Pub Date : 2020-08-07DOI: 10.1109/ColCACI50549.2020.9247855
Luis Felipe Buitrago Castro, Luis Fernando Enriquez Santacruz, M. B. S. Sánchez
Prolonged support or premature weaning of mechanical ventilation leads to several complications of cardiopulmonary physiology. Recently, work of breathing is proposed as an alternative that provides objective information about the weaning process. However, the availability and ease of use of techniques for its estimation in a clinical context are limited. Thus, the application of computerized methods for work of breathing estimation becomes necessary to assist professionals. In this article, we compare the performance of different machine learning techniques in the work of breathing estimation tasks. The problem is divided into two classes: high and low work of breathing, based on information extracted from the pressure, volume, and flow signals recorded by the mechanical ventilator. The classification algorithms used were: support vector machines, neural networks, k nearest neighbors, which were trained and tested on ventilatory signals of subjects with high and low work of breathing. The results show that the classification system can recognize the work of breathing levels with an accuracy of up to 80%.
{"title":"Work of Breathing Estimation during Spontaneous Breathing Test using Machine Learning Techniques","authors":"Luis Felipe Buitrago Castro, Luis Fernando Enriquez Santacruz, M. B. S. Sánchez","doi":"10.1109/ColCACI50549.2020.9247855","DOIUrl":"https://doi.org/10.1109/ColCACI50549.2020.9247855","url":null,"abstract":"Prolonged support or premature weaning of mechanical ventilation leads to several complications of cardiopulmonary physiology. Recently, work of breathing is proposed as an alternative that provides objective information about the weaning process. However, the availability and ease of use of techniques for its estimation in a clinical context are limited. Thus, the application of computerized methods for work of breathing estimation becomes necessary to assist professionals. In this article, we compare the performance of different machine learning techniques in the work of breathing estimation tasks. The problem is divided into two classes: high and low work of breathing, based on information extracted from the pressure, volume, and flow signals recorded by the mechanical ventilator. The classification algorithms used were: support vector machines, neural networks, k nearest neighbors, which were trained and tested on ventilatory signals of subjects with high and low work of breathing. The results show that the classification system can recognize the work of breathing levels with an accuracy of up to 80%.","PeriodicalId":446750,"journal":{"name":"2020 IEEE Colombian Conference on Applications of Computational Intelligence (IEEE ColCACI 2020)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115242435","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 : 2020-08-07DOI: 10.1109/ColCACI50549.2020.9247924
A. Orjuela-Cañón, O. J. Perdomo-Charry, C. H. Valencia-Niño, Leonardo Forero
Commonly, electromyography (EMG) signals have been employed for movements or pattern classification. For this, different digital signals processing methods are applied to extract features, before a classification stage. The present work deals with a proposal based on the use of image classification employing deep learning techniques. The images were obtained from a spectrogram analysis as a previous process of the convolutional neural network employment. Then, a classification of five positions from wrist movements is carried out the model. Results showed that the accuracy is comparable to similar techniques employed with a shallow neural network and a deep neural network applied to the same dataset.
{"title":"Convolutional neural network proposal for wrist position classification from electromyography signals","authors":"A. Orjuela-Cañón, O. J. Perdomo-Charry, C. H. Valencia-Niño, Leonardo Forero","doi":"10.1109/ColCACI50549.2020.9247924","DOIUrl":"https://doi.org/10.1109/ColCACI50549.2020.9247924","url":null,"abstract":"Commonly, electromyography (EMG) signals have been employed for movements or pattern classification. For this, different digital signals processing methods are applied to extract features, before a classification stage. The present work deals with a proposal based on the use of image classification employing deep learning techniques. The images were obtained from a spectrogram analysis as a previous process of the convolutional neural network employment. Then, a classification of five positions from wrist movements is carried out the model. Results showed that the accuracy is comparable to similar techniques employed with a shallow neural network and a deep neural network applied to the same dataset.","PeriodicalId":446750,"journal":{"name":"2020 IEEE Colombian Conference on Applications of Computational Intelligence (IEEE ColCACI 2020)","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123996784","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 : 2020-08-07DOI: 10.1109/ColCACI50549.2020.9248714
D. C. R. Arroyo, A. Florez, D. Flores, R. Romero, Liang Zhao
Echo State Network (ESN) has been widely studied and applied to many problems due to the simplicity of its training phase. This is because since in this network only the output weights are trained, avoiding to deal with the gradient’s vanishing problem presents in most of the recurrent neural networks. However, this technique has been criticized recently because of the echo property limitation and its random topology that may cause chaotic activity in the reservoir layer. In this paper, we present an application of the classic ESN model modifying the reservoir topology to a non-random approaches: clustered and complex networks, as an alternative solution to the chaotic activity problem. Further, the modified and classical models are compared considering two study cases: Rössler and Lorenz systems. Numerical experiments show that the proposed model has a better performance than the classical model.
