Wavelet Transform Feature Extraction Research Articles

Feature extraction method based on wavelet transform for the pressure tactile reflection spectrum of long chirped fiber grating

In this study, a feature extraction method for long chirped fiber grating (LCFBG) pressure tactile sensing spectral signal is proposed and verified. Firstly, the pressure sensing theory and spectral characteristics of LCFBG are analyzed, including spectral pattern, peak, FWHM and other characteristic parameters. The spectral characteristics of LCFBG were analyzed by VI transfer matrix method. Based on the simulation data, different wavelet basis functions are compared and analyzed for the effect of spectral information feature extraction. Bior4.4 wavelet basis function was used to decompose the denoised spectrum at four levels, and the error of spectral zero and extreme point was the minimum, up to 0.4 pm. To verify the extraction effect of LCFBG pressure tactile sensing spectral features, the point contact, surface contact and deformation sensing systems of LCFBG are established, and the experimental results are analyzed by using wavelet transform feature extraction algorithm. The results show that the sensitivity coefficients of the sensing system for point contact force loading and surface contact force loading are 1.04 dBm/N and 1.87 dBm·mm/N respectively. The linearity of the reflection spectrum intensity and the pressure contact force is above 99.5 %. In the deformation sensing experiment, the reflection spectrum has a good linear relationship with the strain change, and the central wavelength interval of the peak and valley determines the width of the deformation region. The relevant research results can provide reference and direction for the feature extraction and recognition of LCFBG tactile sensing spectral signals.

Diagnosis of Pulpitis from Dental Panoramic Radiograph Using Histogram of Gradients with Discrete Wavelet Transform and Multilevel Neural Network Techniques

Radiograph plays the major role of diagnosis, treatment and surgery in the Dental field. There are many types of Intra and extra oral radiographs in which Dental Panoramic Radiograph helps in visualising the full view of the oral cavity. Pulpitis is the dental diseases caused due to the inflammation of the dental pulp from untreated caries, trauma or multiple restorations which leads to Apical Periodontitis. To predict the severity of pulp vitality pulp inflammation has to be evaluated. Radiographs helps the dentist in diagnosing the extent of tooth decay and inflammation. An automatic diagnostic model is proposed using robust algorithms to diagnose pulpits. Dental Panoramic Radiograph is used in the proposed research to diagnose the pulpitis and to classify the normal teeth from the pulpitis. The collected images are pre-processed using Histogram Equalization and filtered using Gaussian and Median filters. Modified K-Means algorithm is applied to segment the bony and teeth area from the oral cavity area. Integral Histogram of Gradients with Discrete Wavelet Transform feature extraction techniques and Multi-Layer Neural Network Classifier is proposed to achieve the accuracy of 91.09% which can be used as an assistive tool by the dentist to diagnose pulpitis.

Automated detection and screening of depression using continuous wavelet transform with electroencephalogram signals

AbstractNearly 264 million people around the globe currently suffer from clinical depression, according to the World Health Organization. Although there are diagnostic techniques and treatments presently used by professionals, they are not always helpful. Herein, we suggest the use of advanced technological methods to diagnose depressed patients correctly. A machine learning approach is presented, which uses the electroencephalogram for diagnostics. The model extracts multiple features by applying a continuous wavelet transform (CWT) for each recording. These recordings are employed to train and test the model, with data gathered from 15 depressed and 15 normal patients. After the features are extracted from these recordings, it is organized into matrix form. The features are dimensionally reduced using kernel‐principal component analysis and principal component analysis techniques, ranked using Student's t‐test, and then labelled as normal or depressed with various classifiers. Accuracies of 99.33% and 99.13% were achieved for the right and left hemispheres of the brain, respectively, and 99.26% for the combined hemispheres of the brain. As compared to the discrete and empirical wavelet transform feature extraction methods, the CWT attained the best results. A depression severity index was also developed, using two features for discriminating the classes: normal versus depressed.

Computer Image Processing System Based on Feature Extraction Technology

This article studies the methods of computer image acquisition, processing and recognition. The study found that the images collected from the computer image processing system often have blurring phenomenon. For this reason, the paper proposes a wavelet transform feature extraction algorithm based on colour images, which can separate the image and the background without greying the colour image. Through experimental research, it is found that this method can efficiently and quickly solve the problem of automatic image registration.

