Xinghui Dong · 2014
thesis or use of any of the information contained in it must acknowledge this thesis as the source of the quotation or information. This thesis evaluates the ability of computational features to estimate perceptual texture similarity. In the first part of this thesis, we conducted two evaluation experiments on the ability of 51 computational feature sets to estimate perceptual texture similarity using two differ-ent evaluation methods, namely, pair-of-pairs based and retrieval based evaluations. These experiments compared the computational features to two sets of human derived ground-truth data, both of which are higher resolution than those commonly used. The first was obtained by free-grouping and the second by pair-of-pairs experiments. Using these higher resolution data, we found that the feature sets do not perform well when compared to human judgements. Our analysis shows that these computational feature sets either (1) only exploit power spectrum information or (2) only compute higher order statistics (HoS) on, at most,
Cited by 10
Bodo Rosenhahn, Harvey Ho, Reinhard Klette · 2005
This article presents a 2D-3D pose estimation algorithm, which relies on texture information on the surface mesh of an object model. The textured surface mesh is rendered in a virtual image and a modified block matching algorithm is applied to determine correspondences between midpoints of surface patches to points in an image. This is used in a point-based 2D-3D pose estimation algorithm to determine the pose and orientation of a 3D object with respect to given image data. We present experiments on various image sequences and show advantages of the chosen approach (e.g.. in the context of varying backgrounds, noise or partial occlusions).
Cited by 0
Brian S. Bourgeois, Carley Walker · 2002
The authors investigate the use of neural networks for the direct estimation of image texture. Unlike previous approaches where networks are used to make decisions on feature vectors derived from traditional techniques, or where a network is trained to perform the function of a traditional technique, the proposed approach uses a network to directly model texture. The envisioned approaches to this method are described. Preliminary results of one-dimensional tests show that a neural network implementation is very adapt at recognizing irregular signals, even in the presence of added noise. This is intended to be applied in a Seafloor Acoustic Imagery via sidescan imagery.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>
Cited by 1
André Gerhardt · 8th International Congress of the Brazilian Geophysical Society · 2003
Over the years several orientation estimation techniques have been developed either as variants of some early methods or as part of particular attribute computations or applications. Designed with specific properties in mind, they usually yield different results for complex signals like seismic data. Most known methods generally require relatively high computational effort. Additionally, all methods are sensitive to noise, so additional pre-filtering of the data is usually required, increasing the computational cost. This paper proposes an orientation estimation method based on the autocovariance function of the data. The proposed approach results in robust and geologically meaningful orientation estimates throughout a seismic dataset. By estimating the orientation from a<br>smoothed version of the autocovariance of the data (not the data itself!), I show that it is possible to achieve better results with considerable less computational effort.
Cited by 0
Ahlborg, Carl · The Knowledge Bank (The Ohio State University) · 2018
The properties of materials are functions of their internal structure. Most engineering materials, including metals, are polycrystalline and composed of microscopic crystals. The crystallographic texture, the description of the orientations of the crystallites in a material, is a key structural indicator of the deformation behavior. Estimating the orientation distribution function (ODF) of a sample allows materials scientists to identify the probability a crystal in a sample is oriented in a certain direction. The current process, which employs a Fourier series expansion over a generalized spherical harmonic basis, leaves much to be desired and is poorly understood by much of the community that uses it. Some limitations include bias introduced by ad-hoc parameters and poor accuracy for small sample sets. The purpose of this study is to develop an algorithm to estimate the ODF using a mixture model that would be free from ad-hoc parameters. To accomplish this, the algorithm uses a mixture of symmetrized Bingham distributions. Using these distributions, it employs a Estimation Maximization (EM) approach to estimate the distribution and reevaluate the data to improve subsequent estimations. It also used a minimum message length (MML) criterion to prevent overfitting, or making too specific estimations based on insufficient data. This algorithm is compared with a similar algorithm developed in tandem using a mixture of symmetrized Bingham distributions but using a Markov Chain Monte Carlo (MCMC) approach instead.
Cited by 0Open access
Jeffrey B. Colombe · Proceedings · 2006
The surfaces of 3D objects may be represented as a connected distribution of surface patches that point in various directions with respect to the observer. Viewpoint-normal patches are those whose tangent plane is perpendicular to the line of sight. Foreshortening of surface patches results from their obliquity, with a directional wavelength compression, and an accompanying 1-dimensional stretching of the spatial frequency distribution. This stretching of spatial frequency distributions was used to generate plausible depth illusions via local foreshortening of surface textures rendered from a stretched spatial frequency envelope. Texture foreshortening cues were exploited by a multi-stage image analysis method that revealed local dominant orientation, degree of orientation dominance, relative power in spatial frequencies at a given orientation, and a measure of local surface obliquity, which provides incomplete but useful information in a multi-cue depth estimation framework.
