Feature Extraction Technology Research Articles

Microwave breast tumor localization using wavelet feature extraction and genetic algorithm-neural network.

Ultra-Wide Band (UWB) microwave breast cancer detection is a promising new technology for routine physical examination and home monitoring. The existing microwave imaging algorithms for breast tumor detection are complex and the effect is still not ideal, due to the heterogeneity of breast tissue, skin reflection, and fibroglandular tissue reflection in backscatter signals. This study aims to develop a machine learning method to accurately locate breast tumor. A microwave-based breast tumor localization method is proposed by time-frequency feature extraction and neural network technology. First, the received microwave array signals are converted into representative and compact features by 4-level Discrete Wavelet Transform (DWT) and Principal Component Analysis (PCA). Then, the Genetic Algorithm-Neural Network (GA-NN) is developed to tune hyper-parameters of the neural network adaptively. The neural network embedded in the GA-NN algorithm is a four-layer architecture and 10-fold cross-validation is performed. Through the trained neural network, the tumor localization performance is evaluated on four datasets that are created by FDTD simulation method from 2-D MRI-derived breast models with varying tissue density, shape, and size. Each dataset consists of 1000 backscatter signals with different tumor positions, in which the ratio of training set to test set is 9:1. In order to verify the generalizability and scalability of the proposed method, the tumor localization performance is also tested on a 3-D breast model. For these 2-D breast models with unknown tumor locations, the evaluation results show that the proposed method has small location errors, which are 0.6076mm, 3.0813mm, 2.0798mm, and 3.2988mm, respectively, and high accuracy, which is 99%, 80%, 94%, and 85%, respectively. Furthermore, the location error and the prediction accuracy of the 3-D breast model are 3.3896mm and 81%. These evaluation results demonstrate that the proposed machine learning method is effective and accurate for microwave breast tumor localization. The traditional microwave-based breast cancer detection method is to reconstruct the entire breast image to highlight the tumor. Compared with the traditional method, our proposed method can directly get the breast tumor location by applying neural network to the received microwave array signals, and circumvent any complicated image reconstruction processing.

Study on feature extraction technology of real-time video acquisition based on deep CNN

Study on feature extraction technology of real-time video acquisition based on deep CNN

A technology of invariant feature extraction of Uav remote sensing image based on fuzzy fractional order function

A technology of invariant feature extraction of Uav remote sensing image based on fuzzy fractional order function

A Three-Phase Current Tacholess Envelope Order Analysis Method for Feature Extraction of Planetary Gearbox under Variable Speed Conditions.

Planetary gearboxes are the key components of large equipment, such as wind turbines, shield machines, etc. The operating state of the planetary gearbox is related to the safety of the equipment as a whole, and its feature extraction technology is essential. In assessing the problem of the non-stationarity of the current signal under variable speed conditions and the difficulty of evaluating the operating state of the planetary gearbox under a tacholess condition, a three-phase current, variable-speed tacholess envelope order analysis method is proposed. Firstly, a tacholess rotation speed estimation is completed by extracting the trend term of the instantaneous frequency of the asynchronous motor’s three-phase currents. The motor slip rate is assumed to be constant. Then, the envelope order analysis signal is obtained by re-sampling in the angular domain. Finally, the features of the envelope order signal are extracted, and a linear discriminant analysis (LDA) algorithm is used to fuse multiple indexes to generate a comprehensive feature reflecting the operating status of the planetary gearbox. The results of the simulation analysis and experimental verification show that the proposed method is effective in evaluating the operating state of the planetary gearbox under variable speed conditions. Compared with the traditional time–frequency ridge extraction method, the tacholess speed estimation method can improve the instantaneous speed estimation accuracy. The comprehensive index of envelope order completes the planetary gearbox state identification process, and a 95% classification accuracy rate is achieved.

Artificial Intelligence Auxiliary Algorithm for Wushu Routine Competition Decision Based on Feature Fusion.

The developments of modern science and technology have significantly promoted the progress of sports science. Advanced technological methods have been widely used in sports training, which has not only improved the scientific level of training but also promoted the continuous growth of sports technology and competition results. Competitive Wushu routine is an important part of Chinese Wushu. The development trend of competitive Wushu routine affects the development of the whole Wushu movement. To improve the training effect of the Wushu routine using artificial intelligence, this paper employed fuzzy information processing and feature extraction technology to analyze the visual features in the process of Wushu competition. The deep neural network-based region segmentation method was employed for implicit feature extraction to examine the shape, texture, and other image features of Wushu routines and improve the recognition performance. The proposed feature extraction model achieved the highest average accuracy of 93.98% accuracy as compared to other contemporary algorithms. Finally, the model was evaluated to validate the superior performance of the proposed method in improving the decision-making ability and effective instruction ability of the martial arts routine competition.

