Pub Date : 2020-10-17DOI: 10.1109/CISP-BMEI51763.2020.9263548
Sun Zhou, Jing Li
Extraction of Sharp-Wave Ripples (SWRs) from brain wave signals plays an important part in various medical studies of mammalian nervous systems. SWRs, playing a crucial role in memory consolidation, are oscillatory patterns in the mammalian brain hippocampus seen on an EEG during immobility and sleep. An SWR is composed of large-amplitude sharp waves accompanied by fast field potential oscillations known as ripple rhythms. However, most of the current commercial software for brain wave processing does not provide with an accurate SWR extraction function. Also, so far there are few literatures that fully explore the ripple detection method. Taking a fuller look at the characteristics of ripple events, an improved pipeline is presented to extract SWRs. The utility of detection based on the large-amplitude feature of SWRs will be weakened by another feature, fast oscillation. Therefore, to shield the extraction from that undesired influence, Hilbert transformation is suggested to restore the analytical signal in complex number field and then to obtain the envelope of the original EEG. Next, Gaussian window is adopted to get rid of some artifacts. Then, the central and the start and end segment of an SWR are successively determined with a sliding window. In addition, considering that the determination of the duration of a ripple also changes the frequency content of a detected, truncated ripple by the spectral leakage effect, which makes it hard to find the actual frequency of the rhythm, we add Hanning window to prevent that effect. From three sets of multi-channel in vivo recording EEG data obtained from different genotypes of mice, we detected SWR events with the proposed method, whose effectiveness and accuracy were validated.
{"title":"A Pipeline for Extraction of Sharp-Wave Ripples from Multi-Channel in vivo Recording EEG","authors":"Sun Zhou, Jing Li","doi":"10.1109/CISP-BMEI51763.2020.9263548","DOIUrl":"https://doi.org/10.1109/CISP-BMEI51763.2020.9263548","url":null,"abstract":"Extraction of Sharp-Wave Ripples (SWRs) from brain wave signals plays an important part in various medical studies of mammalian nervous systems. SWRs, playing a crucial role in memory consolidation, are oscillatory patterns in the mammalian brain hippocampus seen on an EEG during immobility and sleep. An SWR is composed of large-amplitude sharp waves accompanied by fast field potential oscillations known as ripple rhythms. However, most of the current commercial software for brain wave processing does not provide with an accurate SWR extraction function. Also, so far there are few literatures that fully explore the ripple detection method. Taking a fuller look at the characteristics of ripple events, an improved pipeline is presented to extract SWRs. The utility of detection based on the large-amplitude feature of SWRs will be weakened by another feature, fast oscillation. Therefore, to shield the extraction from that undesired influence, Hilbert transformation is suggested to restore the analytical signal in complex number field and then to obtain the envelope of the original EEG. Next, Gaussian window is adopted to get rid of some artifacts. Then, the central and the start and end segment of an SWR are successively determined with a sliding window. In addition, considering that the determination of the duration of a ripple also changes the frequency content of a detected, truncated ripple by the spectral leakage effect, which makes it hard to find the actual frequency of the rhythm, we add Hanning window to prevent that effect. From three sets of multi-channel in vivo recording EEG data obtained from different genotypes of mice, we detected SWR events with the proposed method, whose effectiveness and accuracy were validated.","PeriodicalId":346757,"journal":{"name":"2020 13th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130820130","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-10-17DOI: 10.1109/CISP-BMEI51763.2020.9263648
Hanwu Luo, Wenzheng Li, Wang Luo, Fang Li, Jun Chen, Yuan Xia
Object detection is widely used in many fields, such as intelligent security monitoring, smart city, power inspection, and so on. The object detection algorithm based on deep learning is a kind of storage intensive and computing intensive algorithm which is difficult to achieve on the embedded platform with limited storage and computing resources. In this paper, we choose mobinetv2, a lightweight neural network with few model parameters and strong feature extraction ability, to replace darknet53 as the backbone network of YOLOv3 algorithm. In addition, we use a model compression method based on channel pruning to compress the network model. This method compresses model to detecting objects on embedded ARM platform. Neon instruction and OpenMP technology are further used to optimize and accelerate the intensive computing of convolutional network, and finally achieve a real-time embedded object detection system.