RETRACTED ARTICLE: Big Data Mining and Analysis Based on Convolutional Fuzzy Neural Network

The purpose of this paper is to solve the shortcomings of traditional big data mining and analysis (BDMA); this paper compares and analyzes the denoise effect of wavelet transform and wavelet packet analysis and chooses to use wavelet packet analysis method to denoise the signal and extract fault feature vector. Finally, the extracted feature vectors are divided into two libraries, including training and testing. The fuzzy convolutional neural network (FCNN) is trained by using the training sample database data, the network weights are constantly modified, and the projection of nonlinear is computed between the input features and output features. After the expected recognition accuracy is achieved, the performance of FCNN is evaluated by using the detection samples. The performance comparison and analysis are conducted with the traditional BDMA algorithm, mainly including the comparison of recognition accuracy and learning convergence speed. Two stages of feature extraction and data mining (DM) are constructed on the big data set. The experimental results prove the effectiveness of wavelet transform feature and FCNN in analyzing and mining in big data. Finally, through the example of BDMA, this paper verifies that the organic combination of wavelet transform feature extraction and FCNN algorithm obviously improves the efficiency, meets the requirements of big data analysis, and provides potential application value for BDMA.

Steel strip surface inspection through the combination of feature selection and multiclass classifiers

Purpose During the production process of steel strip, some defects may appear on the surface, that is, traditional manual inspection could not meet the requirements of low-cost and high-efficiency production. The purpose of this paper is to propose a method of feature selection based on filter methods combined with hidden Bayesian classifier for improving the efficiency of defect recognition and reduce the complexity of calculation. The method can select the optimal hybrid model for realizing the accurate classification of steel strip surface defects. Design/methodology/approach A large image feature set was initially obtained based on the discrete wavelet transform feature extraction method. Three feature selection methods (including correlation-based feature selection, consistency subset evaluator [CSE] and information gain) were then used to optimize the feature space. Parameters for the feature selection methods were based on the classification accuracy results of hidden Naive Bayes (HNB) algorithm. The selected feature subset was then applied to the traditional NB classifier and leading extended NB classifiers. Findings The experimental results demonstrated that the HNB model combined with feature selection approaches has better classification performance than other models of defect recognition. Among the results of this study, the proposed hybrid model of CSE + HNB is the most robust and effective and of highest classification accuracy in identifying the optimal subset of the surface defect database. Originality/value The main contribution of this paper is the development of a hybrid model combining feature selection and multi-class classification algorithms for steel strip surface inspection. The proposed hybrid model is primarily robust and effective for steel strip surface inspection.

A High-Accuracy of Transmission Line Faults (TLFs) Classification Based on Convolutional Neural Network

To improve power system reliability, a protection mechanism is highly needed. Early detection can be used to prevent failures in the power transmission line (TL). A classification system method is widely used to protect against false detection as well as assist the decision analysis. Each TL signal has a continuous pattern in which it can be detected and classified by the conventional methods, i.e., wavelet feature extraction and artificial neural network (ANN). However, the accuracy resulting from these mentioned models is relatively low. To overcome this issue, we propose a machine learning-based on Convolutional Neural Network (CNN) for the transmission line faults (TLFs) application. CNN is more suitable for pattern recognition compared to conventional ANN and ANN with Discrete Wavelet Transform (DWT) feature extraction. In this work, we first simulate our proposed model by using Simulink® and Matlab®. This simulation generates a fault signal dataset, which is divided into 45.738 data training and 4.752 data tests. Later, we design the number of machine learning classifiers. Each model classifier is trained by exposing it to the same dataset. The CNN design, with raw input, is determined as an optimal output model from the training process with 100% accuracy.

Automatic annotation of satellite images with multi class support vector machine

Automatic Image Annotation (AIA) is used in image retrieval systems to retrieve the images by predicting tags for images. To achieve image retrieval with high accuracy, an automatic image annotation approach by using Multiclass SVM with the hybrid kernel is proposed. The hybrid kernel is a combination of Radial Basis Function (RBF) and Polynomial Kernel which overcomes the drawbacks of single kernels such as less accuracy, high computational complexity, etc. This technique exploits the Linear Binary Pattern- Discrete Wavelet Transform (LBP-DWT) feature extraction technique to extract the features in horizontal, vertical, and diagonal directions. The experiments suggest that the multiclass SVM can attain a higher accuracy than other conventional SVM with any single kernels. The Multiclass SVM can achieve high accuracy as 95.61% and increases the accuracy by 3.26%, 1.79%, and Kappa coefficient by 3.22%, 2.27% when compared with SVM RBF kernel, polynomial kernel respectively.