Shape Feature Extraction Techniques for Fruits: A Review

Fruits sorting, recognizing, and classifying are essential post-harvest operations, as they contribute to the quality of food industry, thereby increasing the exported quantity of food. Today, an automated system for fruit classification and recognition is very important, especially when exporting to markets where quality of fruit must be high. In this study, the advantages and disadvantages of the various shape-based feature extraction algorithms and technologies that are used in sorting, classifying, and grading of fruits, as well as fruits quality estimation, are discussed in order to provide a good understanding of the use of shape-based feature extraction techniques.

Interaction designers’ perceptions of using motion-based full-body features

Movement-based full-body interactions are increasingly being used in the design of interactive spaces, computer-mediated environments, and virtual user experiences due to the development and availability of diverse sensing technologies. In this context, the role of interaction designers is to find systematic and predictable relationships between bodily actions and the corresponding responses from technology. Sensor-based interaction design relies on sensor data analysis and higher-level feature extraction to improve detection capabilities. However, understanding human movement to inform the design of motion-based interactions is not straightforward if the detection capabilities of interaction technologies are unknown. We aim at understanding the problems and opportunities that practitioners—regardless of their technical background—perceive in using different motion-based full-body features. To achieve this, we conducted four separate focus groups with experienced practitioners, with and without technical backgrounds. We used a framework for the analysis of focus group data in information systems research to identify content areas and draw conclusions. Our findings suggest that most interaction designers, regardless of their technical background, consider motion-based feature extraction to be challenging and time-consuming. However, participants acknowledge they might use designer-interpretable features as a potential tool to foster user behavior exploration. Understanding how practitioners link sensor-based interaction design with feature extraction technology is relevant to design computational tools and reduce the technical effort required from designers to characterize the user’s movement.

Construction of Electronic Educational Game Design Pattern Based on Feature Extraction Technology

With the popularity of computer and Internet, online games and educational games have become very common forms of entertainment in our daily life. Especially for teenagers, all kinds of games have become an indispensable part of their extracurricular activities. The emergence of Educational Games brings entertainment and interest to students’ boring learning, which can realize the change of students from tired of learning to willing to learn. Its purpose is to stimulate learners’ interest in learning and promote learners’ autonomous learning. Whether the design of educational games matches the characteristics of learners is the key factor to determine the success of educational game design. The design of educational games must be based on the characteristics of learners. Based on the feature extraction technology, this paper analyzes the construction method of the design pattern of electronic educational games, strives to balance the educational and game nature of electronic educational games in the design program, and provides a new idea for the construction of teaching resource database.

Design of real-time data analysis system for physical training based on data mining technology

In competitive sports, the most direct and effective way to improve athletes’ competitive level is to strengthen training. At present, the training mode is still mainly based on coaches’ experience and athletes’ physical conditions, and lacks scientific and effective data information as the basis for planning. In this paper, the software is developed in the core construction of recessed Linux, the VIX bus technology is take advantage of gather sports data, and the data mining technology based on collective information characteristic extraction is designed. This paper uses big data mining method to mine the information of sports evaluation, pick up the autocorrelation characteristics of data information flow of sports training real-time data analysis system, constructs the association rule feature set combined with the prior knowledge of sports evaluation, and realizes the database construction of sports training real-time data analysis system by using the association rule feature extraction and big data information fusion processing technology. The simulation results show that the accuracy of sports assess decision-making information mining using the real-time data analysis system of sports training is high, and the dependability of the system is satisfied.

Research on Feature Extraction Technology of Japanese Corpus Resources Based on Rule Matching

Existing methods of relation extraction can be divided into pattern matching-based methods, dictionary-driven methods and machine learning-based methods, among which machine learning-based methods are the mainstream methods of relation extraction at present. Based on the analysis of Snort’s new features and existing rule matching methods, this paper effectively combines Snort algorithm with Japanese language features, and corrects keyword deviation caused by word frequency dependence by introducing multiple eigenvalues such as position factor and part-of-speech factor. Experimental results show that the introduction of part-of-speech factor and position factor is simple and efficient, and can effectively improve the extraction effect of Japanese keywords, which is suitable for short article keywords.