Cited by 0
2013
Abstract-A perspective view of a slanted textured surface shows systematic changes in the density, area, and aspect-ratio of texture elements. These apparent changes in texture element properties can be analyzed to recover information about the physical layout of the scene. However, in practice it is difficult to identify texture elements, especially in images where the texture elements are partially occluded or are themselves textured at a finer scale. To solve this problem, it is necessary to integrate the extraction of texture elements with the recognition of scene layout. We present a method for identifying texture elements while simultaneously recovering the orientation of textured surfaces. A multiscale region detector, based on measurements in a V2G (Laplacian-of-Gaussian) scale-space, is used to construct a set of candidate texture elements. True texture elements are selected from the set of candidate texture elements by finding the planar surface that best predicts the observed areas of the candidate texture elements. Results are shown for a variety of natural textures, including waves, flowers, rocks, clouds, and dirt clods. Zndex Terns-Integration, multiscale structure, natural textures, perspective view, region detection, shape from texture, surface orientation, texture elements, texture gradients, texture homogeneity, threedimensional vision. I.
Cited by 0
Jonas Gårding · IEEE Transactions on Pattern Analysis and Machine Intelligence · 1993
Witkin (1981) has proposed a maximum likelihood (ML) estimator of surface orientation based on the observed directional bias of projected texture elements. However, a drawback of this procedure is that the estimate is only defined indirectly in terms of a set of nonlinear equations. An alternative method is proposed, which allows an estimate of the surface orientation to be computed directly in a single step from certain simple statistics of the image data. We also show that this direct estimate allows Witkin's ML estimate to be computed to within 0.05 degrees in only two or three iterative steps. The performance of the new estimator is demonstrated experimentally and compared to that of the ML estimator, using both synthetic data and real gray-level images.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>
Cited by 38
Anthony Sourice, Guy Plantier, Jean‐Louis Saumet · 2004
The problem of short-term texture orientation is dealt with in this paper through autocorrelation fitting. Actually, the computation of the estimated autocorrelation sequence from a textured oriented image brings about the apparition of an oriented pattern. This pattern is looked upon, in this work, as a Gaussian curve, the parameters of which give an estimated orientation. Those parameters are estimated by a two-dimensional (2-D) curve fitting on a particular region of the autocorrelation. The determination of this region is done automatically thanks to watersheds. The algorithm is developed for synthetic textured images but appears to be effective when applied to real life oriented images.
Cited by 9
Kristoffer Stensbo-Smidt, Kim Steenstrup Pedersen · 2016
The aim is to test whether texture descriptors can extract novel information about physical quantities of galaxies, which normally require spectroscopy to determine. The specific star formation rate (sSFR) is chosen as the quantity in question. Methods We use three sets of features: • Colours from five bandpass filters (no texture information). • Gradient orientation histograms computed at 8 scales, σ. • Shape index histograms computed at 8 scales, σ. Gradient orientation (GO) • GO, θ(x, y; σ), represents first order differential structure. • It is weighted by the scale normalized gradient magnitude, M(x, y; σ). • It is binned to histograms with 8 bins. θ(x, y; σ) = arctan
Cited by 0
G. Davidson, Griffiths, HD · UCL Discovery (University College London) · 2001
Cited by 0
Jun Liu, Lei Wang, Stan Z. Li · 1999
Texture classification is an important area in the field of texture analysis. In this paper, we propose a novel stochastic approach--multiresolution Markov Random Field (MRMRF) model to represent textures and a parameter estimation method based on Markov chain Monte Carlo method is proposed. The parameters estimated from the decomposed subbands can be used as features to classify textures. The classifier used here is nearest linear combination (NLC) which uses the combination of the features of several prototypes of an original texture to fit the features of the query texture. This method is better than NN(nearest neighbor) classifier. The experiment results illustrate the effectiveness of our method. 1. Introduction Texture is one of the basic characteristics of a visible surface and it provides important information for scene interpretation as well as image processing. It plays a crucial role in computer vision and pattern recognition. Understanding texture is a great part in image...
