Locality Sensitive Discriminant Analysis Research Articles

A New Hyperspectral Image Identification Method Based on LSDA and OSELM

To solve the problems of information loss and nonlinear transformation of hyperspectral image dimensionality reduction methods and the problem that recognition methods are sensitive to noise and cannot be trained online, a new hyperspectral image identification method based on locality sensitive discriminant analysis (LSDA) and online sequential extreme learning machine (OSELM) is proposed. Firstly, LSDA is used to reduce the redundant information and data noise, and the local structure and discriminant information of the hyperspectral image are extracted on the basis of reducing the data dimension. Then, a real‐time recognition model based on OSELM is constructed to identify various terrain targets in hyperspectral images through online real‐time updates of data. According to the experimental results of the Pavia University dataset and Salinas dataset, the proposed LSDA‐OSELM not only achieved the maximum overall accuracy (OA) of 97.56 and 97.88 but also obtained the maximum kappa coefficient of 0.9726 and 0.9774. Compared with several new dimensionality reduction methods and identification methods, the proposed LSDA‐OSELM has achieved a leading position in both OA and kappa indicators. The proposed method improves the generalization ability and accuracy of hyperspectral image target recognition models. In the future, its computational complexity and parameter robustness are expected to be further improved and applied in practical application scenarios.

A Novel Computer-Vision Approach Assisted by 2D-Wavelet Transform and Locality Sensitive Discriminant Analysis for Concrete Crack Detection.

This study proposes FastCrackNet, a computationally efficient crack-detection approach. Instead of a computationally costly convolutional neural network (CNN), this technique uses an effective, fully connected network, which is coupled with a 2D-wavelet image transform for analyzing and a locality sensitive discriminant analysis (LSDA) for reducing the number of features. The algorithm described here is used to detect tiny concrete cracks in two noisy adverse conditions and image shadows. By combining wavelet-based feature extraction, feature reduction, and a rapid classifier based on deep learning, this technique surpasses other image classifiers in terms of speed, performance, and resilience. In order to evaluate the accuracy and speed of FastCrackNet, two prominent pre-trained CNN architectures, namely GoogleNet and Xception, are employed. Findings reveal that FastCrackNet has better speed and accuracy than the other models. This study establishes performance and computational thresholds for classifying photos in difficult conditions. In terms of classification efficiency, FastCrackNet outperformed GoogleNet and the Xception model by more than 60 and 80 times, respectively. Furthermore, FastCrackNet's dependability was proved by its robustness and stability in the presence of uncertainties produced by network characteristics and input images, such as input image size, batch size, and input image dimensions.

Feature extraction via prototype margin distance maximizing criterion for subspace learning

Locality sensitive discriminant analysis (LSDA) is an effective feature extraction algorithm, which captures the local structure and label information to build margin for classification. However, the margin of LSDA is easily affected by outliers. To overcome the limitations of LSDA, we propose a new margin-based feature extraction algorithm called prototype margin distance maximizing criterion (PMDMC). Specifically, for each class, we first find its either heterogeneous nearest marginal samples to build the between-class marginal patch. While one side does not characterize the margin. Therefore, for between-class marginal patch, we turn back to find its corresponding within-class marginal patch in the original class. With the between-class marginal patch and within-class marginal patch, we build a novel margin. Finally, we use the Fisher-like criterion to find the optimal feature subspace. In this feature subspace, the margins between different classes are enlarged by maximizing the distance between two heterogeneous marginal patches, and simultaneously minimizing the distance between the prototypes (class means) and their corresponding within-class marginal patches. The obtained margins are more discriminative than before. What's more, under this margin design, the outliers which are usually lying in margins can be pulled close into their respective homogeneous domains. From the comparative experiments performed on the ORL, Yale, and AR face databases, we prove the effectiveness of the proposed PMDMC.

Broad Learning System with Locality Sensitive Discriminant Analysis for Hyperspectral Image Classification

In this paper, we propose a new method for hyperspectral images (HSI) classification, aiming to take advantage of both manifold learning-based feature extraction and neural networks by stacking layers applying locality sensitive discriminant analysis (LSDA) to broad learning system (BLS). BLS has been proven to be a successful model for various machine learning tasks due to its high feature representative capacity introduced by numerous randomly mapped features. However, it also produces redundancy, which is indiscriminate and finally lowers its performance and causes heavy computing demand, especially in cases of the input data bearing high dimensionality. In our work, a manifold learning method is integrated into the BLS by inserting two LSDA layers before the input layer and output layer separate, so the spectral-spatial HSI features are fully utilized to acquire the state-of-the-art classification accuracy. The extensive experiments have shown our method’s superiority.