{"title":"Echo State Network Performance Analysis using Non-random Topologies","authors":"D. C. R. Arroyo, A. Florez, D. Flores, R. Romero, Liang Zhao","doi":"10.1109/ColCACI50549.2020.9248714","DOIUrl":"https://doi.org/10.1109/ColCACI50549.2020.9248714","url":null,"abstract":"Echo State Network (ESN) has been widely studied and applied to many problems due to the simplicity of its training phase. This is because since in this network only the output weights are trained, avoiding to deal with the gradient’s vanishing problem presents in most of the recurrent neural networks. However, this technique has been criticized recently because of the echo property limitation and its random topology that may cause chaotic activity in the reservoir layer. In this paper, we present an application of the classic ESN model modifying the reservoir topology to a non-random approaches: clustered and complex networks, as an alternative solution to the chaotic activity problem. Further, the modified and classical models are compared considering two study cases: Rössler and Lorenz systems. Numerical experiments show that the proposed model has a better performance than the classical model.","PeriodicalId":446750,"journal":{"name":"2020 IEEE Colombian Conference on Applications of Computational Intelligence (IEEE ColCACI 2020)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132454459","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 : 2020-08-07DOI: 10.1109/ColCACI50549.2020.9247874
John Euler Chamorro Fuertes, Jairo Jose Marin Arciniegas, Pablo Bustos García de Castro, Oscar Andrés Vivas Albán
The present paper was developed in order to show the feasibility of using touch-pressure sensors and RoboComp framework, in the PhantomX hexapod robot, so that it can develop displacement. Previously, the robot included some complementary tools to its default version, so that it can develop two types of gait: regular and adaptive. A comparison of results was developed between two types of gates based on the calculation of joints angles from the kinematic model of the robot. For the adaptive gait, a stabilization system was developed with the use of touch-pressure sensors to locate the support points on which the robot’s legs can be kept stable. The results show that the robot can perform movements in a satisfactory way, although a small difference is generated between the trajectories sent and trajectories executed due to the use of some tools and software, this does not prevent good performance in locomotion.
{"title":"Locomotion Control of PhantomX Hexapod Robot with Touch-Pressure Sensor and RoboComp","authors":"John Euler Chamorro Fuertes, Jairo Jose Marin Arciniegas, Pablo Bustos García de Castro, Oscar Andrés Vivas Albán","doi":"10.1109/ColCACI50549.2020.9247874","DOIUrl":"https://doi.org/10.1109/ColCACI50549.2020.9247874","url":null,"abstract":"The present paper was developed in order to show the feasibility of using touch-pressure sensors and RoboComp framework, in the PhantomX hexapod robot, so that it can develop displacement. Previously, the robot included some complementary tools to its default version, so that it can develop two types of gait: regular and adaptive. A comparison of results was developed between two types of gates based on the calculation of joints angles from the kinematic model of the robot. For the adaptive gait, a stabilization system was developed with the use of touch-pressure sensors to locate the support points on which the robot’s legs can be kept stable. The results show that the robot can perform movements in a satisfactory way, although a small difference is generated between the trajectories sent and trajectories executed due to the use of some tools and software, this does not prevent good performance in locomotion.","PeriodicalId":446750,"journal":{"name":"2020 IEEE Colombian Conference on Applications of Computational Intelligence (IEEE ColCACI 2020)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126924318","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 : 2020-08-07DOI: 10.1109/ColCACI50549.2020.9247916
Paola Andrea Zárate Luna, Jesús Alfonso López Sotelo
For approximately 50 years, artificial neural networks have been playing a decisive role in the technological advances of the world, however, their application in the treatment of satellite images has not reached the expected potential since researchers have had to face to several problems such as object recognition, classification and semantic segmentation in images of low spatial resolution due to the high costs generated by building an optimal training and testing data set. This article presents the systematic review of large research literature and the most relevant papers presented in the last decade. The main sources chosen for the review were the IEEE digital library, the indexing of the SCOPUS system database and the Science Direct repository, with a total search of 386 articles related to the case study that after applying different filters, Inclusion and exclusion criteria are deepened in detail with 30 of them, finding an ascending scale in the amount of research developed in recent years, demonstrating the great interest and growth of this type of artificial intelligence technique.
