Co-occurrence Matrix Approach Research Articles

An efficient multi-RVM classification-based ultrasound lung image retrieval approach

This paper proposes an efficient Multi-Relevance Vector Machine (multi-RVM) classification-based ultrasound image retrieval approach for retrieval of lung images relevant to the query image. In our proposed work, a hybrid median filter is used for filtering the training and testing ultrasound lung image. Extraction of feature in the lung image is performed by using the Tamura features and convoluted grey-level co-occurrence matrix approach. The particle swarm optimisation combined differential evolution feature selection approach performs selection of minimum set of features relevant to the query image. Multi-RVM-based classification technique is used to identify the types of lung diseases. Finally, the Hamming-distance-based retrieval technique performs retrieval of similar and relevant lung images from the database. From the performance analysis result, it is clearly evident that the proposed approach achieves better performance in terms of accuracy, sensitivity and specificity, when compared to the existing classification techniques.

Pattern based Dimensionality Reduction Model for Age Classification

The two most popular statistical methods used to measure the textural information of images are the Grey Level Cooccurrence Matrix (GLCM) and Texture Units (TU) approaches. The novelty of the present paper is, it combines TU and GLCM features by deriving a new model called “Pattern based Second order Compressed Binary (PSCB) image” to classify human age in to four groups. The proposed PSCB model reduces the given 5 x 5 grey level image into a 2 x 2 binary image, while preserving the significant features of the texture. The proposed method intelligently compressed a 5x5 window into a 2x2 window and derived TU on them. Thus the derived TU also represents a TU of a 5x5 window. The TU of the proposed PSCB model ranges from 0 to 15, thus it overcomes the previous disadvantages in evaluating TU’s.

Quantitative, Simultaneous PET/MRI for Intratumoral Imaging with an MRI-Compatible PET Scanner

Noninvasive methods are needed to explore the heterogeneous tumor microenvironment and its modulation by therapy. Hybrid PET/MRI systems are being developed for small-animal and clinical use. The advantage of these integrated systems depends on their ability to provide MR images that are spatially coincident with simultaneously acquired PET images, allowing combined functional MRI and PET studies of intratissue heterogeneity. Although much effort has been devoted to developing this new technology, the issue of quantitative and spatial fidelity of PET images from hybrid PET/MRI systems to the tissues imaged has received little attention. Here, we evaluated the ability of a first-generation, small-animal MRI-compatible PET scanner to accurately depict heterogeneous patterns of radiotracer uptake in tumors. Quantitative imaging characteristics of the MRI-compatible PET (PET/MRI) scanner were evaluated with phantoms using calibration coefficients derived from a mouse-sized linearity phantom. PET performance was compared with a commercial small-animal PET system and autoradiography in tumor-bearing mice. Pixel and structure-based similarity metrics were used to evaluate image concordance among modalities. Feasibility of simultaneous PET/MRI functional imaging of tumors was explored by following (64)Cu-labeled antibody uptake in relation to diffusion MRI using cooccurrence matrix analysis. The PET/MRI scanner showed stable and linear response. Activity concentration recovery values (measured and true activity concentration) calculated for 4-mm-diameter rods within linearity and uniform activity rod phantoms were near unity (0.97 ± 0.06 and 1.03 ± 0.03, respectively). Intratumoral uptake patterns for both (18)F-FDG and a (64)Cu-antibody acquired using the PET/MRI scanner and small-animal PET were highly correlated with autoradiography (r > 0.99) and with each other (r = 0.97 ± 0.01). On the basis of these data, we performed a preliminary study comparing diffusion MRI and radiolabeled antibody uptake patterns over time and visualized movement of antibodies from the vascular space into the tumor mass. The MRI-compatible PET scanner provided tumor images that were quantitatively accurate and spatially concordant with autoradiography and the small-animal PET examination. Cooccurrence matrix approaches enabled effective analysis of multimodal image sets. These observations confirm the ability of the current simultaneous PET/MRI system to provide accurate observations of intratumoral function and serve as a benchmark for future evaluations of hybrid instrumentation.

Feature Extraction and Texture Classification in MRI

Automated MRI (Magnetic resonance Imaging) brain tumor segmentation is a difficult task due to the variance and complexity of tumors. In this paper, a statistical structure analysis based tumor segmentation scheme is presented, which focuses on the structural analysis on both tumorous and normal tissues. The basic concept is that local textures in the images can reveal the typical ‘regularities’ of biological structures. Thus, textural features have been extracted using co-occurrence matrix approach. By the analysis of level of correlation we can reduce the number of features to the only significant component .An artificial neural network and fuzzy c-means are used for classification. This approach is designed to investigate the differences of texture features among macroscopic lesion white matter (LWM), normal appearing white matter (NAWM) in magnetic resonance images (MRI) from patients with tumor and normal white matter (NWM).

