Papers by Keyword: Feature Extraction

Abstract: In the recent era- very frequently people come across health issues due to consumption of poor-quality food items- which leads to issues such as food poisoning, vomiting, diarrhea, etc., For a full development of fruits and vegetables, all the nutrients are necessary during its growth. But due to circumstances like soil defects, infections, water scarcity, waterlogging, etc., the vegetables & fruits gets infected with some diseases. So there arises a necessity of a system which inspects for any presence of disease in fruits & vegetables, with reduced manual intervention. This paper provides a detailed overview of a system developed using the Python programming language. Its aim is to recognize and classify various fruits and vegetables, while also identifying any diseases affecting them and determining the specific type of infection. In order to recognize the details accurately, the system is designed to use convolutional neural networks (CNN) and the results are displayed using computer vision techniques. The analysis, implementation, and future improvements of the proposed system are briefed in this paper. For this, we have used Anaconda navigator software (Jupyter notebook, IDLE).

Abstract: Replication, in general, is defined as repeating a study's technique and assessing whether the previous finding re-occurs. Research can become replicable when a person can copy the same content and arrive at the same conclusion as the original study. In this paper, an approach control mechanism for article observation replicas is been proposed. The usage of encrypted pictures or encoded attribute plots has been shown to be successful in preventing unwanted approach to models. The approach's efficiency has only been verified in image organization models and semantic analysis models but not in article recognition models. For the first time, encoded feature plots have proved to be successful in the control of article observation replicas. A safe and efficient technique based on completely homomorphic encryption is used and its usefulness for a variety of real data is been demonstrated. The suggested technique is the first to directly replicate an algorithm on ciphertext, which is one of the best performers on the plaintext feature selection problem. Furthermore, the suggested protocol is extensible to the scenario of more than three data owners.

Abstract: The choice of using speech to control the exoskeleton is based on the number of exoskeletons that are controlled using the EMG signal, where the EMG signal itself has the weakness of the complexity of the signal which is influenced by the position of the electrodes, as well as muscle fatigue. The purpose of this research is to develop an exoskeleton device using voice control based on embedded machine learning on a Raspberry Pi minicomputer. In this study, two feature extraction types namely mel-frequency cepstral coefficient (MFCC) and zero-crossing (ZC), and two machine learning algorithms, namely K-nearest Neighbor (K-NN) and Decision Tree (DT) were evaluated. The hand exoskeleton development consists of 3D hand design, microphone, Raspberry Pi 4B+, PCA9685 servo driver, and servo motor. Microphone was used to record voice commands given. After model evaluation, it was found that the MFCC extraction combined with the K-NN algorithm and the best accuracy (96±7.0%). In the implementation, we found that the accuracy is 79±14.46% and 90±14.14% for open and close commands.

Abstract: Breast cancer is ranked first as the most common cancer case affecting women in the world. Early detection of breast cancer can increase the chances of survival in patients. The role of the radiologist is necessary for the detection of breast cancer, and the radiologists often have limitations in conducting disease consultations with so many patients. The detection gives a subjective result because the process is based on the decision-making of the radiologists. In this work, we proposed a system to detect and classify breast cancer accurately to anticipate delays in patient handling and subjective result. We proposed a digital image processing method using mammograms to classify breast cancer into four categories based on tissue density, namely BI-RADS I, II, III, and IV. The main stages carried out in this research are images processing, feature extraction, data normalization, feature selection, classification, and parameter optimization. This method uses GLCM to extract texture features and two feature selection methods namely, RFE-RF and Chi-Square. The method was tested with various classifiers such as SVM, KNN, Random Forests, and Decision Trees. The hyper-parameters of the classifier were optimized using GridSearch. The final result is measure using accuracy. In this work, Random Forest with the RFE-RF gives the highest accuracy of 99.7%. Feature selection offers a significant impact on improving accuracy. The results of this work prove that our system can classify breast cancer with high accuracy. So that our system can solve problems to assist radiologists in screening mammograms and help make decisions to diagnose patients with breast cancer based on density.

