Texture Classification System Research Articles

Soil Texture Classification Algorithm Using RGB Characteristics of Soil Images

Soil texture has an important influence on agriculture, affecting crop selection, movement of nutrients and water, soil electrical conductivity, and crop growth. Soil texture has traditionally been determined in the laboratory using pipette and hydrometer methods that require a considerable amount of time, labor, and expense. Recently, in-situ soil texture classification systems using optical diffuse reflectometry or mechanical resistance have been reported, especially for precision agriculture where more data is needed than in conventional agriculture. This paper is a part of overall research to develop a soil texture classification system using image processing. Application of image processing was motivated by simple traditional approaches such as visual inspection and the “hand-feel” method. In this paper, the potential of soil texture classification using RGB histograms was investigated. Seven sites representing major Korean paddy soil series were selected, 4-6 core samples up to a 50-cm depth were obtained from each site, and each sample was segmented by 5-cm intervals. For each segmented soil sample, four surface images were taken using a miniaturized CCD camera, and texture fractions were determined by the pipette method. Scatter plots showed linear patterns between silt content and histogram variables such as brightness, skewness, and mode - brightness. When 5% averaged silt content was linearly regressed with mode - brightness, R2, RMSEC, and RMSEP were 0.96, 2.2%, and 6.3%, respectively. When soils were classified using USDA criteria, the laboratory method and the in-situ image processing method produced the same results for 48% of the samples.

An optimum feature extraction method for texture classification

Texture can be defined as a local statistical pattern of texture primitives in observer’s domain of interest. Texture classification aims to assign texture labels to unknown textures, according to training samples and classification rules. In this paper a novel method, which is an intelligent system for texture classification is introduced. It used a combination of genetic algorithm, discrete wavelet transform and neural network for optimum feature extraction from texture images. An algorithm called the intelligent system, which processes the pattern recognition approximation, is developed. We tested the proposed method with several texture images. The overall success rate is about 95%.

Optimized texture operators for the automated design of image analysis systems: Non-linear and oriented kernels vs. gray value co-occurrence matrices

The rapid development in image processing technology allows to tackle applications of increasing complexity. For efficient design of application specific systems, design automation techniques are required. This paper reports on activities for automated texture classification system design employing non-linear oriented kernels (NLOK) configured by evolutionary optimization techniques and swarm optimization (PSO). First and second order statistical features of the dominating kernels in automatically adapted regions of interests serve as features for the texture classification. Our approach was tested with benchmark and application data from leather inspection and was found to be viable and competitive in both cases. The optimized feature set was tested versus features computed from gray value co-occurrence matrices (COOC) with non-optimized parameters. The classification rates for NLOK were significantly higher than for COOC (> 75% vs. 90% vs. < 90%). Regarding the two optimization approaches, PSO outperformed genetic algorithms (GA) in optimization time and in accuracy for the benchmark data and in accuracy for application data. Additionally, the COOC parameters were adapted as well and yielded higher classification rates than without adaptation.

Comparison of wavelet-SVM and wavelet-adaptive network based fuzzy inference system for texture classification

Texture image is a low-level image feature. Nowadays, there are many different applications involving texture analysis, including endoscope, ultrasonographic images recognition, document classification, radar image classification, and texture-based image retrieval. The texture images analysis and classification becomes an important topic. The greatest hardness of texture analysis and classification in the past was the deficiency of enough methods to characterize. Multi-resolution analysis methods such as wavelet and wavelet packet decompositions are more superior to other classic statistical methods. In this paper, a comparison of wavelet-support vector machine (W-SVM) and wavelet-adaptive network based fuzzy inference system (W-ANFIS) approaches for texture image classification is presented. Both W-SVM and W-ANFIS methods are used for classification of the 22 texture images obtained from Brodatz image album. There, 50 64 x 64 image regions were randomly selected (overlapping or nonoverlapping) for each of these 22 images. 25 of these image regions and other 25 of these image regions are used for training and testing of the W-SVM and W-ANFIS methods, respectively.

Wavelet transform and adaptive neuro-fuzzy inference system for color texture classification

The wavelet domain features have been intensively used for texture classification and texture segmentation with encouraging results. More of the proposed multi resolution texture analysis methods are quite successful, but all the applications of the texture analysis so far are limited to gray scale images. This paper investigates the usage of Wavelet transform (WT) and Adaptive neuro-fuzzy inference system (ANFIS) for color texture classification problem. The proposed scheme composed of a wavelet domain feature extractor and an ANFIS classifier. Both entropy and energy features are used on wavelet domain. Different color spaces are considered in the experimental studies. The performed experimental studies show the effectiveness of the wavelet transform and ANFIS structure for color texture classification problem. The overall success rate is over 96%.

