Hamidullah Binol

Kernel Fukunaga–Koontz Transform Subspaces for Classification of Hyperspectral Images With Small Sample Sizes

In this letter, a novel supervised classification approach is presented for the classification of hyperspectral images using kernel Fukunaga-Koontz transform (KFKT). The Fukunaga-Koontz transform (FKT) is originally a powerful target detection method used in remote sensing tasks, and it is an especially good classification tool for two-class problems. The traditional FKT method has been kernelized for increasing the nonlinear discrimination ability and capturing higher order of statistics of data. The proposed approach in this letter aims to solve the multiclass problem by regarding one class as target that is tried to be separated from the remaining classes (as clutter) like one-against-all methodology. The KFKT provides superior performance in the classification of hyperspectral data even using small number of samples because of nonlinear separability of data in higher dimensional space. The experiments confirm that KFKT has better and promising results than FKT and support vector machine in classification of hyperspectral images.

A expert system for stomach cancer images with artificial neural network by using HOG features and linear discriminant analysis: HOG_LDA_ANN

In this study, normal (n), benign (b), and malign (m) stomach image cells have taken from faculty of Medicine the Fırat University with Light Microscope help. Total number of stomach images are 180 which be 60 n, 60 b, and 60 m. 90 of these 180 stomach images have been used for testing purposes and 90 have used for training purposes. The histograms of oriented gradient (HOG) feature vectors have been obtained for normal, benign, and malign original stomach images. The size of these HOG feature vectors is 46900×180. High-dimensional of these HOG feature vectors is reduced to lower-dimensional with Linear Discriminant Analysis (LDA). These low-dimensional data are 180×180. These low-dimensional data are classified as normal benign and malign by artificial neural network (ANN) classification. Thus, HOG_LDA_ANN method for stomach cancer images have developed. Diagnostic accuracy of classification results with this method has found as 88.9%. According to the other methods, this result has higher accuracy result. And this result has found in a shorter time.

A heuristic-based band selection approach to improve classification accuracy in hyperspectral images

Variable Neighborhood Search (VNS) is one of the methods, called metaheuristic, which are based on searching the solution space quickly to get optimal or approximately optimal solution. This method is based on the systematically neighborhood change in search area and generally used to achieve the optimal solution in a short time in high dimensional search space. Examining the data including large scale of information such as hyperspectral images and eliminating redundant features (bands) is quite important for computation time and target classification/detection performance. In this study, band selection as a dimension reduction procedure is employed to hyperspectral images using VNS method. Then the classification was done for different selections of the spectral bands with the spectral angle mapper (SAM) and support vector machine (SVM) on hyperspectral Indian Pine image. The experimental results show that the VNS-based dimension reduction algorithm can improve classification performance in high dimensional hyperspectral data.

Classification on hyperspectral images using enhanced covariance descriptor

Pattern classification is a vital area of computer vision. Classification of hyperspectral images is difficult and complex due to their high-dimensional characteristics. Covariance descriptor is often used in the area of pattern recognition on 2-dimensional images. In this study, we propose a different approach to classical covariance descriptor in hyper-spectral image classification. The proposed approach covers partial covariance matrix with efficient features instead of classical method. The performance of new approach is compared with the recent work in that area. We used AVIRIS hyperspectral data for implementations.

Recognition of the stomach cancer images with probabilistic HOG feature vector histograms by using HOG features

In this study, normal (n), benign (b), and malign (m) stomach image cells have taken from faculty of Medicine the Fırat University with Light Microscope help. Total number of stomach images are 180 which be 60 n, 60 b, and 60 m. 90 of these 180 stomach images have been used for testing purposes and 90 have been used for training purposes. The histograms of oriented gradient (HOG) feature extraction method were used for these images. HOG feature vectors were obtained by plotting HOG features on normal, benign, and malign original stomach images. Using these HOG property vectors, histograms of normal, benign, and malignant stomach images were plotted. Bins and h histogram values were obtained from these drawn histograms. A bandwidth range that can be distinguished between normal, benign, and malignant stomach images was calculated by comparing the bins and h values obtained for normal (n), benign (b) and malign (m) images. This bandwidth range was found to be 0.09–0.22. According to this bandwidth range, the accuracy result of stomach cancer images is found as 100%. When the h values of the HOG feature vector between these bandwidths are examined, the h values of normal and benign stomach images are found to be higher than those of a malignant stomach image. Between this bandwidth, the h value of the normal stomach image was found to be higher than the benign stomach image.