Abstract: The increasing need for prosthetic hands for people with disabilities is one reason for innovation in the field of prosthetic hands to create the best prosthetic hand technology. In the design of EMG-based prosthetic hands, this is determined by several things, among others, the selection of features. The selection of the right features will determine the accuracy of the prosthetic hand Therefore, the purpose of this study is to analysis the time domain feature to obtain the best feature in classifying the hand motion. The contribution of this work is able to detect 4 movements in real time, namely hand close, flexion, extension, and relax. The Electromyograph signal is tapped using an electromyograph (EMG) dry electrode sensor in which there is a circuit of EMG instrumentation amplifier. Furthermore, the analog EMG signal data is processed through the ADC (Analog to Digital Converter) by using MCP3008 device. EMG signal data is processed in Raspberry Pi. A feature extraction process is applied to reduce data and determine the characteristics of each hand movement. Feature extraction used is MAV (mean absolute value), SSI (sign slope integral), VAR (variance), and RMS (root mean square). From the results of the four-time domain feature, then the best feature extraction is determined using scatter plot and Euclidean distance. The results that have been carried out on ten people with each person doing ten sets of movements (hand close, flexion, extension, relax), showing the best Euclidean distance results, is the RMS feature, with a value of 2608.07. This data is the result of the best feature extraction analysis through the method of calculating the distance of feature extraction data using Euclidean distance. This analysis of time domain feature is expected to be useful for further experiment in machine learning implementation so that it can be obtained an effective prosthetic hand.

Abstract: Rapid progress in Micro-Electromechanical System (MEMS) technique is making inertial sensors increasingly miniaturized, enabling it to be widely applied in people’s everyday life. Recent years, research and development of wireless input device based on MEMS inertial measurement unit (IMU) is receiving more and more attention. In this paper, a survey is made of the recent research on inertial pens based on MEMS-IMU. First, the advantage of IMU-based input is discussed, with comparison with other types of input systems. Then, based on the operation of an inertial pen, which can be roughly divided into four stages: motion sensing, error containment, feature extraction and recognition, various approaches employed to address the challenges facing each stage are introduced. Finally, while discussing the future prospect of the IMU-based input systems, it is suggested that the methods of autonomous and portable calibration of inertial sensor errors be further explored. The low-cost feature of an inertial pen makes it desirable that its calibration be carried out independently, rapidly, and portably. Meanwhile, some unique features of the operational environment of an inertial pen make it possible to simplify its error propagation model and expedite its calibration, making the technique more practically viable.

Abstract: Degradation state assessment of bearing is an important part of prognostic and health management (PHM) in rotating machinery. Generally, the energy distribution of frequency band is sensitive to degradation state for rolling bearing. Hence, a novel assessment method based on variational mode decomposition (VMD) and energy distribution is proposed in this work. Firstly, the VMD is used to decompose raw vibration signal into several components with different scales and frequency bands. These components is capable of reflecting the local characteristic of vibration signal. Then, the energy distribution of these components is utilized as feature vector. Finally, the different bearing states can be classified by the scatter plots of the first several principal components after principal component analysis (PCA). The analysis of an experimental dataset demonstrates the effectiveness of this methods. The comparative analysis shows the VMD is superior to traditional empirical mode decomposition (EMD) methods.

Abstract: For the situation that it is difficult to diagnose rolling bearings fault effectively for small samples, so it proposes a feature extraction method of rolling bearing based on local mean decomposition (LMD) energy feature. Due to the frequency domain distribution of vibration signals will change when different faults occur in rolling bearings, so it can use LMD energy feature method to extract the fault features of rolling bearings. The instances analysis and extracted results show that the LMD energy feature can extract the vibration signal fault feature of rolling bearings effectively.

Abstract: A CAD/CAM system has become a dominant system in modern production techniques. In the attempts to facilitate the transfer of design data and geometry between the huge varieties of the International working CAD/CAM systems, the International Standard for the Exchange of Product coding system (STEP) is introduced (ISO Step AP-203). Extraction of the entities is the first step in the automated programming either for process planning, product classification or in programming for Numerical Control Machine Tools. This step is the aim of this paper in a line of several steps to build a complete CAD/CAM system for manufacturing of complicated products. A methodology is developed to extract each entity. Software has been built using Visual Studio C# - Version 2010. Interface is designed giving all the entities and showing the geometrical shape of the product. Different cases have been tested to verify the developed software and full successfulresults were obtained.

Abstract: The essential features and scale of sensor data was discussed to monitor the tool anomaly in the machining process from the pattern variation of large scale sensor data such as vibration and effective power. The cycle data, the time series sensor data collected with an acceleration or power sensor in one periodical machining of the given groove shape, had been measured periodically. In this study, the graphic pattern formed by overwriting the time series cycle data on a specific coordinate system was treated as the “big sensor data”. The big data from the effective power sensor can stably respond to the cutting power changes and showed a strong possibility as a detecting device for tool anomaly such as abrasive wear and chipping. While the big data from the acceleration sensor only responded to a big event like the chattering vibration. The number of cycle data needed to generate the big sensor data also affected on the detection sensitivity for tool anomaly. It had been required a family of time series sensor data enough to represent the cutting power change as a visual graphic pattern.