ICA-based multi-colour space texture classification system

Presented is a novel method which uses independent component analysis (ICA) for systematically partitioning and combining textural features extracted from different colour spaces, in a multiple classifier based system, for colour texture classification. Results obtained illustrate that the proposed ICA-based feature-partitioning and classifier combination system produces more accurate results compared to a system that combines classifiers applied to features extracted from individual colour spaces.

Texture classification system using colour space fusion

A novel colour texture classification system is presented based on an ensemble of independent classifiers each assigned to a different colour representation model. Gaussian Markov random fields features are used in a study to illustrate this previously unexplored approach and its potential to combine information from different colour spaces to improve accuracy. Experimental results illustrate that colour space information fusion results in significant improvements in accuracy.

Development and validation of an in vivo analysis tool to identify changes in carotid plaque tissue types in serial 3-D ultrasound scans

We have developed a three-dimensional (3-D) B-mode acquisition system suitable for imaging carotid plaques in vivo. A texture classification system using 157 statistical and textural algorithms, previously developed in our laboratory and shown to predict the contents of in vitro carotid plaques, was applied to in vivo 3-D image sets obtained from patients with both symptomatic and asymptomatic carotid artery plaques. Delineation of plaque boundaries is more difficult using in vivo images than in vitro images of excised plaques embedded in agar. This study has examined inter- and intraobserver variability studies to assess the degree of selectivity of the plaque region-of-interest (ROI) and assess the degree of repeatability for potential use in comparing serial scans. An interobserver limit of agreement of ± 12.9% and an intraobserver limit of repeatability of < 2% were obtained. These results show that the plaque ROI selection is subjective, but is repeatable within acceptable limits. (E-mail: Colemandp@cardiff.ac.uk)

Wavelet domain statistical hyperspectral soil texture classification

This communication presents an automatic soil texture classification system using hyperspectral soil signatures and wavelet-based statistical models. Previous soil texture classification systems are closely related to texture classification methods, where images are used for training and testing. In this study, we develop a novel system using hyperspectral soil textures, which provide rich information and intrinsic properties about soil textures, where two wavelet-domain statistical models, namely, the maximum-likelihood and hidden Markov models, are incorporated for the classification task. Experimental results show that these methods are both reliable and robust.

Image Descreening by GA-CNN-Based Texture Classification

This work proposes an image-descreening technique based on texture classification using a cellular neural network (CNN) with template trained by genetic algorithm (GA), called GA-CNN. Instead of using the fixed filters for image descreening, we are equipped with a more pliable mechanism for classifications in screening patterns. Using CNN makes it possible to get an accurate texture classification result in a faster speed by its superiority of implementable hardware and the flexible choices of templates. The use of the GA here helps us to look for the most appropriate template for CNNs more adaptively and methodically. The evolved parameters in the template for CNNs can not only provide a quicker classification mechanism but also help us with a better texture classification for screening patterns. After the class of screening patterns in the querying images is determined by the trained GA-CNN-based texture classification system, the recommendatory filters are induced to solve the screening problems. The induction of the classification in screening patterns has simplified the choice of filters and made it valueless to determine a new structured filter. Eventually, our comprehensive methodology is going to be topped off with more desirable results and the indication for the decrease in time complexity.

Upgrading processes in BIF-derived iron ore deposits: implications for ore genesis and downstream mineral processing

Ore textures in Australian BIF-derived ores are quite complex and have formed the basis for ore genesis models, which includes hypogene upgrading or supergene upgrading then regional metamorphism for the microplaty hematite ores and supergene upgrading for the martite-goethite ores. Despite a high level of ore texture knowledge, little is known about the spatial distribution of ore textures within deposits or their relationship to downstream processing. This paper describes a new iron ore texture classification system to understand better the vertical textural zonation characteristics of Australian BIF-derived deposits, the interpreted role of recent supergene upgrading processes on modifying primary ore textures and their implications for downstream mineral processing. The iron ore textural classification scheme presented uses textural groupings defined on the basis of similarities in mineralogy, ore texture, porosity, mineral associations and hardness. The scheme is non-genetic and has been successfully applied to deposits in the Pilbara, Yilgarn and Gawler cratons. The main ore texture groups include dense martite/hematite, microplaty hematite, microplaty hematite-goethite, martite-goethite, goethite-martite and goethite-rich. Each group can be further subdivided into physically hard to softer subcategories. The ore textural groups can be divided into those interpreted to be associated with primary replacement of BIF and secondary textures resulting from more recent modification by near-surface hydration or dehydration processes. A schematic vertical section through BIF-derived deposits consists (from the surface) of detritals underlain by a carapace, both of which may have been eroded, followed by a hydrated zone with replacement of primary ore martite and microplaty hematite by vitreous goethite. The hydrated zone is underlain by 0·2-6 m thick zones of intense dehydration and textural infilling of porosity with hematite and hydrohematite. Immediately below the dehydration zone is a strong zone of leaching, where more porous and friable ore textures occur,followed underneath by a gradational contact with primary ore. Secondary supergene processes can impact on beneficiation with hydration and leaching resulting in more difficult density separations at the concentrator stage due to reductions in density differences between ore and gangue and loss of heavy media into pores. Supergene modification of primary BIF-derived ore textures can substantially change lump metallurgical properties, with hydration and dehydration favourably increasing lump yield, but reducing resistance to thermal shock whilst leaching will give the opposing outcome. The lump properties of ore textural groups appear to be relatively consistent within and between the iron ore deposits examined.