{"title":"Embedded Object Detection System Based on Deep Neural Network","authors":"Hanwu Luo, Wenzheng Li, Wang Luo, Fang Li, Jun Chen, Yuan Xia","doi":"10.1109/CISP-BMEI51763.2020.9263648","DOIUrl":"https://doi.org/10.1109/CISP-BMEI51763.2020.9263648","url":null,"abstract":"Object detection is widely used in many fields, such as intelligent security monitoring, smart city, power inspection, and so on. The object detection algorithm based on deep learning is a kind of storage intensive and computing intensive algorithm which is difficult to achieve on the embedded platform with limited storage and computing resources. In this paper, we choose mobinetv2, a lightweight neural network with few model parameters and strong feature extraction ability, to replace darknet53 as the backbone network of YOLOv3 algorithm. In addition, we use a model compression method based on channel pruning to compress the network model. This method compresses model to detecting objects on embedded ARM platform. Neon instruction and OpenMP technology are further used to optimize and accelerate the intensive computing of convolutional network, and finally achieve a real-time embedded object detection system.","PeriodicalId":346757,"journal":{"name":"2020 13th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130919594","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-10-17DOI: 10.1109/CISP-BMEI51763.2020.9263670
Xia Li, Qing Chang
Accurate non-rigid registration of chest radiographs facilitates image diagnosis and occupies an important position in medical image analysis. In this paper, we proposed a non-rigid registration framework that combines the advantages of B-spline FFD (free form deformation) and inertial demons. The proposed method applied B-spline FFD to match structures in the lung area and prevent lesion being destroyed; at the same time, the inertial demons model is used to refine the detail of results observed by FFD. Temporal subtraction images created from the chest radiography image pairs are given to demonstrate the registration accuracy. Multiple experiments on clinical data have shown that the proposed algorithm is more accurate in chest radiographs registration than the widely used B-spline FFD and demons algorithm alone.
{"title":"A Hybrid Nonrigid Medical Image Registration Method on Chest Radiography","authors":"Xia Li, Qing Chang","doi":"10.1109/CISP-BMEI51763.2020.9263670","DOIUrl":"https://doi.org/10.1109/CISP-BMEI51763.2020.9263670","url":null,"abstract":"Accurate non-rigid registration of chest radiographs facilitates image diagnosis and occupies an important position in medical image analysis. In this paper, we proposed a non-rigid registration framework that combines the advantages of B-spline FFD (free form deformation) and inertial demons. The proposed method applied B-spline FFD to match structures in the lung area and prevent lesion being destroyed; at the same time, the inertial demons model is used to refine the detail of results observed by FFD. Temporal subtraction images created from the chest radiography image pairs are given to demonstrate the registration accuracy. Multiple experiments on clinical data have shown that the proposed algorithm is more accurate in chest radiographs registration than the widely used B-spline FFD and demons algorithm alone.","PeriodicalId":346757,"journal":{"name":"2020 13th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"92 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133535882","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-10-17DOI: 10.1109/CISP-BMEI51763.2020.9263594
B. Wang, Wenming Ma
With the rapid development of science and technology, huge amounts of information fill people’s lives, and the accompanying information overload phenomenon has become an urgent problem to be solved. Because the recommendation system can quickly find the products users want in the massive item information, to a certain extent It has attracted much attention to solve the problem of information overload. Matrix factorization is a commonly used technique in recommendation systems. It can effectively improve the recommendation effect when the scoring matrix is sparse. However, due to its own reasons, matrix factorization has many problems such as sparseness, cold start, and low interpretability. In the field of deep learning, because the normalization technology BatchNorm can optimize the training process, accelerate the training speed and make the training results more stable, it has been studied by a large number of scholars. In this paper, Matrix Factorization Based on BatchNorm and Preference Bias is proposed. BatchNorm is combined with matrix factorization, user and item preferences are added, and Adam algorithm is used for optimization. Experiments show that the algorithm in this paper has a good recommendation effect on sparse matrix.