Stacked Auto-Encoder Based CNC Tool Diagnosis Using Discrete Wavelet Transform Feature Extraction

Machining processes are critical and widely used components in the manufacturing industry because they help to precisely make products and reduce production time. To keep the previous advantages, a machine tool should be installed at the designated place and condition of the machine tool should be maintained appropriately to working environment. In various maintenance methods for keeping the condition of machine tool, condition-based maintenance can be robust to unpredicted accidents and reduce maintenance costs. Tool monitoring and diagnosis are some of the most important components of the condition based maintenance. This paper proposes stacked auto-encoder based CNC machine tool diagnosis using discrete wavelet transform feature extraction to diagnose a machine tool. The diagnosis model, which only uses cutting force data, cannot sufficiently reflects tool condition. Hence, we modeled diagnosis model using features extracted from a cutting force, a current signal, and coefficients of the discrete wavelet transform. The experimental results showed that the model which uses feature data has better performance than the model that uses only cutting force data. The feature based models are lower false negative rate (FNR) and false positive rate. Moreover, squared prediction error using normalized residual vector also reduced FNR because normalization reduces weight bias.

Lung Nodule Detection: Image Enhancement using Fuzzy Rule Based Contrast Limited Adaptive Histogram Equalization and Entropy Weighted Residual Convolution Neural Network Method in CT

Computed Tomography (CT) images are read by several lung nodule detection methods. The early step of contrast enhancement is mandatory because of low contrast in original image and further techniques of image processing are with unsatisfactory results. Hence this process are resulted an enhanced image of clearly discrete lung area from background. Image enhancement, feature extraction, and classification are three primary steps. In this work, Rule based Contrast Limited Adaptive Histogram Equalization (FRCLAHE) perform image enhancement step followed by feature extraction and Fuzzy Rule (FR) determines the contrast value. From rules upper contrast value are determined then image is enhancement from CLAHE. In the second, the feature extraction is conducted using the Fuzzy Continuous Wavelet Transform (FCWT) and Gray Level Feature Extraction (GLCM). After this step, the classification is completed using the Entropy Weighted Residual Convolution Neural Network (EWRCNN). Finally, the results are evaluated between the samples, compared to FP reduction with Faster R-CNN alone, the inclusion of rule‐based classification lead to an improvement in detection accuracy for the CAD system. These preliminary results demonstrate the feasibility of the proposed EWRCNN approach to lung nodule detection and FP reduction on CT images.

Non-destructive analysis for the in-flexible-containers liquid composition based on WTFE-NPLS method

In order to achieve the non-destructive analysis of complex mixed solution in flexible container, the effect of flexible containers on spectral analysis was studied in this work. To reduce the influence of the difference in flexible containers on the spectral analysis of complex mixed solution, the difference of absorption spectra between the complex solution and the flexible container was studied. A method that using N-way Partial Least Square (N-PLS) established a quantitative model between the concentration of the measured substance and the spectral feature vectors extracted by wavelet transform (WTFE) is proposed to suppress the influence of the difference in the flexible container. To verify the feasibility of the method, an experiment was designed by using a mixed solution of intra-lipid and India-ink as an analytical sample and by using the container of polyvinyl chloride (PVC) as the flexible container of the analytical sample. The results of modeling by using PLS, WTFE-PLS (Partial Least Square regression combining Wavelet transform feature extraction), and WTFE-NPLS (N-way Partial Least Square regression combining Wavelet transform feature extraction) were compared. The experimental results show that the WTFE-NPLS method effectively suppresses the influence of the difference in flexible containers on the spectral analysis of complex mixed solution. It lays the foundation for the non-destructive spectral analysis of blood or other complex mixed solution in flexible containers.

Two-dimensional empirical wavelet transform based supervised hyperspectral image classification

Hyperspectral image classification is one of the major field of application for hyperspectral imaging systems. Though hyperspectral data gives accurate results than their multispectral counterparts, they are computationally more complex due to their high dimensionality. One of the classical problem while dealing with supervised hyperspectral classification is the class imbalance problem that arises due to the limited availability of samples for training. In order to deal with high dimensionality, many feature mining techniques has been proposed in literature for hyperspectral images. In this paper, we propose a hyperspectral image classification method based on two-dimensional Empirical Wavelet Transform (2D-EWT) feature extraction and compare it with that of Image Empirical Mode Decomposition (IEMD) based extracted features and raw features. Here, the focus is upon the fact that the number of features trained should be less than what is to be tested. Since the computational time for classification is also of prime importance, only some of the fast and best of the classifiers are selected. Sparse-based classifiers are one of the fast and efficient method for supervised classification of hyperspectral images. Subspace Pursuit (SP) and Orthogonal Matching Pursuit (OMP) algorithms are used in our experiments for sparse-based classification. Other classifiers used are Support Vector Machine (SVM) and Hybrid Support Vector Selection and Adaptation (HSVSA). The proposed methodology gives improved performance in terms of classification evaluation measures for hyperspectral image classification task.