A survey of ship target feature extraction based on video surveillance

Ship target feature extraction plays an important role in ship recognition and tracking based on video surveillance. However, the current feature extraction of ship targets based on video surveillance does not have a complete summary, so this article reviews these methods. This article summarizes the main ship target feature extraction methods used in ship target recognition and tracking in different application environments, including traditional artificial ship feature extraction and automated learning ship feature extraction. It focuses on the core ideas, advantages and disadvantages of the representative methods in each type of problem, and discusses the current problems in the feature extraction of ship targets and the future development of feature extraction technology.

Pronunciation correction of students in music classroom based on computer voice simulation

In order to improve the pronunciation effect of students in music classroom, based on computer speech simulation technology, this paper combines speech recognition technology and speech feature extraction technology to summarize the various acoustic parameters of speech and the perception of the information by the human ear, and establishes the physical model and digital model of the speech signal. Through the analysis of voice signals, this paper selects several acoustic parameters that can reflect individual characteristics, and studies the methods of their extraction and adjustment to construct a pronunciation correction system for music students based on computer voice simulation. Finally, this paper designs experiments to verify the performance of the system constructed in this paper. The research results show that the system constructed in this paper has certain practical effects and enriches the expressive power of machine speech. The process of synthesizing speech is simple, and the practical effect is good, which meets the current music teaching needs.

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.

Multi-Physical Tracking and Sensing Technology of Switchgear Partial Discharge Based on Edge Computing and Smart Gateway

In view of the high bottleneck of current partial discharge detection technology, strong professionalism and heavy detection tasks, the partial discharge process is greatly affected by the equipment operating environment and condition factors. The paper proposes a switching device partial discharge multiphysics based on edge computing. Research on perception and state tracking technology. The article aims at the edge computing technology of multi-physical partial discharge signal feature extraction, the construction of multi-physical tracking over-limit early warning model, and the development and trial production of intelligent gateway technology. The research found that the edge computing algorithm proposed in the paper can effectively reduce the cost and power consumption under the premise of ensuring the performance of the detection system.

3D Face Modeling Algorithm for Film and Television Animation Based on Lightweight Convolutional Neural Network

Through the analysis of facial feature extraction technology, this paper designs a lightweight convolutional neural network (LW-CNN). The LW-CNN model adopts a separable convolution structure, which can propose more accurate features with fewer parameters and can extract 3D feature points of a human face. In order to enhance the accuracy of feature extraction, a face detection method based on the inverted triangle structure is used to detect the face frame of the images in the training set before the model extracts the features. Aiming at the problem that the feature extraction algorithm based on the difference criterion cannot effectively extract the discriminative information, the Generalized Multiple Maximum Dispersion Difference Criterion (GMMSD) and the corresponding feature extraction algorithm are proposed. The algorithm uses the difference criterion instead of the entropy criterion to avoid the “small sample” problem, and the use of QR decomposition can extract more effective discriminative features for facial recognition, while also reducing the computational complexity of feature extraction. Compared with traditional feature extraction methods, GMMSD avoids the problem of “small samples” and does not require preprocessing steps on the samples; it uses QR decomposition to extract features from the original samples and retains the distribution characteristics of the original samples. According to different change matrices, GMMSD can evolve into different feature extraction algorithms, which shows the generalized characteristics of GMMSD. Experiments show that GMMSD can effectively extract facial identification features and improve the accuracy of facial recognition.

Monopulse Feature Extraction and Fault Diagnosis Method of Rolling Bearing under Low-Speed and Heavy-Load Conditions

According to the rolling bearing local fault vibration mechanism, a monopulse feature extraction and fault diagnosis method of rolling bearing under low-speed and heavy-load conditions based on phase scan and CNN is proposed. The synchronous collected speed signal is used to calculate bearing phase function and divide fault monopulse periods. The monopulse waveforms of multiple fault periods are scanned and ensemble averaged to suppress noise interference and detail feature loss at the same time of feature extraction. By iteratively calibrating phase function, the feature matrix containing bearing fault information can be obtained. Finally, CNN is used to recognize and classify different bearing states. The experimental and analysis results show that bearing fault diagnosis can be achieved. The total recognition rates of constant and variable speed samples are 99.67% and 99.89%, respectively. The trained network has fast convergence speed and good generalization ability for different fault sizes and working conditions. Further experiments show that the method can also accurately identify different bearing degradation states. The total recognition rates of constant and variable speed samples are 96.67% and 95.56%, respectively. The limited errors are concentrated between the degradation states with the same type weak fault. The experimental results using Case Western Reserve University bearing data show that feature extraction and network training are better, and the recognition rates of 5 bearing states are all 100%. Therefore, the proposed method is an effective rolling bearing feature extraction and fault diagnosis technology.

