Image Processing

Pel-Recursive Motion Estimation Using the Expectation-Maximization Technique and Spatial Adaptation

by Vania V. Estrela

MULTIFRAME BLIND DECONVOLUTION, SUPER-RESOLUTION, AND SATURATION CORRECTION VIA INCREMENTAL EM

by Suvrit Sra

Text Detection of Two Major Indian Scripts in Natural Scene Images

by Aruni Roy Chowdhury

Co-authored with Dr. Ujjwal Bhattacharya and Dr. Swapan K. Parui

In this article, we present a robust scheme for detection of Devanagari or Bangla texts in scene images. These are the... more In this article, we present a robust scheme for detection of Devanagari or Bangla texts in scene images. These are the two most popular scripts in India. The proposed scheme is primarily based on two major characteristics of such texts - (i) variations in stroke thickness for text components of a script are low compared to their non-text counterparts and (ii) presence of a headline along with a few vertical downward strokes originating from this headline. We use the Euclidean distance transform to verify the general characteristics of texts in (i). Also, we apply the probabilistic Hough line transform to detect the characteristic headline of Devanagari and Bangla texts. Further, similarity and adjacency measures are applied to identify text regions, which do not satisfy the verification in (ii). The proposed approach has been simulated on a repository of 120 images taken from Indian roads and the results are encouraging. Also, we have discussed the applicability of the proposed scheme for detection of English texts. Towards this end, we have considered the training and test samples from the image database of ICDAR 2003 Robust Reading Competition.

Some observations from radiometric '8 bit'data of sediment thin sections based on alternative petrographic image analysis method

by Sabyasachi Maiti

Improved road crack detection based on one-class Parzen density estimation and entropy reduction

by José Jasnau Caeiro

Automatic Rural Road Extraction from Sets of Connected Edge Points

by José Jasnau Caeiro

Data-Driven Motion Estimation with Spatial Adaptation

by Vania V. Estrela

Alessandra Martins Coelho, Vania V. Estrela

Besides being an ill-posed problem, the pel-recursive computation of 2-D optical flow raises a
wealth of issues,... more Besides being an ill-posed problem, the pel-recursive computation of 2-D optical flow raises a
wealth of issues, such as the treatment of outliers, motion discontinuities and occlusion. Our
proposed approach deals with these issues within a common framework. It relies on the use of a
data-driven technique called Generalized Cross Validation (GCV) to estimate the best
regularization scheme for a given moving pixel. In our model, a regularization matrix carries
information about different sources of error in its entries and motion vector estimation takes into
consideration local image properties following a spatially adaptive. Preliminary experiments
indicate that this approach provides robust estimates of the optical flow

Desenvolvimento De Uma Ferramenta Computacional Para O Processamento De Imagens Estereoscópicas

by Vania V. Estrela

Blind signal separation and identification of mixtures of images

by Vania V. Estrela

Sub-pixel accuracy edge fitting by means of B-spline

by Vania V. Estrela

Estimating motion with principal component regression strategies

by Vania V. Estrela

Jensen-Bregman LogDet Divergence for Efficient Similarity Computations on Positive Defnite Tensors

by Anoop Cherian

Covariance matrices provide an easy platform for fusing multiple features compactly and as a result have found immense... more Covariance matrices provide an easy platform for fusing multiple features compactly and as a result have found immense success in several computer vision applications, including activity recognition, visual surveillance, and diffusion tensor
imaging. An important task in all of these applications is to compute the distance
between covariance matrices using a (dis)similarity function, for which the natural
choice is the Riemannian metric corresponding to the manifold inhabited by these
matrices. As this Riemannian manifold is not flat, the dissimilarities should take
into account the curvature of the manifold. As a result such distance computations
tend to slow down, especially when the matrix dimensions are large or gradients
are required. Further, suitability of the metric to enable efficient nearest neighbor
retrieval is an important requirement in the contemporary times of big data analytics. To alleviate these difficulties, this paper proposes a novel dissimilarity measure
for covariances, the Jensen-Bregman LogDet Divergence (JBLD). This divergence
enjoys several desirable theoretical properties, at the same time is computationally less demanding (compared to standard measures). To address the problem
of efficient nearest neighbor retrieval on large covariance datasets, we propose a
metric tree framework using kmeans clustering on JBLD. We demonstrate the superior performance of JBLD on covariance datasets from several computer vision
applications.

