PCA-SIFT: a more distinctive representation for local image descriptors

By Yan Ke, Rahul Sukthankar. 2004

3,915 citations recorded by the source corpus.

Subfields: 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 Computer Vision and Pattern Recognition

Citations in the corpus

Fisher Vectors Derived from Hybrid Gaussian-Laplacian Mixture Models for Image Annotation (2014, methodology): “Therefore, and since PCA suffers from the implicit assumption of underlying Gaussian Distribution [16], the Independent Component Analysis (ICA) [22] is chosen in our experiments.”
Merging chrominance and luminance in early, medium, and late fusion using Convolutional Neural Networks (2015, methodology): “Other features are available but are not discussed here, some examples are the Gradient location-orientation histogram (GLOH) [Mikolajczyk and Schmid, 2005] with 17 locations, 16 orientation bins in a log-polar grid and uses Principal Component Analysis (PCA) to reduce the dimensions to 128; the PCA”
A sparse object category model for efficient learning and exhaustive recognition (2005, methodology): “Inspired by the performance of PCA-SIFT in region matching [12], we utilize a gradient-based PCA approach in contrast to the intensity-based PCA approach of [6].”
Image retrieval based on micro-structure descriptor (2011, neutral): “Many of them are distribution-based, such as the SIFT descriptor [19], PCA-SIFT descriptor [20], GLOH descriptor [21], SURF descriptor [22], shape context [23], generalized correlograms [24] and CS-LBP algorithm [25].”
The Pascal Visual Object Classes Challenge 2006 ( VOC 2006 ) Results (2006, methodology): “Those initial negative sets were then semi-automatically enriched by the iterative procedure described in [2] (i.”
The devil is in the details: an evaluation of recent feature encoding methods (2011, methodology): “Given a set x 1 , . . . , x N ∈ R D of N training descriptors (to which PCA reduction may have been applied), k-means seeks K vectors µ 1 , . . . , µ K ∈ R D and a data-to-means assignments q 1 , . . . , q N ∈ {1, . . . , K} such that the cumulative approximation error ∑ N i=1 x i − µ q i 2 is minim”
MIFS-ND: A mutual information-based feature selection method (2014, methodology): “…attribute evaluation techniques, such as ReliefF (Kira & Rendell, 1992), Chi squared (Liu & Setiono, 1995), Correlation Feature Selection (CFS) (Hall & Smith, 1999) and Principal components analysis (Ke & Sukthankar, 2004), also have been provided in the Weka software platform (Hall et al., 2009).”
Automatic landslide detection from remote-sensing imagery using a scene classification method based on BoVW and pLSA (2013, methodology): “The detected keypoints are represented by local descriptors such as the scale-invariant feature transform (SIFT) descriptor (Lowe 2004) and the PCA-SIFT descriptor, etc. (Ke and Sukthankar 2004).”
A Survey of Augmented Reality (2015, methodology): “, 2006], which uses approximations of the DoG and orientation histograms algorithms to increase feature detection and feature description speeds, and PCA-SIFT [Ke and Sukthankar, 2004], which uses principal component analysis to reduce the descriptor length from 128 to 36 to increase feature matchin”
Bags of Binary Words for Fast Place Recognition in Image Sequences (2012, methodology): “Apart from GPU implementations [16], there are other similar features that try to reduce this computation time by, for example, approximating the SIFT descriptor [17] or reducing the dimensionality [18].”
Evaluating bag-of-visual-words representations in scene classification (2007, methodology): “The detected keypoints are depicted using the PCA-SIFT descriptor, which is a 36-dimensional real-valued feature vector [5].”
An efficient parts-based near-duplicate and sub-image retrieval system (2004, methodology): “For interest point representation, we use PCA-SIFT [11], a local descriptor that has been shown to be both more distinctive and compact than the original SIFT [13] descriptor.”
VisualRank: Applying PageRank to Large-Scale Image Search (2008, background): “Matching local descriptors efficiently has received tremendous research attention in recent years [32], [33], [34], [35], [36], [37].”
Detection and Localization of Multiple Objects (2006, methodology): “We adapted the PCA-SIFT implementation used by [1].”