A Reliable Auto-Robust Analysis of Blood Smear Images for Classification of Microcytic Hypochromic Anemia Using Gray Level Matrices and Gabor Feature Bank

Accurate blood smear quantification with various blood cell samples is of great clinical importance. The conventional manual process of blood smear quantification is quite time consuming and is prone to errors. Therefore, this paper presents automatic detection of the most frequently occurring condition in human blood—microcytic hyperchromic anemia—which is the cause of various life-threatening diseases. This task has been done with segmentation of blood contents, i.e., Red Blood Cells (RBCs), White Blood Cells (WBCs), and platelets, in the first step. Then, the most influential features like geometric shape descriptors, Gray Level Co-occurrence Matrix (GLCM), Gray Level Run Length Matrix (GLRLM), and Gabor features (mean squared energy and mean amplitude) are extracted from each of the RBCs. To discriminate the cells as hypochromic microcytes among other RBC classes, scanning is done at angles (0, 45, 90, and 135). To achieve high-level accuracy, Adaptive Synthetic (AdaSyn) sampling for imbalance learning is used to balance the datasets and locality sensitive discriminant analysis (LSDA) technique is used for feature reduction. Finally, upon using these features, classification of blood cells is done using the multilayer perceptual model and random forest learning algorithms. Performance in terms of accuracy was 96%, which is better than the performance of existing techniques. The final outcome of this work may be useful in the efforts to produce a cost-effective screening scheme that could make inexpensive screening for blood smear analysis available globally, thus providing early detection of these diseases.

Development of breast papillary index for differentiation of benign and malignant lesions using ultrasound images

Papillary breast lesions include a wide spectrum of pathologies ranging from benign to malignant. The word papillary originates from finger-like projections, or papules, which are seen when these lesions are projected under a microscope. Papillary breast lesions have an array of radiological features at presentations; hence differentiation between benign and malignant based on imaging features is challenging. Histopathological diagnosis is crucial for the distinction and further management of the lesions. Traditionally, tumor and ductal excision is the treatment of choice for malignant and atypical or benign papilloma with imaging discordance. However, current clinical practice guidance advocates complete surgical excision, even for asymptomatic and purely benign papillomas diagnosed on core needle biopsy, as they are highly associated with atypia and malignant upstage on subsequent surgery. Computer aided diagnosis (CAD) is a non-invasive method of diagnosing medical signals/images using advanced image processing followed by soft computing techniques. In this study, we have developed a non-invasive CAD system for differentiating benign versus malignant papillary breast lesions using bi-dimensional empirical mode decomposition (BEMD) and the discrete cosine transform (DCT) followed by locality sensitive discriminant analysis (LSDA). The developed model is validated using a large collection of ultrasound images of papillary breast lesions, and achieved a maximum performance of 98.63% accuracy. We have also developed a breast papillary index, which may in the future be used as a substitute for the conventional soft computing techniques. The developed model can be utilized as a tool to assist radiologists in their routine clinical practice after validation with a larger database.

A novel hybrid approach for automated detection of retinal detachment using ultrasound images

Retinal detachment (RD) is an ocular emergency, which needs quick intervention to preclude permanent vision loss. In general, ocular ultrasound is used by ophthalmologists to enhance their judgment in detecting RD in eyes with media opacities which precludes the retinal evaluation. However, the quality of ultrasound (US) images may be degraded due to the presence of noise, and other retinal conditions may cause membranous echoes. All these can influence the accuracy of diagnosis. Hence, to overcome the above, we are proposing an automated system to detect RD using texton, higher order spectral (HOS) cumulants and locality sensitive discriminant analysis (LSDA) techniques. Our developed method is able to classify the posterior vitreous detachment and RD using support vector machine classifier with highest accuracy of 99.13%. Our system is ready to be tested with more diverse ultrasound images and aid ophthalmologists to arrive at a more accurate diagnosis.

Automated focal EEG signal detection based on third order cumulant function

Epilepsy is a chronic neurological disorder which occurs due to recurrent seizures. The epilepsy surgery is the only cure of epileptic seizures as it cannot be controlled with medication. Hence, it becomes the primary task to localize the epileptogenic zone for successful epilepsy surgery. The epileptic surgical area can be recognized by the focal intracranial electroencephalogram (EEG) signals. In this paper, a nonlinear third-order cumulant has been proposed for the classification of the non-focal and focal intracranial EEG signals efficiently. The attributes are measured from the logarithm of the diagonal slice of third-order cumulant. It provides relevant subtle information about the nonlinear dynamics of EEG signals. A data reduction technique, locality sensitive discriminant analysis (LSDA), has been introduced to map the measured features at higher dimensional space and ranked them according to the probability of discrimination. The achieved results reveal that the ranked LSDA features with the support vector machine (SVM) classifier have yielded maximum classification accuracy of 99% on the Bern Barcelona EEG database. Thus, the proposed algorithm helps a clinician to localize the epileptogenic zone for successful brain surgery.