Slope Collapse Detection Based on Image Processing

In this paper, we study a task of slope collapse detection (SCD) for river embankment and formulate it as the tasks of motion detection and image recognition. Specifically, we introduce an SCD method based on motion detection and image recognition technologies to help inspector attendants detect the slope collapse. In this method, we use the foreground motion detection algorithm to identify the slope collapse of the scene of the river embankment. Since the moving targets in the foreground may not only be the slope collapse but also maybe some biology, we further use the image feature extraction and image recognition technology to recognize the foreground motion area, thus eliminating the influence of the biology on the detection results. Experimental results on the relevant scene data show that the proposed method can identify the slope collapse in real-time, and can effectively eliminate the motion interference of the biology, which has a high practical value.

Research on Wind Turbines Fault Diagnosis Technology Based on CMS Data Feature Extraction

As a rich clean and environmentally friendly renewable resources, wind energy has emerged as a strategic choice for countries around the world. Because the wind turbines often operate in severe working conditions such as variable load and large temperature difference they are prone to failures and possible shutdowns. The shutdowns however seriously affect the economic benefits of the wind turbines. Initiative maintenance has become a worldwide recognized scientific method for planning and determining preventive maintenance work, the implementation of this strategy relies on real-time condition monitoring and fault signal identification methods. The condition monitoring of wind turbine can help master the health state and power generation performance of wind turbine, so as to timely formulate maintenance strategies and adopt technical modification measures to improve power generation performance, reduce the down time of wind turbine, avoid the occurrence of major faults, save maintenance cost and improve power generation capacity. Therefore, a condition monitoring system is built on a wind turbine of Zhangjiakou, and a systematic signal analysis method is proposed, time-domain synchronous averaging technology, based on variable period, impulse signal feature extraction technology based on Teager and signal decomposition technology based on CEEMD. The proposed method realizes the signal analysis and feature extraction of non-stationary nonlinear, weak signal and frequency aliasing signals, and successfully diagnose the gearbox secondary meshing failure during the long-term monitoring. This confirms that the monitoring system methods and signal analysis technology proposed in this paper can effectively realize the condition monitoring and fault diagnosis of wind turbines.

Research on the Deviation Analysis of Utilization Hours in Wind Power Plants Based on Regulatory Data Fusion and Feature Extraction & Processing Technologies

The continuous elevation of wind power consumption pressure has proposed higher requirements for the technical means, which comprehensively support the wind power consumption. The power generation level of a wind power plant is measured through the utilization hours at present, but the factors influencing the utilization hours are scarcely analyzed and quantified. In order to accurately figure out the reasons for the different utilization hours of wind power plants located in the same area, the total power prediction, power plant data, standalone data, resources, etc. were summarized, and the contribution degrees of five dimensions—resources, equipment availability, AGC command balance degree, AGC command following performance, and power generation capacity—to the utilization hours were analyzed and quantified through the data fusion and feature extraction, in an effort to find the main factors influencing the utilization hours in different power plants. The customized measures were proposed according to the conclusions in order to increase the utilization hours for the power plants. The derived computing method was applied to the wind power plant data acquired by a company in a province, followed by the difference analysis of the utilization hours. It was found that in Mulan sectional area, the reason for partially low utilization hours in Jifeng and Xunfeng Wind Power Plants was resource factor, that for Daheishan and Songshan Wind Power Plants lied in equipment availability, and that for Jufeng Wind Power Plant was possibly equipment following performance or equipment power generation capacity. In addition, the AGC command balance degrees in all target power plants were at low levels.

Investigation of Biometric Identification Technology Based on biological Fingerprints and Facial Features

Biometric identification is largely dependent on feature extraction technology. As feature extraction techniques are increasingly mature, scholars have gradually turned their attention to the relevant problems between biometric characteristics. This paper reviews the characteristic extraction method of face and fingerprint analyzes the feature classification extraction method based on empirical knowledge and the depth learning-based computer logic sampling extraction method and compares existing solutions from the angle of image processing. Based on feature extraction, it will have prospected in the future biometric identification model of progress.