{"title":"Matrix Factorization Based on BatchNorm and Preference Bias","authors":"B. Wang, Wenming Ma","doi":"10.1109/CISP-BMEI51763.2020.9263594","DOIUrl":"https://doi.org/10.1109/CISP-BMEI51763.2020.9263594","url":null,"abstract":"With the rapid development of science and technology, huge amounts of information fill people’s lives, and the accompanying information overload phenomenon has become an urgent problem to be solved. Because the recommendation system can quickly find the products users want in the massive item information, to a certain extent It has attracted much attention to solve the problem of information overload. Matrix factorization is a commonly used technique in recommendation systems. It can effectively improve the recommendation effect when the scoring matrix is sparse. However, due to its own reasons, matrix factorization has many problems such as sparseness, cold start, and low interpretability. In the field of deep learning, because the normalization technology BatchNorm can optimize the training process, accelerate the training speed and make the training results more stable, it has been studied by a large number of scholars. In this paper, Matrix Factorization Based on BatchNorm and Preference Bias is proposed. BatchNorm is combined with matrix factorization, user and item preferences are added, and Adam algorithm is used for optimization. Experiments show that the algorithm in this paper has a good recommendation effect on sparse matrix.","PeriodicalId":346757,"journal":{"name":"2020 13th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115488096","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-10-17DOI: 10.1109/CISP-BMEI51763.2020.9263655
Jiaxin Cheng, Jun Zhong, Handing Wang, Xu Tang, Changzhe Jiao, Hong Zhou
In this paper, we proposed an effective method to obtain the R wave concept to estimate heart rate from electrocardiogram signals produced by a wearable electro-cardiogram(ECG) device. The multiple instance adaptive co-sine/coherent estimator(MI-ACE) is a multiple instance learning method that can learn the target concept from imprecisely labeled data. However, the R wave concepts estimated by MI-ACE are dependent on initialization strategy of MI-ACE. Thus, the heart rate estimation results are undetermined with different initialization. Evolutionary algorithm is a global optimization method that simulates natural processes. To overcome this problem, we pro-posed the evolutionary optimized MI-ACE algorithm(MI-ACE-Evo) which combines MI-ACE with an evolutionary optimization to learn the R wave target concept, which will make heart rate estimation more effective and not affected by varies initialization of MI-ACE. The experimental results show that the R wave concept learned by MI-ACE-Evo is more discriminative and the heartrate estimation results are superior to that of the original MI-ACE method.
{"title":"Evolutionary Optimized Multiple Instance Concept Learning for Beat-to-Beat Heart Rate Estimation from Electrocardiograms","authors":"Jiaxin Cheng, Jun Zhong, Handing Wang, Xu Tang, Changzhe Jiao, Hong Zhou","doi":"10.1109/CISP-BMEI51763.2020.9263655","DOIUrl":"https://doi.org/10.1109/CISP-BMEI51763.2020.9263655","url":null,"abstract":"In this paper, we proposed an effective method to obtain the R wave concept to estimate heart rate from electrocardiogram signals produced by a wearable electro-cardiogram(ECG) device. The multiple instance adaptive co-sine/coherent estimator(MI-ACE) is a multiple instance learning method that can learn the target concept from imprecisely labeled data. However, the R wave concepts estimated by MI-ACE are dependent on initialization strategy of MI-ACE. Thus, the heart rate estimation results are undetermined with different initialization. Evolutionary algorithm is a global optimization method that simulates natural processes. To overcome this problem, we pro-posed the evolutionary optimized MI-ACE algorithm(MI-ACE-Evo) which combines MI-ACE with an evolutionary optimization to learn the R wave target concept, which will make heart rate estimation more effective and not affected by varies initialization of MI-ACE. The experimental results show that the R wave concept learned by MI-ACE-Evo is more discriminative and the heartrate estimation results are superior to that of the original MI-ACE method.","PeriodicalId":346757,"journal":{"name":"2020 13th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114485559","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-10-17DOI: 10.1109/CISP-BMEI51763.2020.9263627
Sijia Sun, Hangfang Zhao
The perturbations of sound speed profiles (SSPs) has great influence on sound propagation. Empirical orthogonal functions (EOFs) are often used to simplify the description of sound speed profiles. However, when the unevenness of seawater, such as internal wave and turbulence exists, the regularization operation will result in a significant decrease in the reconstruction accuracy of sound speed. In this paper, the dictionary learning, a form of unsupervised machine learning, is used to generate non-orthogonal entries of sound speed profiles, OMP algorithm is used in sparse coding, while K-SVD algorithm is used in dictionary updating. Because dictionary learning does not require the use of orthogonal conditions, it is more flexible for training data, and thus can use fewer atomic combinations to achieve higher reconstruction accuracy. The reconstruction performance of EOFs and LDs was tested with HYCOM data. The results show that compared with EOFs, LDs can better explain the perturbations of sound speed profiles with a few entries. Dictionary learning can improve the sparsity of sound speed profiles and improve the reconstruction accuracy of sound speed profiles.
