China
Africa
Explore chapters and articles related to this topic
Multimodal Ambulatory Fall Risk Assessment in the Era of Big Data
Published in Ervin Sejdić, Tiago H. Falk, Signal Processing and Machine Learning for Biomedical Big Data, 2018
Due to the specific characteristics of our images, texture seems to be an appropriate feature for describing their contents (e.g., brick, tiles, rocks, carpet) [79]. Texture analysis has been an active research area, and numerous algorithms have been proposed based on different models, for example, gray-level co-occurrence (GLC) matrices and Markov random field (MRF) model [79,88]. In recent works, wavelets have become very popular due to their capacity to provide multiresolution analysis. In particular, the Gabor transform has mathematical and biological properties resembling the characteristics of human visual cortical cells, such as extracting texture features from images for segmentation, object detection, and biometric identification applications.
A Study on Feature Extraction and Classification Techniques for Melanoma Detection
Published in P. Madhumathy, M. Vinoth Kumar, R. Umamaheswari, Machine Learning and IoT for Intelligent Systems and Smart Applications, 2021
S. Poovizhi, T. R. Ganesh Babu, R. Praveena, J. Kirubakaran
Mallat [22] proposed a mathematical tool called wavelet transform for image processing applications where wavelets give both time and frequency information of the signal. A wavelet provides multiresolution analysis i.e represents the image on more than one scale. The advantage of multiresolution analysis is that the features undetected at one scale can be detected by the other scale. Wavelets can represent discontinuity of the image with fewer coefficients.
An Intelligent Scheme for Classification of Shunt Faults Including Atypical Faults in Double-Circuit Transmission Line
Published in Almoataz Y. Abdelaziz, Shady Hossam Eldeen Abdel Aleem, Anamika Yadav, Artificial Intelligence Applications in Electrical Transmission and Distribution Systems Protection, 2021
Valabhoju Ashok, Anamika Yadav, Mohammad Pazoki, Almoataz Y. Abdelaziz
The DWT is a time-frequency-based multi-resolution analysis technique to decompose the signal into diverse levels. The original signal is decomposed into two components: approximation coefficient and detail coefficient [14]. A data window of 1-cycle is chosen from the fault instant and subsequently DWT is employed to extract detail coefficients at level-6 of the fault signal using the “db4” wavelet.
Assessment of autonomic response in 6–12-month-old babies during the interaction with robot and avatar by means of thermal infrared imaging
Published in Quantitative InfraRed Thermography Journal, 2023
C. Filippini, D. Cardone, D. Perpetuini, A.M. Chiarelli, L.A. Petitto, A. Merla
Concerning the second approach, in order to analyse the thermal signals at different frequencies, but without loss of timing information, a wavelet transform was performed on both signals. In fact, thermal signals are nonstationary in nature, and thermal response time in infants can be different from subject to subject. The frequency bandwidth preserved by the filter included bands in which the skin blood flow oscillations occur, corresponding to the endothelium-related metabolic (0.008–0.02 Hz), neurogenic (0.02–0.05 Hz), myogenic (0.05–0.15 Hz), respiratory (0.15–0.4 Hz) and cardiac regulations (0.4–2 Hz) [39]. Exploring these frequency bands allowed us to investigate the principal factor that contributed to the signal frequency content under different experimental conditions. A continuous wavelet transform (Morse wavelet, gamma equal to 3 and time-bandwidth product equal to 60) was performed to investigate the frequency at which the highest signal power occurred (fHP). The wavelet transform provides a multiresolution analysis by decomposing the signals into a time-frequency space. The signal power content can thus be evaluated as a function of frequency and time [40].
A Hybrid PSO-ANN-based Fault Classification System for EHV Transmission Lines
Published in IETE Journal of Research, 2022
Pranav D. Raval, Ashit S. Pandya
A typical Discrete Wavelet Transform (DWT) uses two filter banks successively, low-pass filter and high-pass filter, as shown in Figure 3. It demonstrates the process of Multiresolution Analysis (MRA) done by the use of Wavelet transform. The signal S(n) is applied to two filter banks put subsequently as low-pass and high-pass filter. This is illustrated by Equations (1) and (2) how a signal S(n) is worked upon with a dyadic operation by a mother wavelet transform successively. The operations under MRA show the signal is firstly separated in two parts in the frequency band by the use of low-pass and high-pass filters. The output of the low-pass filter is further split and given to a second phase of low-pass and high-pass filter banks. The multiple filter banks used in the process of decomposition of signal S(n) extract the spectral information at every stage.
Glioma grade classification using wavelet transform-local binary pattern based statistical texture features and geometric measures extracted from MRI
Published in Journal of Experimental & Theoretical Artificial Intelligence, 2019
Manu Gupta, Venkateswaran Rajagopalan, B. V. V. S. N. Prabhakar Rao
Texture has been one of the most important characteristics that have been used to identify spatial variations or heterogeneity in brain tumour images (Davnall et al., 2012; Materka & Strzelecki, 1998). LBP (Ojala, Pietikäinen, & Mäenpää, 2000) is a powerful method to describe the local spatial structure of an image. It is used widely in computer vision applications, such as face recognition, because of its computational efficiency and high discriminative power (Sharma, Hussain, & Jurie, 2012). In this study, DWT and local contrast property of LBP algorithm are utilised to derive texture properties of tumorous MR images through multiresolution analysis. The advantage of multiresolution analysis–based feature extraction is that it gathers information at more than one resolution. The features that are not detected at one resolution might get detected at another resolution. In this work, texture features are extracted at multiple resolutions by computing LBP for sub-bands of tumorous MR images obtained from DWT.