A new approach for finding an appropriate combination of texture parameters for classification

This article is an attempt to suggest a new approach for eliminating the lengthy process of selecting various parameters for extracting texture features and to quantify the relative importance of the parameters affecting textural classification. A multivariate data analysis technique called ‘conjoint analysis’ has been used in the study to analyse the relative importance of these parameters. Results indicate that the choice of texture feature and window size have higher relative importance in the classification process than quantization level or the choice of image band for extracting texture feature. Results of the classification of an Indian urban environment using spatial property (texture) have also been reported. It was observed that the classification incorporating texture features using grey level co-occurrence matrix and wavelet-based approach improves the overall accuracy in a statistically significant manner in comparison to pure spectral classification.

Fractional differential approach to detecting textural features of digital image and its fractional differential filter implementation

This paper mainly discusses fractional differential approach to detecting textural features of digital image and its fractional differential filter. Firstly, both the geometric meaning and the kinetic physical meaning of fractional differential are clearly explained in view of information theory and kinetics, respectively. Secondly, it puts forward and discusses the definitions and theories of fractional stationary point, fractional equilibrium coefficient, fractional stable coefficient, and fractional grayscale co-occurrence matrix. At the same time, it particularly discusses fractional grayscale co-occurrence matrix approach to detecting textural features of digital image. Thirdly, it discusses in detail the structures and parameters of n×n any order fractional differential mask on negative x-coordinate, positive x-coordinate, negative y-coordinate, positive y-coordinate, left downward diagonal, left upward diagonal, right downward diagonal, and right upward diagonal, respectively. Furthermore, it discusses the numerical implementation algorithms of fractional differential mask for digital image. Lastly, based on the above-mentioned discussion, it puts forward and discusses the theory and implementation of fractional differential filter for digital image. Experiments show that the fractional differential-based image operator has excellent feedback for enhancing the textural details of rich-grained digital images.

An Ordinal Co-occurrence Matrix Framework for Texture Retrieval

We present a novel ordinal co-occurrence matrix framework for the purpose of content-based texture retrieval. Several particularizations of the framework will be derived and tested for retrieval purposes. Features obtained using the framework represent the occurrence frequency of certain ordinal relationships at different distances and orientations. In the ordinal co-occurrence matrix framework, the actual pixel values do not affect the features, instead, the ordinal relationships between the pixels are taken into account. Therefore, the derived features are invariant to monotonic gray-level changes in the pixel values and can thus be applied to textures which may be obtained, for example, under different illumination conditions. Described ordinal co-occurrence matrix approaches are tested and compared against other well-known ordinal and nonordinal methods.

Textural analysis of IRS-1D panchromatic data for land cover classification

Manifestation of texture in high spatial resolution optical satellite data has an inherent potential to provide land cover information. A study has been conducted to explore this utility with IRS-1D panchromatic (PAN) data through its grey values and the derived textural information. A typical test site with a wide range of textures was subjected to textural analysis by the Grey Level Co-occurrence Matrix (GLCM) approach. Entropy at an Inter Pixel Distance (IPD) of 1 and Correlation at an IPD of 9 were found to be the optimised textural features. The data has been independently classified using PAN grey levels, textural information, and PAN combined texture, which showed significantly improved classification accuracy with the addition of textural information. Similar results were obtained over two other independent sites, validating the consistent performance of the optimised textural features.

Texture anisotropy in 3-D images

Two approaches to the characterization of three-dimensional (3-D) textures are presented: one based on gradient vectors and one on generalized co-occurrence matrices. They are investigated with the help of simulated data for their behavior in the presence of noise and for various values of the parameters they depend on. They are also applied to several medical volume images characterized by the presence of microtextures and their potential as diagnostic tools and tools for quantifying and monitoring the progress of various pathologies is discussed. No firm medical conclusions can be drawn as not enough clinical data are available. The gradient based method appears to be more appropriate for the characterization of microtextures. It also shows more consistent behavior as a descriptor of pathologies than the generalized co-occurrence matrix approach.

Noise robustness of texture features

This note examines the noise robustness of two sets of texture features, one set derived from the popular multichannel filtering approach, and the other from the benchmark grey level co-occurrence matrix approach. Comparative experimental results are presented. The results clearly demonstrate the superiority of the multichannel approach.