{"title":"Sparse Representation of Sound Speed Profiles Based on Dictionary Learning","authors":"Sijia Sun, Hangfang Zhao","doi":"10.1109/CISP-BMEI51763.2020.9263627","DOIUrl":"https://doi.org/10.1109/CISP-BMEI51763.2020.9263627","url":null,"abstract":"The perturbations of sound speed profiles (SSPs) has great influence on sound propagation. Empirical orthogonal functions (EOFs) are often used to simplify the description of sound speed profiles. However, when the unevenness of seawater, such as internal wave and turbulence exists, the regularization operation will result in a significant decrease in the reconstruction accuracy of sound speed. In this paper, the dictionary learning, a form of unsupervised machine learning, is used to generate non-orthogonal entries of sound speed profiles, OMP algorithm is used in sparse coding, while K-SVD algorithm is used in dictionary updating. Because dictionary learning does not require the use of orthogonal conditions, it is more flexible for training data, and thus can use fewer atomic combinations to achieve higher reconstruction accuracy. The reconstruction performance of EOFs and LDs was tested with HYCOM data. The results show that compared with EOFs, LDs can better explain the perturbations of sound speed profiles with a few entries. Dictionary learning can improve the sparsity of sound speed profiles and improve the reconstruction accuracy of sound speed profiles.","PeriodicalId":346757,"journal":{"name":"2020 13th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"66 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116018667","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-10-17DOI: 10.1109/CISP-BMEI51763.2020.9263540
Xiao Sun, Jindong Xu
The existing remote sensing image dehazing methods based on deep learning networks usually use pairs of clear images and corresponding haze images to train the model. However, pairs of clear images and their haze counterparts are extremely lacking, and synthetically haze images could not accurately simulate the real haze generation process in real-world scenarios. To address this problem, a cascade method combining two GANs (generative adversarial networks) is proposed. It contains a learning-to-haze GAN (UGAN) and learning-to-dehaze GAN (PAGAN). UGAN learns how to haze remote sensing images with unpaired clear and haze images sets, and then guides the PAGAN to learn how to correctly dehaze such images. To reduce the discrepancy between real haze and synthetic haze images, we added self-attention mechanism to PAGAN. The details can be generated using cues from all feature locations. Moreover, the discriminator could check that highly detailed features in distant portions of the images that are consistent with each other. Compared with other dehazing methods, this algorithm does not require numerous pairs of images to train the network repeatedly. And the results show that the cascaded generative adversarial networks has visual and quantitative effectiveness for the removal of haze, thin clouds.