Elucidation of ball bearing performance utilizing product functions of vibration signals and locality sensitive discriminant analysis

Dimensionality is a decent term utilized for the dimension-related issues of feature vectors for the condition assessment of bearings. It has been a challenging task while dealing with the characteristic information conservation of the sampled data. Feature vectors with higher dimensions provide an accurate description of the condition, while lower dimension vectors are easy to be classified. These conditions make the dimensionality of feature space a big challenge in the performance evaluation of critical machine parts like bearings. This paper presents a judicious and wise application of a linear approach, locality sensitive discriminant analysis along with the rational use of local mean decomposition in performance degradation assessment. The combination effectively solves the problems of insufficient training samples and large dimensionality of fault features, which imparts excessive noise and causes loss of competent hidden characteristics. In such cases, local structures of feature space are more crucial than the global one. These obstacles can be answered by imparting locality sensitive discriminant analysis as an initial operation. Locality sensitive discriminant analysis is a linear dimensionality reduction tool, which explores the precise projections that amplify the margin between data points and prepare a conservation aid to preserve faulty information of bearings. The steps are followed to achieve the same: decomposition of vibration signal into product functions; calculation of fault features; third, higher dimensionality of the features is reduced with the implementation of locality sensitive discriminant analysis and some prime features are selected with the proposed criterion; the reduced features are further clustered and a trained model of assessment is prepared. Finally, health indicator is calculated from the trained model and test features. The proposed technique is verified on two bearing datasets. The superiority of the technique has been envisaged by comparing the method with different similar assessment methods i.e. time domain features, linear discriminant analysis, and principal component analysis.

Locality sensitive discriminant projection for feature extraction and face recognition

As an effective feature extraction method, locality sensitive discriminant analysis (LSDA) utilizes the neighbor relationship of data to characterize the manifold structure of data and uses label information of data to adapt to classification tasks. However, the performance of LSDA is affected by outliers and the destruction of local structure. Aiming at solving the limitations of LSDA, a locality sensitive discriminant projection (LSDP) algorithm is proposed. LSDP minimizes the distance of intraclass neighbor samples to maintain local structure and minimizes the intraclass non-neighbor samples to increase the compactness of intraclass samples after projection. The problem of outliers is alleviated by increasing the compactness of intraclass samples in subspace. At the same time, we redefine the weights of interclass neighbor samples to maintain the neighbor relationship of different labels samples. Holding the local structure of interclass samples maintains the manifold structure of data. Experiments on face datasets demonstrate the effectiveness of the LSDP algorithm.

Adaptive learning in the detection of Movement Related Cortical Potentials improves usability of associative Brain-Computer Interfaces.

Brain-computer interfaces have increasingly found applications in motor function recovery in stroke patients. In this context, it has been demonstrated that associative-BCI protocols, implemented by means the movement related cortical potentials (MRCPs), induce significant cortical plasticity. To date, no methods have been proposed to deal with brain signal (i.e. MRCP feature) non-stationarity. This study introduces adaptive learning methods in MRCP detection and aims at comparing a no-adaptive approach based on the Locality Sensitive Discriminant Analysis (LSDA) with three LSDA-based adaptive approaches. As a proof of concept, EEG and force data were collected from six healthy subjects while performing isometric ankle dorsiflexion. Results revealed that adaptive algorithms increase the number of true detections and decrease the number of false positives per minute. Moreover, the markedly reduction of BCI system calibration time suggests that these methods have the potential to improve the usability of associative-BCI in post-stroke motor recovery.

A NEW TECHNIQUE FOR CLASSIFICATION OF FOCAL AND NONFOCAL EEG SIGNALS USING HIGHER-ORDER SPECTRA

Epilepsy is a neurological disorder characterized by epileptic seizures inside the human brain. An authentic localization of epileptogenic area will help the clinicians for a successful epilepsy surgery. The epileptogenic area can be characterized by the focal electroencephalogram (EEG) signals. Hence, in this article, a bispectrum-based approach is implemented to characterize the focal EEG signals. The highest twenty-five magnitudes of bispectrum from the principal domain are used as features. The locality sensitive discriminant analysis (LSDA), data reduction technique, is implemented to reduce the number of attributes. The ranked LSDA attributes are input to the support vector machine (SVM) classifier yielding 96.2% classification accuracy using the entire Bern Barcelona EEG database. Hence, the proposed technique can be employed to confirm the epileptogenic area for successful epilepsy surgery and can be employed in the community health care centers and hospitals.

