3D and Infrared Face Reconstruction from RGB data using Canonical Correlation Analysis

Reiter, Michael J.; Dormer, R.; Langs, Georg; Bischof, Horst

doi:10.1109/icpr.2006.24

Cited by 48 publications

(30 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Relatively few attempts have been made on directly converting between IR and VS faces. Reiter et al ( [4]) have proposed an algorithm, which applies CCA to map images between near IR and VS. However, near IR images capture reflected photons from the subjects in a similar way that the VS images work.…”

Section: Related Workmentioning

confidence: 99%

“…In optimization, it favors adjustments that keep the VS patch related to the input: Consider two adjustment vectors ∆x 1 and ∆x 2 with equal norms. Then the cost of these two adjustments in terms of the observation energy in Eq (4) shows that the adjustment that changes less correlated components costs less observation energy.…”

Section: Markov Random Field Of Patchesmentioning

confidence: 99%

“…(v) σ φ and σ ψ are set to 1. In Figure 6, the results of our method on middle-wavelength IR (MWIR) images are compared to that of holistic CCA/EigenTransformation [4][5], patch-based LLE [8] and the MRF model [10]. Our method yields the most visually plausible Figure 7.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Hallucinating faces from thermal infrared images

Hao

Zhang

et al. 2008

2008 15th IEEE International Conference on Image Processing

View full text Add to dashboard Cite

This paper addresses the face hallucination problem of converting thermal infrared face images into photo-realistic ones. It is a challenging task because the two modalities are of dramatical difference, which makes many developed linear models inapplicable. We propose a learning-based framework synthesizing the normal face from the infrared input. Compared to the previous work, we further exploit the local linearity in not only the image spatial domain but also the image manifolds. We have also developed a measurement of the variance between an input and its prediction, thus we can apply the Markov random field model to the predicted normal face to improve the hallucination result. Experimental results show the advantage of our algorithm over the existing methods. Our algorithm can be readily generalized to solve other multi-modal image conversion problems as well.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Markov Random Field Of Patchesmentioning

confidence: 99%

See 1 more Smart Citation

Hallucinating faces from thermal infrared images

Hao

Zhang

et al. 2008

2008 15th IEEE International Conference on Image Processing

View full text Add to dashboard Cite

show abstract

“…CCA and its variants have been applied to many different face-analysis applications, such as face and facial-expression recognition [32] [33], 3D and infrared face reconstruction [34], face super-resolution [35] [29], etc. One of the previous AAM works [36] applied CCA to efficiently model the dependency between texture residuals and model parameters in the searching step, which improves the convergence speed.…”

Section: Canonical Correlation Analysismentioning

confidence: 99%

Shape-appearance-correlated active appearance model

Zhou

Lam

2016

Pattern Recognition

View full text Add to dashboard Cite

Among the challenges faced by current active shape or appearance models, facial-feature localization in the wild, with occlusion in a novel face image, i.e. in a generic environment, is regarded as one of the most difficult computer-vision tasks. In this paper, we propose an Active Appearance Model (AAM) to tackle the problem of generic environment. Firstly, a fast face-model initialization scheme is proposed, based on the idea that the local appearance of feature points can be accurately approximated with locality constraints. Nearest neighbors, which have similar poses and textures to a test face, are retrieved from a training set for constructing the initial face model. To further improve the fitting of the initial model to the test face, an orthogonal CCA (oCCA) is employed to increase the correlation between shape features and appearance features represented by Principal Component Analysis (PCA).With these two contributions, we propose a novel AAM, namely the shape-appearance-correlated AAM (SAC-AAM), and the optimization is solved by using the recently proposed fast simultaneous inverse compositional (Fast-SIC) algorithm. Experimental results demonstrate a 5-10% improvement on controlled and semi-controlled datasets, and with around 10% improvement on wild face datasets in terms of fitting accuracy compared to other state-of-the-art AAM models.

show abstract

“…Melzer et al [11] used Canonical Correlation Analysis (CCA) to infer the pose of the object from the gray-level images, and Reiter et al's method [12] learns the depth information from RGB images also using CCA. CCA aims to find two sets of projection directions for two training sets, with the property that the correlation between the projected data is maximized.…”

Section: Related Workmentioning

confidence: 99%

Converting Thermal Infrared Face Images into Normal Gray-Level Images

Dou

Zhang

Hao

et al.

Computer Vision – ACCV 2007

View full text Add to dashboard Cite

Abstract. In this paper, we address the problem of producing visible spectrum facial images as we normally see by using thermal infrared images. We apply Canonical Correlation Analysis (CCA) to extract the features, converting a many-to-many mapping between infrared and visible images into a one-to-one mapping approximately. Then we learn the relationship between two feature spaces in which the visible features are inferred from the corresponding infrared features using Locally-Linear Regression (LLR) or, what is called, Sophisticated LLE, and a Locally Linear Embedding (LLE) method is used to recover a visible image from the inferred features, recovering some information lost in the infrared image. Experiments demonstrate that our method maintains the global facial structure and infers many local facial details from the thermal infrared images.

show abstract

3D and Infrared Face Reconstruction from RGB data using Canonical Correlation Analysis

Cited by 48 publications

References 13 publications

Hallucinating faces from thermal infrared images

Hallucinating faces from thermal infrared images

Shape-appearance-correlated active appearance model

Converting Thermal Infrared Face Images into Normal Gray-Level Images

Contact Info

Product

Resources

About