{"title":"Remote Sensing Images Dehazing Algorithm based on Cascade Generative Adversarial Networks","authors":"Xiao Sun, Jindong Xu","doi":"10.1109/CISP-BMEI51763.2020.9263540","DOIUrl":"https://doi.org/10.1109/CISP-BMEI51763.2020.9263540","url":null,"abstract":"The existing remote sensing image dehazing methods based on deep learning networks usually use pairs of clear images and corresponding haze images to train the model. However, pairs of clear images and their haze counterparts are extremely lacking, and synthetically haze images could not accurately simulate the real haze generation process in real-world scenarios. To address this problem, a cascade method combining two GANs (generative adversarial networks) is proposed. It contains a learning-to-haze GAN (UGAN) and learning-to-dehaze GAN (PAGAN). UGAN learns how to haze remote sensing images with unpaired clear and haze images sets, and then guides the PAGAN to learn how to correctly dehaze such images. To reduce the discrepancy between real haze and synthetic haze images, we added self-attention mechanism to PAGAN. The details can be generated using cues from all feature locations. Moreover, the discriminator could check that highly detailed features in distant portions of the images that are consistent with each other. Compared with other dehazing methods, this algorithm does not require numerous pairs of images to train the network repeatedly. And the results show that the cascaded generative adversarial networks has visual and quantitative effectiveness for the removal of haze, thin clouds.","PeriodicalId":346757,"journal":{"name":"2020 13th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"381 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123483949","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-10-17DOI: 10.1109/CISP-BMEI51763.2020.9263647
Rubing Xi, Lei Jin
The channels of the multi-temporal SAR image have strong scattering target distribution in different positions. Focus on this, this paper propose the intensity segregation representation model for the multi-temporal SAR image restoration. This new variational regularization model based on the intensity separation of the multi-temporal SAR image is composed of two sub-models. The first one is a variational regularization model for the intensity component of the image, where the noise is assumed to be multiplicative, and the regularization term is the total variation. A fixed point iterative algorithm is used to solve the Euler-Lagrangian equation of the first sub-model. The second sub-model is the vectorial variational regularization model for the vector component of the image, which is obtained by the assumption that the noise is multiplicative. And the vectorial total variation norm of the vector defined on the unit sphere is obtained. A partial differential equation method is used to get the differential iterative algorithm to solve the Euler-Lagrangian equation of the second sub-model. In this paper, the intensity separation model is applied to the multi-temporal SAR image despeckling. The strong scattering target is well preserved while the good efficient of despeckling is obtained. In summary, this method is proved to highly promote the ability of distinguish different kinds of surface target of the multi-temporal SAR image.
{"title":"An Intensity Separated Variational Regularization Model for Multichannel Image Enhancement","authors":"Rubing Xi, Lei Jin","doi":"10.1109/CISP-BMEI51763.2020.9263647","DOIUrl":"https://doi.org/10.1109/CISP-BMEI51763.2020.9263647","url":null,"abstract":"The channels of the multi-temporal SAR image have strong scattering target distribution in different positions. Focus on this, this paper propose the intensity segregation representation model for the multi-temporal SAR image restoration. This new variational regularization model based on the intensity separation of the multi-temporal SAR image is composed of two sub-models. The first one is a variational regularization model for the intensity component of the image, where the noise is assumed to be multiplicative, and the regularization term is the total variation. A fixed point iterative algorithm is used to solve the Euler-Lagrangian equation of the first sub-model. The second sub-model is the vectorial variational regularization model for the vector component of the image, which is obtained by the assumption that the noise is multiplicative. And the vectorial total variation norm of the vector defined on the unit sphere is obtained. A partial differential equation method is used to get the differential iterative algorithm to solve the Euler-Lagrangian equation of the second sub-model. In this paper, the intensity separation model is applied to the multi-temporal SAR image despeckling. The strong scattering target is well preserved while the good efficient of despeckling is obtained. In summary, this method is proved to highly promote the ability of distinguish different kinds of surface target of the multi-temporal SAR image.","PeriodicalId":346757,"journal":{"name":"2020 13th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"71 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117179230","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-10-17DOI: 10.1109/CISP-BMEI51763.2020.9263669
Zhesen Cui, Xiaolei Hou, Hu Zhou, Wei Lian, Jinran Wu
The slime mould algorithm (SMA) is a recently developed meta-heuristic optimization algorithm which is based on the oscillation mode of slime mould in nature. However, the SMA is often trapped in local optima for global continuous optimization problems. To strengthen SMA’s exploration for global optimum, we propose a modified SMA, which takes randomization based on a Levy distribution instead of the traditional uniform one, namely LF-SMA. Our LF-SMA is integrated with Levy-flight guidance to its optimal paths for connecting food with excellent exploratory propensity. Experimental results show that the proposed LF-SMA achieves better performance in 13 benchmark test functions and one investigated engineering case in terms of both computation cost and solution.