Cited by 0
Soumya Ray · 2016
Synthetic Aperture Radar (SAR) is a powerful signal processing system that finds its application in diverse fields. This system achieves finer resolution in both range and azimuth. However, coherent nature of the same makes it susceptible to signal dependent, multiplicative noise known as speckle. It’s granularity is responsible for degradation of radiometric resolution which in turn makes visual interpretation a difficult task. Furthermore, presence of such noise degrades accuracy of detector, classifier designed for SAR clutter data. This entitles the development of SAR despeckling filters, and has become a major research area over past few years. In this research work, SAR clutter data is modeled using multiplicative models such as K, W and CR-CG distributions which gives prior knowledge regarding texture and amplitude speckle statistics. This knowledge is further utilized to formulate Γ-MAP, β-MAP, CR-MAP estimate of texture amplitude which in turn is helpful for designing MAP based despeckling filters. Furthermore, estimators for the parameters associated with texture estimate using MoLC and EM algorithm are formulated. Few MMSE based despeckling filters such as Lee, Kuan are also analyzed for extracting texture from clutter data. Experiments are carried out on 1-look real SAR clutter data and 3-look synthetic data to examine the suitability and applicability of these filters in despeckling areas of diverge kind. Furthermore, effectiveness of the same is assessed in terms of mean preservation and speckle suppression using recently proposed statistical measures, MPI and MPSSI. Experimental results clearly illustrates that CR-MAP filter outperforms state-of-the-art filters in preserving mean while effectively suppressing speckle for SAR clutter data with varying degree of roughness.
Cited by 0
David J. Nott, Richard J. Wilson · Signal Processing · 1997
Cited by 20
Nan Dong, Fuqiang Liu, Zhipeng Li · 2014
This paper presents a technique for crowd density estimation in surveillance images, which needs neither individual detection and tracking nor a complex training. This is done by building a set of feature templates for different crowd density scenes, and calculating the similarity between templates and features that are extracted from surveillance video frames. These templates can be selected by staff according to the situation of surveillance scenes. Thus our approach can be deployed with minimal setup for a new site. In order to get sparse features, a generative model of sparse texture representation is improved for crowd scene description: firstly, multi-scale local image patch is generated to deal with perspective projection; secondly, a novel statistic descriptor, Gray-Gradient Dependence Matrix, is introduced to extract features; thirdly, an adaptive clustering is utilized to identify clusters. By computing the weighted average of these clusters, a more compact representation of the image can be obtained. Three aspects of experimental results show that the proposed approach is efficient and accurate in crowd density estimation.
Cited by 0
Yong Xia, Dagan Feng, Rong-Chun Zhao · IEEE Transactions on Image Processing · 2006
Multifractal analysis is becoming more and more popular in image segmentation community, in which the box-counting based multifractal dimension estimations are most commonly used. However, in spite of its computational efficiency, the regular partition scheme used by various box-counting methods intrinsically produces less accurate results. In this paper, a novel multifractal estimation algorithm based on mathematical morphology is proposed and a set of new multifractal descriptors, namely the local morphological multifractal exponents is defined to characterize the local scaling properties of textures. A series of cubic structure elements and an iterative dilation scheme are utilized so that the computational complexity of the morphological operations can be tremendously reduced. Both the proposed algorithm and the box-counting based methods have been applied to the segmentation of texture mosaics and real images. The comparison results demonstrate that the morphological multifractal estimation can differentiate texture images more effectively and provide more robust segmentations.
Cited by 103Open access
Hao Zhang, Guillaume Lopez, Ran Tao et al. · 2012
Cited by 6Open access
Mohammed En-nadhir Zighem, Abdelkrim Ouafi, Salah Eddine Bekhouche et al. · 2017
Automatic age estimation via faces is one of the dominant topics nowadays. In this paper, we proposed a novel approach for age estimation from RGB color face image. Despite the large number of methods for age estimation, their performance remains unsatisfactory for RGB images. The idea is to extract features from three color band instead of extract features from gray-scale. The experimental results show the superiority of our approach compared to that of the state-of-the-art methods that use the PAL dataset.
Cited by 0
Deguchi, Morishita · IEEE Transactions on Computers · 1978
A new approach to texture characterization and texture-based image partitioning is presented. In this approach the gray level of each pixel of an image is estimated from a weighted sum of gray levels of its neighbor pixels. The weight coefficients are determined so that the mean square estimation error is minimized. The obtained two-dimensional estimator characterizes the texture of the sample and can be used to discriminate other classes of textures. This texture characterization technique is applied to image partitioning, and two partitioning procedures are developed. Experimental results obtained by their application are also shown.
Cited by 47
Mohamed El Mounir Abdelaziz, Thierry Chonavel, Abdeldjalil Aïssa El Bey et al. · HAL (Le Centre pour la Communication Scientifique Directe) · 2013
International audience
Cited by 0