Integral histogram: a fast way to extract histograms in Cartesian spaces

By Fatih Murat Porikli. Computer Vision and Pattern Recognition, 2005

826 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/fbf4d36e787e2e5c8444e1a2229b821e9cd68adf

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

A fast local descriptor for dense matching (2008, background): “This is also better than integral ima ge-like computations of histograms [19] in which all the gradient vectors contribute the same: We can very efficiently reduce the influence of gradient norms from distant locations.”
Center-surround divergence of feature statistics for salient object detection (2011, background): “Therefore, we need to estimate local distributions of basic feature results: we split every basic feature layer into an integral histogram [19].”
The PASCAL visual object classes challenge 2006 (VOC2006) results (2006, methodology): “[6] F.”
Object Tracking Benchmark (2015, methodology): “To exploit local directional edge information, histograms of oriented gradients (HOGs) [18] have been adopted for tracking [73] with the integral histogram [62].”
Spatio-temporal Depth Cuboid Similarity Feature for Activity Recognition Using Depth Camera (2013, methodology): “To reduce computational cost, we use integral histograms [17] to compute the depth histograms rapidly.”
The Pascal Visual Object Classes Challenge 2006 ( VOC 2006 ) Results (2006, methodology): “We start with our kernel-based bag of visual features approach [6].”
Real-Time Tracking via On-line Boosting (2006, methodology): “done very efficiently using integral images and integral histograms as data structures [21].”
Pedestrian Detection: An Evaluation of the State of the Art (2012, methodology): “Zhu et al. [46] sped up HOG features by using integral histograms [47].”
Pose primitive based human action recognition in videos or still images (2008, methodology): “In a first basic Matlab implementation we achieve about 1 FPS (dependent on the image resolution) in run-mod using integral histograms [ 17 ] to speed up HOG computation (without NMF computation this could be increased up to 5 FPS).”
Finding People in Images and Videos (2006, methodology): “The centre-surround HOG architecture avoids these block specific computations, which allows to be optimised for very rapid computation via integral histograms [Porikli 2005]. Centre-Surround HOGs implement an alternative centre-surround style cell normalisation scheme. They create β “orientation ima”
Covariance Tracking using Model Update Based on Lie Algebra (2006, methodology): “After constructing tensors of integral images corresponding to each feature dimension and multiplication of any two feature dimensions, the covariance matrix of any arbitrary rectangular region can be computed independent of the region size.”
On-line Boosting and Vision (2006, methodology): “Note that we use integral images and integral histograms [22] as efficient data structures, which allow a very fast calculation of all these feature types.”
Fisher and VLAD with FLAIR (2014, background): “Simple histograms can be efficiently computed with multi-dimensional integral images [19], but use prohibitive amounts of memory at higher dimensionalities.”
Region Covariance: A Fast Descriptor for Detection and Classification (2006, methodology): “We search the target image for the same locations and sizes, and fast construction of histograms are performed through integral histograms [7].”
Shape and Appearance Context Modeling (2007, methodology): “The single pass inspection of f(x) is the k-dimensional extension of the 2-dimensional version described in [25, 18].”
Visual Tracking via Locality Sensitive Histograms (2013, methodology): “They are computed at each pixel location, and have been proved to be very useful for many tasks like the bilateral filtering [25, 16], median filtering [25, 12] and tracking [15, 1].”
Semantic texton forests for image categorization and segmentation (2008, methodology): “If region r is rectangular, the histograms and class distributions can be calculated very quickly using integral histograms [24].”