Intent recognition of torso motion using wavelet transform feature extraction and linear discriminant analysis ensemble classification

Abstract In this paper, a multi-sensor, multi-classifier approach for intent recognition of human torso motion is presented. A linear discriminant analysis based classifier is used, and the extraction of time-frequency domain features through the use of the wavelet transform is discussed. In addition, a weighted multi-classifier combination method for combining outputs of multiple classifiers into a single coherent output is implemented. The approach was evaluated on physiological data collected from three human participants. Results show up to 97% accuracy in classifying flexion and extension motions of the torso.

Fault Diagnosis of Helical Gearbox through Vibration Signals using J48 Decision Tree and Wavelet

Objectives: Gear plays an efficient role in power transmission. Minor faults in gears can lead to severe faults. The vibration analysis can be used for determining the causes of the faults which are raised while ongoing operation. This study determines the usage of machine learning algorithm for condition monitoring of helical gearbox. Methods/Statistical Analysis: The vibration signals were taken by using accelerometers from helical gearbox in which artificial faults were incorporated before testing. By using Discrete Wavelet Transform (DWT) feature extraction was done. The feature selection and feature classification was done by using J48 algorithm and subsequent results were observed. Findings: The classification accuracy of helical gearbox using Discrete Wavelet Transform was observed to be 89.28% which itself shows its efficiency. In feature extraction maximum accuracy of 89.06% was obtained by sym 8 wavelet. During feature selection and classification many modifications in algorithm were made i.e. minimum number of object, confidence factor etc. Suitable readings of the modifications were applied and feature classification was done. Improvements: Different Discrete Wavelet Transforms were compared taken from vibration signal proved Sym 8 Discrete Wavelet Transform is the best one to be used in this scenario. The methodology yielded a satisfactory classification accuracy of 89.28%, which is higher than what was obtained by similar experiments with different methodology till date. The results and their analysis are discussed in the study. The performance of this methodology may be further improved by using different classifiers and different wavelets.

A New Method Based for Diagnosis of Breast Cancer Cells from Microscopic Images: DWEE—JHT

In these days, there are many various diseases, whose diagnosis is very hardly. Breast cancer is one of these type diseases. In this paper, accuracy diagnosis of normal, benign, and malign breast cancer cell were found by combining mean success rates Jensen Shannon, Hellinger, and Triangle measure which connected with each other. In this article, an diagnostic method based on feature extraction Discrete Wavelet Entropy Energy (DWEE) and Jensen Shannon, Hellinger, Triangle Measure (JHT) Classifier for diagnosis of breast cancer. This diagnosis method is called as DWEE--JHT this paper. With this diagnosis method have found optimal feature subset using discrete wavelet transform feature extraction. Then these convenient features are given to Jensen Shannon, Hellinger, Triangle Measure (JHT) classifier. Then, between classifiers which are Jensen Shannon, Hellinger, and triangle distance have been validated the measures via relationships. Afterwards, breast cancer cells are classified using Jensen Shannon, Hellinger, and Triangle distance. Mean success rate of 16 feature vector with Jensen Shannon classifier is found % 97.81. Mean success rate of 16 feature vector with Hellinger classifier is found % 97.75. Mean success rate of 16 feature vector with Triangle classifier is found % 97.87. By averaging of results obtained from these 3 classifiers are found as 97.81 % average of accuracy diagnosis.

Facial Expression Recognition Based on MATLAB

In today's social life, communication between people is essential to do anything important way. The facial expressions are rich in human behavior information, it is very important means of communication, as a carrier of information, expression can convey a lot of voice can not convey the information. Expression recognition field of pattern recognition is a new task, is an essential part of intelligent machines. This paper studies the discrete wavelet transform feature extraction and expression using MATLAB software image feature extraction and treatment with an elastic template matching algorithm to do the appropriate test expression recognition.