Particle‐Size Distributions

Conventional soil texture classification systems use different definitions of particle‐size distributions (PSDs). For example, sand in the International Soil Science Society (ISSS) system equals the combined separate limits of coarse silt and sand in the USDA system. Because relationships between texture and other soil properties are affected by these differences, the ability to merge survey data in environmental studies is limited. Previous research calculated two PSD statistics, namely the geometric mean particle diameter (dg) and its standard deviation (σg), which do not depend on separate limits. We expanded the development of the PSD statistics dg and σg to compare the USDA and ISSS systems, develop relationships with soil properties, include rock fragments, and simplify the USDA texture classification to facilitate the use of soil survey data in environmental research. We found that (i) for equal clay and sand fractions, the texture of a soil sample as described by the USDA system has larger dg and σg values than in ISSS; (ii) for equal clay and sand fractions, soil samples have larger values of cation‐exchange capacity (CEC) in the ISSS than in the USDA system; (iii) small differences between some of the traditional 12 USDA classes are reflected in the dg and σg values for samples containing rocks, thereby presenting a rationale for simplification; and (iv) with this rationale, the 12 USDA classes were aggregated into five classes.

On texture classification

Texture analysis has found wide application in, say, remote sensing, medical diagnosis, and quality control. There are many ways to classify image texture and many approaches split the problem into extraction followed by classification. We describe feature extraction using the new Statistical Geometrical Features in comparison with Liu's features, features from the Fourier transform using geometrical regions, the Statistical Grey Level Dependency Matrix and the Statistical Feature Matrix. We also include a formal analysis concerning rotational-invariance in the Statistical Geometric Features. Classification techniques considered here include the K-Nearest Neighbour Rule, the Error Back-propagation method and the new Generating-Shrinking Algorithm. A particular consideration is scale-invariance in the feature space since this implies that textures can be classified as the same, even when the overall illumination level differs. Experimental evaluation on the whole Brodatz texture set shows that the Statistical Geometrical Features can give the best performance for all the considered classifiers, that the Genera ting-Shrinking Algorithm can offer better performance over the Error Back-Propagation method and that the K-Nearest Neighbour Rule's performance is comparable with that of the Generating-Shrinking Algorithm. Also, the combination of the Statistical Geometrical Features with the Generaling-Shrinking Algorithm constitutes one of the best texture classification systems considered.

Markov random fields for texture classification

Texture features obtained by fitting generalized Ising, auto-binomial, and Gaussian Markov random fields (MRFs) to homogeneous textures are evaluated and compared by visual examination and by standard pattern recognition methodology. The MRF model parameters capture the strong cues for human perception, such as directionality, coarseness, and/or contrast. This paper is a comparative study of MRF model-based features. A comparison of classifying natural textures and sandpaper textures using nearest neighbor (NN), quadratic, and Fisher classifiers, suggests that both texture feature extraction and classifier design should be simultaneously considered in designing an optimal texture classification system.

Feature-based texture classification of side-scan sonar images using a neural network approach

A texture classification system for side-scan sonar images by using a trained multilayer feedforward neural network (MFNN) is presented. The system classified textures by exploiting principal feature patterns, giving a high correct-classification rate. Experimental examples for the classification of side-scan sonar images are provided.

Texture classification using the cortex transform

Texture classification has been the subject of extensive research during the last several years. The key step in any texture classification process is the choice of the set of features to be used to characterize the texture. In this paper, the strategy for developing a classification system is to consider texture classification as performed by the human visual system; therefore the feature extraction stage is based on the cortex transform. A new and simple algorithm for implementing the cortex transform is developed; a texture classification system based on it is evaluated. The initial classification results are promising and appear to be robust with respect to a noisy environment.

Texture Profile Method

SUMMARYA texture profile method was developed that uses the A. D. Little flavor profile method as a model. A texture profile is defined as the organoleptic analysis of the texture complex of a food in terms of its mechanical, geometrical, fat, and moisture characteristics, the degree of each present, and the order in which they appear from first bite through complete mastication. The texture profile analysis requires a panel of judges with prior knowledge of the texture classification system, use of standard rating scales, and proper panel procedures with regard to the mechanics of testing and sample control.