles, pores, stubble hair, etc.”
A Comprehensive Survey on Pose-Invariant Face Recognition (2015, background): “For example, Romdhani and Vetter [2005] proposed a new ﬁtting algorithm that employs multiple features including pixel intensity, edges, and specular highlights.”