DORIS PROJECT: THE EUROPEAN DOWNSTREAM SERVICE FOR LANDSLIDES AND SUBSIDENCE RISK MANAGEMENT

by Andrea Ciampalini

International Geoscience & Remote Sensing Symposium 2012

Artificial Neural Network for License Plate Localization

by Nabil Belhaj

In this project, we developed a license plate localization system based on artificial neural networks (ANNs) ; a... more In this project, we developed a license plate localization system based on artificial neural networks (ANNs) ; a multilayer perceptron (MLP) with 2400 input units, one hidden layer containing 70 neurons and one output. Back-Propagation algorithm was used for the learning process.
The dataset used was obtained from real images issued from real automatic radars and provided by the National Traffic Division. Very good results were obtained giving a 73% of localization success.

Red Palm Weevil (Rynchophorus Ferrugineous) Recognition by Image Processing Techniques

by Ghulam Mubashar Hassan

American Journal of Agriculture and Biological Sciences, ISI indexed, Vol. 6, No. 3, July-Sep. 2011, Page 365-376. Peer reviewed.
DOI: 10.3844/ajabssp.2011.365.376

Problem statement: Red palm weevil is the most destructive insect for palm trees all over the world. This research is... more Problem statement: Red palm weevil is the most destructive insect for palm trees all over the world. This research is part of developing an automated wireless red palm weevil detection and control system. The focus for this study was to develop red palm Weevil recognition system which can detect RPW in an image and can be used in wireless image sensor network which will be part of entire proposed system. Approach: Template based recognition techniques were used. Two general recognition methods i.e., Zernike and Regional Properties and an algorithm combining them were used. Besides that, a novel technique for detecting Rostrum of RPW named as ‘Rostrum Analysis’ was proposed and used for recognition, a conclusive algorithm based on all three techniques was also proposed, 319 test images of RPW and 93 images of other insects which found in RPW habitat were used. Results: It was found that both general techniques i.e., Regional Properties and Zernike Moments methods perform reasonably in recognizing RPW. The algorithm based on both these methods performs better than individual methods. The Rostrum Analysis outperforms better than both the earlier methods and proposed algorithm using all three analytical techniques gives best results among all discussed techniques in recognizing RPW as well as other insects. Conclusion: The most balanced and efficient recognition technique is to use the proposed conclusive algorithm which is combination of Regional Properties, Zernike Moments and Rostrum Analysis techniques. The maximum time for processing an image is 0.47 sec and the results obtained in recognizing the RPW and other insects are 97 and 88% respectively.

Tools for Reordering: Commonplacing and the Space of Words in Linnaeus's Philosophia Botanica, Intellectual History Review, 20 (2010), 227-252

by Matthew Daniel Eddy

Author: Matthew Daniel Eddy

Recent studies on commonplacing have shown that it flourished as an important information management tool and, in some... more Recent studies on commonplacing have shown that it flourished as an important information management tool and, in some cases, it functioned as a method (methodus) that facilitated the ordering of natural history systems. In what follows in this essay, I wish to extend this point by examining the role played by heads in the work of Carolus Linnaeus (Carl von Linné). I address two core questions. First, what were the economies of attention that guided his commonplacing techniques? Second, what type of impact did his note-taking skills have upon the way that he spatially arranged information in texts? Whereas intellectual historians sometimes tend to focus on the role that he played as the unique originator of modern botanical and zoological classification systems, I approach his work merely as one example in a long tradition of commonplacing and graphic design that originated in the Renaissance, but which had become an indispensable organisational tool used to create knowledge systems in the leading research centres of Enlightenment Europe.