AUTOMATED GLAUCOMA DETECTION USING CENTER SLICE OF HIGHER ORDER STATISTICS

Glaucoma is one of the leading causes of blindness. The raised intraocular pressure is one of the important modifiable risk factor causing glaucomatous optic nerve damage. Glaucomatous optic nerve damage is seen as increase in the cupping of the optic disc and loss of neuroretinal rim. An automated detection system using nonlinear higher order statistics (HOS) based method is used to capture the detailed information present in the fundus image efficiently. The center slice of bispectrum and bicepstrum are applied on fundus images. Various features are extracted from the diagonal of these central slices. In order to reduce the number of features the locality sensitive discriminant analysis (LSDA) data reduction technique method is implemented. The ranked LSDA features are fed to support vector machine (SVM) classifier with various kernels for automated glaucoma detection. The simulation is performed on two databases. The proposed algorithm has yielded classification accuracy of 98.8% and 95% using entire private and public databases, respectively. The proposed technique achieved the highest classification accuracy, hence, confirm the diagnosis of ophthalmologists and can be employed in the community health care centers and hospitals.

Dictionary‐induced least squares framework for multi‐view dimensionality reduction with multi‐manifold embeddings

This study proposes a novel dimensionality reduction (DR) method for multi‐view datasets. The principal component analysis (PCA) idea of minimising least squares reconstruction errors is extended to consider both data distribution and penalty weights called dictionary to recover outliers free global structures from missing and noisy data points. In this way, PCA is viewed as a special instance of the authors’ proposed dictionary induced least squares framework (DLS). Furthermore, to appropriately handle multi‐view DR, we combine the DLS with multiple manifold embeddings (DLSME). Therefore it can obtain lower projections while maintaining a balance between preserving global structures with DLS and local structures with multi‐manifold embeddings. Extensive experiments on object and face recognition datasets verify that the DLS achieves better classification results with lower dimensional projections than PCA. Also, on many multi‐view datasets of visual recognition and web image annotation, the DLSME method demonstrates more effectiveness than Graph‐Laplacian PCA (gLPCA), robust PCA‐optimal mean, canonical correlation analysis (CCA), bilinear models (BLM), neighbourhood preserving embedding, locality preserving projections, and locality sensitive discriminant analysis.

RETRACTED ARTICLE: Automated detection of chronic kidney disease using higher-order features and elongated quinary patterns from B-mode ultrasound images

Chronic kidney disease (CKD) is a continuing loss of kidney function, and early detection of this disease is fundamental to halting its progression to end-stage disease. Numerous methods have been proposed to detect CKD, mainly focusing on classification based upon peripheral clinical parameters and quantitative ultrasound parameters that must be manually calculated, or on shear wave elastography. No studies have been found that detect the presence or absence of CKD based solely from one B-mode ultrasound image. In this work, we propose an automated system to detect chronic kidney disease utilizing only the automatic extraction of features from a B-mode ultrasound image of the kidney, with a database of 405 images. Higher-order bispectrum and cumulants, and elongated quinary patterns, are extracted from each image to provide a final total of 24,480 features per image. These features were subjected to a locality sensitive discriminant analysis (LSDA) technique, which provides 30 LSDA coefficients. The coefficients were arranged according to their t value and inserted into various classifiers, to yield the best diagnostic accuracy using the least number of features. The best performance was obtained using a support vector machine and a radial basis function, utilizing only five features, resulting in an accuracy of 99.75%, a sensitivity of 100%, and a specificity of 99.57%. Based upon these findings, it is evident that the technique accurately and automatically identifies subjects with and without CKD from B-mode ultrasound images.

Automated detection of diabetic foot with and without neuropathy using double density-dual tree-complex wavelet transform on foot thermograms

