RASL: Robust alignment by sparse and low-rank decomposition for linearly correlated images

By YiGang Peng, Arvind Ganesh, John Wright, Wenli Xu, Yi Ma. 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2010

1,045 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computational Mechanics, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/3a6c9b81cc77f2d66dfec3d567a125b526c58ddd

How researchers see this paper

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

In Aerospace Engineering

In Computational Mechanics

In Computer Vision and Pattern Recognition

Citations in the corpus

Real Image Denoising With Feature Attention (2019, methodology): “Model-based algorithms include non-local self-similarity (NSS) [16, 11, 18], sparsity [27, 45], gradient methods [43, 52, 50], Markov random field models [48], and external denoising priors [8, 57, 39].”
A collaborative framework for 3D alignment and classification of heterogeneous subvolumes in cryo-electron tomography (2010, background): “, 2008) and (Peng et al., 2012), which was designed for robustly aligning linearly correlated heterogeneous images as a part of batch image preprocessing for compression, pattern recognition and data mining tasks.”
Parallel Multi-Block ADMM with o(1 / k) Convergence (2013, neutral): “See [11, 1, 3, 29, 4, 24, 22, 31, 23] and the references therein for a number of examples.”
GAN-Supervised Dense Visual Alignment (2021, background): “Subsequent work in this area assumes the data lies on a low-rank subspace [43, 64] or factorizes images as a composition of color, appearance and shape [61] to establish dense correspondences between instances of the same object category.”
Moving Object Detection by Detecting Contiguous Outliers in the Low-Rank Representation (2011, background): “Here we use the incremental refinement [57], [60] to solve thi s parametric motion estimation problem: at each iteration, we update τ̂ by a small increment ∆τ and linearizeD ◦ τ asD ◦ τ̂ + Jτ̂∆τ , whereJτ̂ denotes the Jacobian matrix ∂D ∂τ |τ=τ̂ .”
Robust Visual Tracking via Structured Multi-Task Sparse Learning (2012, background): “(ii) Columns of X can be aligned to columns of Dt as in Peng et al. (2012) to solve the geometric transformation.”
An Edge Traffic Flow Detection Scheme Based on Deep Learning in an Intelligent Transportation System (2021, methodology): “To do so, various structure compression and algorithm acceleration models were proposed, e.g., parameter pruning [28], parameter sharing [29], low-rank decomposition [30], compact convolution kernel [31] and knowledge distillation [32].”
Multi-label Learning with Missing Labels Using Mixed Dependency Graphs (2018, methodology): “= arg min where we deﬁne denotes the singular value soft-thresholding operator (Peng et al. 2012), utilizing the soft-thresholding operator S λ ( B ij ) = sign ( B ij ) max ( 0 , | B ij |− λ) and the SVD decomposition wherethesoft-thresholdingoperator S γ 1 ρ 1 ( · ) and E aredeﬁned as above.”
Divide-and-Conquer Matrix Factorization (2011, background): “Examples of the matrix completion problem include collaborative filt ering for recommender systems, link prediction for social networks, and click prediction for web se arch, while applications of robust matrix factorization arise in video surveillance [2], graphical m odel selection [4], document m”
Competition on counter measures to 2-D facial spoofing attacks (2011, methodology): “Classifier 1 - Non-Rigid Motion: The nonrigid motions in real faces are detected by a batch image alignment technique proposed in [21], called “RASL”.”
Image classification by non-negative sparse coding, low-rank and sparse decomposition (2011, background): “Over the past few years, the low-rank matrix recovery problem has been widely studied [22-25] and successfully applied to many applications, such as image processing [22], web data mining [26], and bioinformatic data analysis [27].”
Dense Semantic Correspondence Where Every Pixel is a Classifier (2015, methodology): “[15] present a parametric approach to the general alignment problem, in a similar vein to congealing [3] or RASL [16].”
Weighted Nuclear Norm Minimization and Its Applications to Low Level Vision (2016, methodology): “2011) has been successfully utilized inmany computer vision tasks (Ji et al. 2010; Liu et al. 2010; Zhang et al. 2012b; Liu et al. 2012; Peng et al. 2012).”
Nuclear Norm Based Matrix Regression with Applications to Face Recognition with Occlusion and Illumination Changes (2014, methodology): “where ~ A is the aligned training set, which can be calculated through an alignment method as suggested in [52].”