Intelligent drowsy eye detection using image mining

Cell phones, MP4 players, and GPS all distract drivers, forcing them to divide their attention to what is happening outside and what is happening inside the vehicle. Preventing drivers from falling asleep at the wheel and drifting off on the road is an increasingly hard task but a solution can be achieved with today's technology: by capturing driver's eye gaze and thus the level of their exhaustion, the potential for safer roads is real. Today, image mining systems can automatically extract important information from an image dataset which can then be used to classify images and image relationship that been applied in different areas. The basic challenge in image mining research area is to determine how image level pixel representation in an original image can be used and constructed to then identify meaningful information or form relationships. In this research Intelligent Drowsy Eye Detection, using an Image Mining (IDEDIM) system is proposed. The proposed architecture would use different feature extraction techniques and three data mining classification techniques to aid with accurate information collection. Two thousand left and right eye images were used to test the developed system by Discrete Wavelet Transform (DWT), Statistical features, and Local binary pattern (LBP) feature extraction techniques. The extracted features were used as input for Decisions Tree C5.0, K Nearest Neighbor (KNN), and the Support Vector Machine (SVM) Classifier. After several experimental sets, the C5.0 and KNN classifier were performing better than SVM classifier. Based on the results, we recommend the inclusion of LBP and DWT in conjunction with C5.0 with KNN as a classification technique. To validate the achieved results, Receiver Operating Characteristic (ROC) curve is used to compare among the proposed classifiers. The proposed system can be integrated with the existing subsystem into real time Drowsy Detection System to achieve excellent accuracy and performance.

Evaluation of red wines antioxidant capacity by means of a voltammetric e-tongue with an optimized sensor array

In this work, two sets of voltammetric sensors -prepared using different strategies- have been combined in an electronic tongue to evaluate the complete antioxidant profile of red wines. To this aim, wine samples were analyzed with the whole set of sensors. In order to reduce the large dimensionality of the data set while keeping the relevant information provided by the sensors, two different methods of feature selection and data compression were used (the kernels method and Discrete Wavelet Transform feature extraction method). Then, the coefficients obtained were used as the input variables of Principal Component Analysis (to evaluate the capability of discrimination. Partial-least squares regression (PLS) and artificial neural networks (ANNs) were performer to build the quantitative prediction models that allowed the quantification of the antioxidant capacity of the tested wines.

Two-dimensional wavelet transform feature extraction for porous silicon chemical sensors

Designing reliable, fast responding, highly sensitive, and low-power consuming chemo-sensory systems has long been a major goal in chemo-sensing. This goal, however, presents a difficult challenge because having a set of chemo-sensory detectors exhibiting all these aforementioned ideal conditions are still largely un-realizable to-date. This paper presents a unique perspective on capturing more in-depth insights into the physicochemical interactions of two distinct, selectively chemically modified porous silicon (pSi) film-based optical gas sensors by implementing an innovative, based on signal processing methodology, namely the two-dimensional discrete wavelet transform. Specifically, the method consists of using the two-dimensional discrete wavelet transform as a feature extraction method to capture the non-stationary behavior from the bi-dimensional pSi rugate sensor response. Utilizing a comprehensive set of measurements collected from each of the aforementioned optically based chemical sensors, we evaluate the significance of our approach on a complex, six-dimensional chemical analyte discrimination/quantification task problem. Due to the bi-dimensional aspects naturally governing the optical sensor response to chemical analytes, our findings provide evidence that the proposed feature extractor strategy may be a valuable tool to deepen our understanding of the performance of optically based chemical sensors as well as an important step toward attaining their implementation in more realistic chemo-sensing applications.

Wavelet Transform Feature Extraction for Chip form Recognition during Carbon Steel Turning

Cutting force sensor monitoring and wavelet decomposition signal processing were implemented for feature extraction and pattern recognition of chip form typology during turning of 1045 carbon steel. The wavelet packet transform was applied for the analysis of the detected cutting force signals by representing them in a time-frequency domain and providing for the extraction of wavelet packet statistical features. The latter were used to construct wavelet packet feature vectors, ranked according to the number of overlapping elements related to favourable or unfavourable chip forms that cause noise in the pattern recognition procedure (lower number, lower noise, higher rank). The eight highest ranked wavelet packet feature vectors were selected as inputs to a neural network decision-making system on chip form acceptability. Subsequently, a data refinement procedure was employed to improve the neural network performance in the chip form identification process.