Topic
Spectrogram
About: Spectrogram is a research topic. Over the lifetime, 5813 publications have been published within this topic receiving 81547 citations.
Papers published on a yearly basis
Papers
More filters
TL;DR: A sliding window algorithm of short frame length is suitable for differentiate the Mel-spectrogram of bird sound and the GRU network is connected and used as a classifier to directly output the prediction results.
24 citations
TL;DR: The time-series analysis method is used to accurately measure the similarity of the micro-Doppler signatures between the training and test data, thus providing improved gesture classification and it is shown that the DTW method achieves high gesture classification rates and is robust to time misalignment.
Abstract: Hand and arm gesture recognition using radio frequency (RF) sensing modality proves valuable in man–machine interfaces and smart environments. In this paper, we use the time-series analysis method to accurately measure the similarity of the micro-Doppler (MD) signatures between the training and test data, thus providing improved gesture classification. We characterize the MD signatures by the maximum instantaneous Doppler frequencies depicted in the spectrograms. In particular, we apply two machine learning (ML) techniques, namely, the dynamic time warping (DTW) method and the long short-term memory (LSTM) network. Both methods take into account the values as well as the temporal evolution and characteristics of the time-series data. It is shown that the DTW method achieves high gesture classification rates and is robust to time misalignment.
24 citations
TL;DR: A new, bilinear, cross-term suppressed and alias-free time-frequency representation (TFR) that has a higher resolution than the spectrogram with the same window width and is applied to interference excision in direct-sequence spread-spectrum communications.
24 citations
TL;DR: The TF-MUSIC algorithm is suitable for extracting a target response whose spectrum changes significantly during the observation, and the usefulness of this method was demonstrated.
24 citations
TL;DR: A deep convolutional neural network model capable to efficiently detect and classify epilepsy seizures based on the EEG spectrogram images is developed and proved to be a powerful tool in classifying EEG signals.
Abstract: Epilepsy disease is one of the most prevalent neurological disorders caused by malfunction of large symptoms number of neurons. That’s lead us to propose an automated approach to classify Electroencephalography (EEG) signals of the aforementioned pathology. To realize an efficient seizures detection the output of our classification is divided into three classes; normal, pre-ictal and ictal class. In fact, we propose to use the Short-Time Fourier Transform (STFT) as a non-stationary signal processing technique to extract useful information from the EEG signals. After that, we transform the STFT into a spectrogram image which will be used as an input in the classification process. In this context, we developed a deep convolutional neural network (CNN) model capable to efficiently detect and classify epilepsy seizures based on the EEG spectrogram images. It should be noted that the database used in this work is the publicly available EEG data set of Bonn University. In order to evaluate the performance of the proposed classification method several metrics are calculated, such as; sensitivity, specificity, accuracy and precision. The experimental results prove that the proposed method is a powerful tool in classifying EEG signals with a high average accuracy rate of 98.22%.
24 citations