Weed Leaf Image Segmentation by Deformable Templates

by Anne-Gaelle Ausseil

A Novel Iterative Blind Deconvolution Using Morphology

by Lopamudra Kundu

Handbook for the GREAT08 Challenge: An image analysis competition for cosmological lensing

by Sreekumar Thaithara Balan

Sarah Bridle, John Shawe-Taylor, Adam Amara, Douglas Applegate, Sreekumar T. Balan, Joel Berge, Gary Bernstein, Hakon Dahle, Thomas Erben, Mandeep Gill, Alan Heavens, Catherine Heymans, F. William High, Henk Hoekstra, Mike Jarvis, Donnacha Kirk, Thomas Kitching, Jean-Paul Kneib, Konrad Kuijken, David Lagatutta, Rachel Mandelbaum, Richard Massey, Yannick Mellier, Baback Moghaddam, Yassir Moudden, Reiko Nakajima, Stephane Paulin-Henriksson, Sandrine Pires, Anais Rassat, Alexandre Refregier, Jason Rhodes, Tim Schrabback, Elisabetta Semboloni, Marina Shmakova, Ludovic van Waerbeke, Dugan Witherick, Lisa Voigt, David Wittman, Annals of Applied Statistics, 2009,vol. 3, p. 6-37

The GRavitational lEnsing Accuracy Testing 2008 (GREAT08) Challenge focuses on a problem that is of crucial importance... more The GRavitational lEnsing Accuracy Testing 2008 (GREAT08) Challenge focuses on a problem that is of crucial importance for future observations in cosmology. The shapes of distant galaxies can be used to determine the properties of dark energy and the nature of gravity, because light from those galaxies is bent by gravity from the intervening dark matter. The observed galaxy images appear distorted, although only slightly, and their shapes must be precisely disentangled from the effects of pixelisation, convolution and noise. The worldwide gravitational lensing community has made significant progress in techniques to measure these distortions via the Shear TEsting Program (STEP). Via STEP, we have run challenges within our own community, and come to recognise that this particular image analysis problem is ideally matched to experts in statistical inference, inverse problems and computational learning. Thus, in order to continue the progress seen in recent years, we are seeking an infusion of new ideas from these communities. This document details the GREAT08 Challenge for potential participants. Please visit www.great08challenge.info for the latest information.

Image Analysis for Cosmology: Results from the GREAT10 Galaxy Challenge

by Sreekumar Thaithara Balan

T. D. Kitching, S. T. Balan, S. Bridle, N. Cantale, F. Courbin, M. Gentile, M. S. S. Gill, S. Harmeling, C. Heymans, M. Hirsch, T. Kacprzak, D. Kirkby, D. Margala, R. J. Massey, P. Melchior, G. Nurbaeva, K. Patton, J. Rhodes, B. T. P. Rowe, A. N. Taylor, M. Tewes, M. Viola, D. Witherick, L. Voigt, J. Young, J. Zuntz, MNRAS, 2012

In this paper we present results from the weak lensing shape measurement GRavitational lEnsing Accuracy Testing 2010... more In this paper we present results from the weak lensing shape measurement GRavitational lEnsing Accuracy Testing 2010 (GREAT10) Galaxy Challenge. This marks an order of magnitude step change in the level of scrutiny employed in weak lensing shape measurement analysis. We provide descriptions of each method tested and include 10 evaluation metrics over 24 simulation branches. GREAT10 was the first shape measurement challenge to include variable fields; both the shear field and the Point Spread Function (PSF) vary across the images in a realistic manner. The variable fields enable a variety of metrics that are inaccessible to constant shear simulations including a direct measure of the impact of shape measurement inaccuracies, and the impact of PSF size and ellipticity, on the shear power spectrum. To assess the impact of shape measurement bias for cosmic shear we present a general pseudo-Cl formalism, that propagates spatially varying systematics in cosmic shear through to power spectrum estimates. We also show how one-point estimators of bias can be extracted from variable shear simulations. The GREAT10 Galaxy Challenge received 95 submissions and saw a factor of 3 improvement in the accuracy achieved by shape measurement methods. The best methods achieve sub-percent average biases. We find a strong dependence in accuracy as a function of signal-to-noise, and indications of a weak dependence on galaxy type and size. Some requirements for the most ambitious cosmic shear experiments are met above a signal-to-noise ratio of 20. These results have the caveat that the simulated PSF was a ground-based PSF. Our results are a snapshot of the accuracy of current shape measurement methods and are a benchmark upon which improvement can continue. This provides a foundation for a better understanding of the strengths and limitations of shape measurement methods.