Statistical methods to compare the texture features of machined surfaces

Texture studies play a paramount role in many image processing applications. In this paper an attempt is made to study the textural features of machined surfaces (grinding, milling and shaping) using the most widely used statistical methods, viz. co-occurrence matrix approach, the amplitude varying rate statistical approach (AVRS) and the run length matrix approach. Textural features derived from these matrices are studied and analysed. A new matrix for the qualitative evaluation of surfaces, namely the gray-level difference-pixel distance matrix, is presented and its usefulness in texture analysis is analysed. The features calculated from these matrices are correlated with surface parameters, such as roughness, and the different features are studied for classification of these surfaces.

Vector choice in the texture spectrum approach

Abstract Recently the texture spectrum approach has been proposed as a statistical method for texture analysis, and applied to remotely sensed images. In the present study, this method is generalized to a space of M vectors and N grey level intervals for the elements of texture units, instead of M = 8 and N = 3 in the earlier studies of the texture spectrum. In this way, the texture unit set can be defined in a neighbourhood of 3 pixels by 3 pixels, 5 pixels by 5 pixels, 7 pixels by 7 pixels or other forms and sizes, and the co-occurrence matrix approach is unified to the texture spectrum method with the extreme case of M = 1. Several combinations of M and N have been evaluated to classify an imagery composed of six natural textures. The results show maximum discrimination for M = 5, followed by M = 4. In this way we minimize the calculation lime needed to maximize the accuracy of the classification.

Un nouvel espace de texture

Abstract Based on the texture spectrum approach, a new texture space is presented. In this way, a texture can be defined and represented by a space of N dimensions with M grey levels. In this generalized method, the co-occurrence matrix approach and the texture spectrum are the two extreme cases of N =2 and N =9, respectively. The texture features, usually used in the texture analysis domain, can be redefined in this new texture space.

An efficient threshold-evaluation algorithm for image segmentation based on spatial graylevel co-occurrences

The co-occurrence matrix approach, considering locally spatial variation of graylevels in image data, has demonstrated its application in image thresholding. Diverse measures based upon this type of matrices were proposed for selecting optimal thresholds by fully evaluating all possible values. We explore computational parallelism and present a modified computing procedure that is easily hardware-implemented and more efficient in both time and space complexities. Analysis shows that our algorithm is most promising for locally adaptive thresholding.

Texture features based on texture spectrum

Texture is an important spatial feature useful for identifying objects or regions of interest in an image. This paper presents a new set of textural measures derived from the texture spectrum. The proposed features extract textural information of an image with a more complete respect of texture characteristics (in all the eight directions instead of only one displacement vector used in the co-occurrence matrix approach). Some of Brodatz's natural texture images have been used for evaluating the performance of the features in the characterization and discrimination of the texture aspects of images. Some of the traditional texture features, extracted from the co-occurrence matrix, have also been used for comparison with the new measures in the discrimination of different lithological units of an airborne synthetic aperture radar image. Promising results are obtained and presented in this paper.

Cloud field classification based upon high spatial resolution textural features: 2. Simplified vector approaches

The huge volume of data being collected in global climate studies makes it necessary to develop efficient automatic data analysis methods. While most cloud classification algorithms are based upon multispectral signatures, there is growing use of textural features. The results given in Part 1 of this study demonstrate that textural features computed from the Gray Level Cooccurrence Matrix (GLCM) approach produce high cloud classification accuracies. The present study compares classification results derived from two vector approaches, Sum and Difference Histogram (SADH) and Gray Level Difference Vector (GLDV), with those from the GLCM approach. It is found that the SADH approach produces accuracies equivalent to those obtained using GLCM, but with greater ability to resolve error clusters; also, there is a 30% savings in run time and a 50% savings in storage requirements. The GLDV approach suffers a slight degradation in classification accuracy but has a 40% savings in run time and an 87% savings in storage requirements. Textural features are not highly sensitive to moderate variations in cloud threshold selection. However, the whole cloud, rather than only the brightest portions of the cloud, produce the highest classification accuracies. A very important result is that spatial information content and classification accuracy are preserved even at lower radiometric resolutions with effective gray levels of 16. means that significantly low resolution digitized versions of satellite imagery retain essentially the full spatial information content of the original digital data. Substitution of digitized imagery can significantly reduce the expense of many remote sensing studies.

Characterization of textures by ‘Eigenfilters’

The grey tones of all pixels of a 3 × 3 neigbourhood are combined into a feature vector which is thought to be attached to the central pixel. After calculation of the corresponding variance-covariance matrix over a homogeneously textured region the eigenvectors of the matrix are obtained. They can be interpreted as adapted filter masks for which the designation ‘eigenfilters’ is proposed. For some applications they could replace the empirical filters in Laws ‘texture energy’ measures. It is shown how the approach using eigenfilters relates to Laws' approach and to the co-occurrence matrix approach.