Food inspection using hyperspectral imaging and SVDD

Nowadays food inspection and evaluation is becoming significant public issue, therefore robust, fast, and environmentally safe methods are studied instead of human visual assessment. Optical sensing is one of the potential methods with the properties of being non-destructive and accurate. As a remote sensing technology, hyperspectral imaging (HSI) is being successfully applied by researchers because of having both spatial and detailed spectral information about studied material. HSI can be used to inspect food quality and safety estimation such as meat quality assessment, quality evaluation of fish, detection of skin tumors on chicken carcasses, and classification of wheat kernels in the food industry. In this paper, we have implied an experiment to detect fat ratio in ground meat via Support Vector Data Description which is an efficient and robust one-class classifier for HSI. The experiments have been implemented on two different ground meat HSI data sets with different fat percentage. Addition to these implementations, we have also applied bagging technique which is mostly used as an ensemble method to improve the prediction ratio. The results show that the proposed methods produce high detection performance for fat ratio in ground meat.

Covariance descriptor fusion for target detection

Target detection is one of the most important topics for military or civilian applications. In order to address such detection tasks, hyperspectral imaging sensors provide useful images data containing both spatial and spectral information. Target detection has various challenging scenarios for hyperspectral images. To overcome these challenges, covariance descriptor presents many advantages. Detection capability of the conventional covariance descriptor technique can be improved by fusion methods. In this paper, hyperspectral bands are clustered according to inter-bands correlation. Target detection is then realized by fusion of covariance descriptor results based on the band clusters. The proposed combination technique is denoted Covariance Descriptor Fusion (CDF). The efficiency of the CDF is evaluated by applying to hyperspectral imagery to detect man-made objects. The obtained results show that the CDF presents better performance than the conventional covariance descriptor.

Supervised target detection in hyperspectral images using one-class Fukunaga-Koontz Transform

A novel hyperspectral target detection technique based on Fukunaga-Koontz transform (FKT) is presented. FKT offers significant properties for feature selection and ordering. However, it can only be used to solve multi-pattern classification problems. Target detection may be considered as a two-class classification problem, i.e., target versus background clutter. Nevertheless, background clutter typically contains different types of materials. That’s why; target detection techniques are different than classification methods by way of modeling clutter. To avoid the modeling of the background clutter, we have improved one-class FKT (OC-FKT) for target detection. The statistical properties of target training samples are used to define tunnel-like boundary of the target class. Non-target samples are then created synthetically as to be outside of the boundary. Thus, only limited target samples become adequate for training of FKT. The hyperspectral image experiments confirm that the proposed OC-FKT technique provides an effective means for target detection.

A supervised discriminant subspaces-based ensemble learning for binary classification

To enable feature extraction and reduction in pattern recognition applications, discriminant subspace analysis-based algorithms are used. Among the better-known discriminant subspace techniques for two-pattern recognition is the Fukunaga-Koontz transform (FKT). This technique has been modified to a non-linear version with the aid of kernel machines. This has enabled an increase in its non-linear discrimination ability, apart from securing higher statistics of data. The performance of kernel FKT (KFKT) is, however, dependent on the choice of suitable kernels and their inherent parameters. The aim of this paper is to ascertain the difficulties of ensemble learning with a finite set of base kernels on FKT subspaces. The study presents a new approach to tackling the issues of multiple kernel learning (MKL) on FKT. For this, a better kernel function is designed by either linearly or non-linearly combining numerous pre-chosen kernels into the algorithm. KFKTs were used with sub-kernel learners with a diverse set of kernels, each with different parameters. Weighted and unweighted fusions were employed in order to combine the predictions of sub-learners. The eventual results proved that the ensemble of KFKTs was far better than the single KFKTs as far as classification performance went.

An efficient classifier design for remote sensing hyperspectral imagery

Among the various classifiers, the Support Vector Data Description (SVDD) is a well-known strong classifier since it uses nonparametric boundary approach that constructs the minimum hypersphere enclosing the target objects as much as possible. The SVDD has been used in many studies for classification, anomaly and target detection problems on airborne or spaceborne remote sensing hyperspectral images (HSI). In this paper, we have designed an efficient classifier using ensemble method with SVDD. As an ensemble approach, we have selected bagging technique with majority voting. To verify the performance improvement, we have tested the proposed classifier for Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) hyperspectral data. AVIRIS is a proven instrument in the realm of Earth remote sensing and has been flown on airborne platforms. The results show that the ensemble method based on bagging produces better performance than the conventional SVDD.