{"title":"Modified Slime Mould Algorithm via Levy Flight","authors":"Zhesen Cui, Xiaolei Hou, Hu Zhou, Wei Lian, Jinran Wu","doi":"10.1109/CISP-BMEI51763.2020.9263669","DOIUrl":"https://doi.org/10.1109/CISP-BMEI51763.2020.9263669","url":null,"abstract":"The slime mould algorithm (SMA) is a recently developed meta-heuristic optimization algorithm which is based on the oscillation mode of slime mould in nature. However, the SMA is often trapped in local optima for global continuous optimization problems. To strengthen SMA’s exploration for global optimum, we propose a modified SMA, which takes randomization based on a Levy distribution instead of the traditional uniform one, namely LF-SMA. Our LF-SMA is integrated with Levy-flight guidance to its optimal paths for connecting food with excellent exploratory propensity. Experimental results show that the proposed LF-SMA achieves better performance in 13 benchmark test functions and one investigated engineering case in terms of both computation cost and solution.","PeriodicalId":346757,"journal":{"name":"2020 13th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"55 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125861478","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-10-17DOI: 10.1109/CISP-BMEI51763.2020.9263630
Xin Liu, Qihan Hu, H. Yuan, Cuiwei Yang
Due to the presence of motion artifacts (MAs), heart rate monitoring using PPG sensors in daily life and physical exercise is challenging, and there have been many studies on MA removal algorithms. However, most studies do not consider the quality evaluation of PPG signal before the MA removal. In this way, removing the MA directly regardless of whether there is motion artifact signal is not only a waste of computing resources, but also easy to introduce new noise. In this paper, the MA detection in PPG signal is performed by dividing the original signal into 6s signal segments and calculating the amplitude mean difference function (AMDF). Then the obtained AMDF is converted into a 2-D image through the Gramian Angular Field (GAF), and then classified by the Convolutional Neural Networks (CNN) classifier, so as to distinguish the contaminated signal and clean signal. In the subsequent processing, only the contaminated signal needs to remove the MAs, and the clean signal segment can be directly used for heart rate estimation. In this study, we achieve a classification accuracy of 0.966 in the local database, and a classification accuracy of 0.946 in the BIDMC PPG and Respiration Dataset published by PhysioNet. With the combination of feature extraction and SVM classifier, the proposed method has significantly improved the results.
{"title":"Motion Artifact Detection in PPG Signals Based on Gramian Angular Field and 2-D-CNN","authors":"Xin Liu, Qihan Hu, H. Yuan, Cuiwei Yang","doi":"10.1109/CISP-BMEI51763.2020.9263630","DOIUrl":"https://doi.org/10.1109/CISP-BMEI51763.2020.9263630","url":null,"abstract":"Due to the presence of motion artifacts (MAs), heart rate monitoring using PPG sensors in daily life and physical exercise is challenging, and there have been many studies on MA removal algorithms. However, most studies do not consider the quality evaluation of PPG signal before the MA removal. In this way, removing the MA directly regardless of whether there is motion artifact signal is not only a waste of computing resources, but also easy to introduce new noise. In this paper, the MA detection in PPG signal is performed by dividing the original signal into 6s signal segments and calculating the amplitude mean difference function (AMDF). Then the obtained AMDF is converted into a 2-D image through the Gramian Angular Field (GAF), and then classified by the Convolutional Neural Networks (CNN) classifier, so as to distinguish the contaminated signal and clean signal. In the subsequent processing, only the contaminated signal needs to remove the MAs, and the clean signal segment can be directly used for heart rate estimation. In this study, we achieve a classification accuracy of 0.966 in the local database, and a classification accuracy of 0.946 in the BIDMC PPG and Respiration Dataset published by PhysioNet. With the combination of feature extraction and SVM classifier, the proposed method has significantly improved the results.","PeriodicalId":346757,"journal":{"name":"2020 13th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"60 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124633499","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：