Diabetic foot is the most common problem among diabetic patients, mainly due to peripheral vascular and neuropathy induced capillary perfusion changes. These pathogenic factors cause superficial temperature changes that can be qualitatively and visually documented using infrared thermography (IRT). Hence, IRT can potentially be used to evaluate the diabetic foot. However, it is tedious to manually interpret these subtle temperature variations by inspecting the feet thermal image. Therefore, an automated system to detect diabetic foot with and without neuropathy is proposed. In this study, 51 healthy individuals and 66 diabetic patients (33 with and 33 without neuropathy) are considered. The segmented plantar foot thermograms are decomposed into coefficients using double density-dual tree-complex wavelet transform (DD-DT-CWT). Several entropy and texture features are extracted from the decomposed images of left, right and bilateral foot. These features are reduced using various dimensionality reduction techniques and subsequently ranked using F-values. The ranked features are fed individually into the different classifiers one by one. The developed system yielded 93.16% accuracy, 90.91% sensitivity and 98.04% specificity using only four locality sensitive discriminant analysis (LSDA) features obtained from bilateral foot thermal images with k-nearest neighbour (kNN) classifier. This automated diabetic foot detection system can be introduced in polyclinics and hospitals to clinically support the clinicians to confirm their manual diabetic foot diagnosis.

Novel expert system for glaucoma identification using non-parametric spatial envelope energy spectrum with fundus images

Glaucoma is the prime cause of blindness and early detection of it may prevent patients from vision loss. An expert system plays a vital role in glaucoma screening, which assist the ophthalmologists to make accurate decision. This paper proposes a novel technique for glaucoma detection using optic disk localization and non-parametric GIST descriptor. The method proposes a novel area based optic disk segmentation followed by the Radon transformation (RT). The change in the illumination levels of Radon transformed image are compensated using modified census transformation (MCT). The MCT images are then subjected to GIST descriptor to extract the spatial envelope energy spectrum. The obtained dimension of the GIST descriptor is reduced using locality sensitive discriminant analysis (LSDA) followed by various feature selection and ranking schemes. The ranked features are used to build an efficient classifier to detect glaucoma. Our system yielded a maximum accuracy (97.00%), sensitivity (97.80%) and specificity (95.80%) using support vector machine (SVM) classifier with nineteen features. Developed expert system also achieved maximum accuracy (93.62%), sensitivity (87.50%) and specificity (98.43%) for public dataset using twenty six features. The proposed method is efficient and computationally less expensive as it require only nineteen features to model a classifier for the huge dataset. Therefore the proposed method can be effectively utilized in hospitals for glaucoma screening.

A Lip Reading System Based on OLSDA

Lip features extraction is a very crucial part in lip reading system, accurate and effective features can promote the recognition rate of system. In this paper, we propose a features extraction method which is based on geometrical model and Optimized Locality Sensitive Discriminant Analysis (OLSDA). First, the key points in the lip area are extracted artificially, lip visual features are obtained by mathematical treatment. Then, the dimension of these features are reduced by OLSDA and processed features maintain the ability of recognize. We verify the method by SVM classifier and experimental results show that this system has a high recognition rate.

Locality Sensitive Discriminant Analysis for Group Sparse Representation-Based Hyperspectral Imagery Classification

This letter proposes to integrate the locality sensitive discriminant analysis (LSDA) with the group sparse representation (GSR) for a hyperspectral imagery classification. The LSDA is to project the data set to a lower-dimensional subspace to preserve local manifold structure and discriminant information, while the GSR is to encode the projected testing set as a sparse linear combination of group-structured training samples for classification. The proposed approach, denoted as LSDA-GSR classifier (GSRC), is evaluated using two real hyperspectral data sets. Experimental results demonstrate that it can provide considerable improvement to the original counterparts, i.e., SRC and GSRC, with a relatively low computational cost.

Automated diabetic macular edema (DME) grading system using DWT, DCT Features and maculopathy index

The cause of diabetic macular edema (DME) is due to prolonged and uncontrolled diabetes mellitus (DM) which affects the vision of diabetic subjects. DME is graded based on the exudate location from the macula. It is clinically diagnosed using fundus images which is tedious and time-consuming. Regular eye screening and subsequent treatment may prevent the vision loss. Hence, in this work, a hybrid system based on Radon transform (RT), discrete wavelet transform (DWT) and discrete cosine transform (DCT) are proposed for an automated detection of DME. The fundus images are subjected to RT to obtain sinograms and DWT is applied on these sinograms to extract wavelet coefficients (approximate, horizontal, vertical and diagonal). DCT is applied on approximate coefficients to obtain 2D-DCT coefficients. Further, these coefficients are converted into 1D vector by arranging the coefficients in zig-zag manner. This 1D signal is subjected to locality sensitive discriminant analysis (LSDA). Finally, various supervised classifiers are used to classify the three classes using significant features. Our proposed technique yielded a classification accuracy of 100% and 97.01% using two and seven significant features for private and public (MESSIDOR) databases respectively. Also, a maculopathy index is formulated with two significant parameters to discriminate the three groups distinctly using a single integer. Hence, our obtained results suggest that this system can be used as an eye screening tool for diabetic subjects for DME.