近50年来，人工神经网络在世界技术进步中发挥了举足轻重的作用，然而，由于构建最优训练和测试数据集的成本高昂，研究人员不得不面对低空间分辨率图像中的目标识别、分类和语义分割等问题，人工神经网络在卫星图像处理中的应用并没有达到预期的潜力。本文对近十年来的大量研究文献和最相关的论文进行了系统回顾。综述选取的主要来源为IEEE数字图书馆、SCOPUS系统数据库的索引和Science Direct repository，共检索了与案例研究相关的386篇文章，在应用不同的筛选后，对其中30篇的纳入和排除标准进行了详细的深化，发现近年来的研究数量呈上升趋势，显示了这类人工智能技术的巨大兴趣和增长。
{"title":"Systematic Literature Review: Artificial Neural Networks Applied in Satellite Images","authors":"Paola Andrea Zárate Luna, Jesús Alfonso López Sotelo","doi":"10.1109/ColCACI50549.2020.9247916","DOIUrl":"https://doi.org/10.1109/ColCACI50549.2020.9247916","url":null,"abstract":"For approximately 50 years, artificial neural networks have been playing a decisive role in the technological advances of the world, however, their application in the treatment of satellite images has not reached the expected potential since researchers have had to face to several problems such as object recognition, classification and semantic segmentation in images of low spatial resolution due to the high costs generated by building an optimal training and testing data set. This article presents the systematic review of large research literature and the most relevant papers presented in the last decade. The main sources chosen for the review were the IEEE digital library, the indexing of the SCOPUS system database and the Science Direct repository, with a total search of 386 articles related to the case study that after applying different filters, Inclusion and exclusion criteria are deepened in detail with 30 of them, finding an ascending scale in the amount of research developed in recent years, demonstrating the great interest and growth of this type of artificial intelligence technique.","PeriodicalId":446750,"journal":{"name":"2020 IEEE Colombian Conference on Applications of Computational Intelligence (IEEE ColCACI 2020)","volume":"64 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121096083","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 : 2020-08-07DOI: 10.1109/ColCACI50549.2020.9247926
Elizabeth Martinez, C. Calderón, Hans Garcia, H. Arguello
A Brain tumour is a collection of an abnormal mass of tissue that can be grown as cancerous. This pathology can be detected using noninvasive techniques such as CT and MR. Despite CT can form a three-dimensional computer model by taking multiple X-rays shots, the MRI scans are highly preferred since they do not use ionizing energy on its captures and they also provide sufficient information to confirm a diagnosis, however, MRI scans have a lot of noise which can reduce the accuracy of the diagnosis. Therefore, many works in the state of the art try to solve these issue using first a filtering method to clear the noise and then a semantic classification algorithm such feature pyramid network, mask R CNN and random forest classifiers trained over the images acquired with MRI technique extracting grayscale intensity, spatial proximity and texture similarity features, however, segmentation image using these methods does not have sufficient accuracy. Thus, this work proposes to look forward over the FLAIR images on the BRATS 2015 training dataset that is composed by 155 captures of axial cuts from where the principal and adjacent layers that have the highest amount of information are used to reformulate and increase data features that lead on a pixel-based classifier U-net proposed performs a semantic segmentation with a precision of 76%, which improves in up to 23% precision compared with the random forest-based method that obtained a 53% of precision.
{"title":"MRI Brain Tumour Segmentation using a CNN Over a Multi-parametric Feature Extraction","authors":"Elizabeth Martinez, C. Calderón, Hans Garcia, H. Arguello","doi":"10.1109/ColCACI50549.2020.9247926","DOIUrl":"https://doi.org/10.1109/ColCACI50549.2020.9247926","url":null,"abstract":"A Brain tumour is a collection of an abnormal mass of tissue that can be grown as cancerous. This pathology can be detected using noninvasive techniques such as CT and MR. Despite CT can form a three-dimensional computer model by taking multiple X-rays shots, the MRI scans are highly preferred since they do not use ionizing energy on its captures and they also provide sufficient information to confirm a diagnosis, however, MRI scans have a lot of noise which can reduce the accuracy of the diagnosis. Therefore, many works in the state of the art try to solve these issue using first a filtering method to clear the noise and then a semantic classification algorithm such feature pyramid network, mask R CNN and random forest classifiers trained over the images acquired with MRI technique extracting grayscale intensity, spatial proximity and texture similarity features, however, segmentation image using these methods does not have sufficient accuracy. Thus, this work proposes to look forward over the FLAIR images on the BRATS 2015 training dataset that is composed by 155 captures of axial cuts from where the principal and adjacent layers that have the highest amount of information are used to reformulate and increase data features that lead on a pixel-based classifier U-net proposed performs a semantic segmentation with a precision of 76%, which improves in up to 23% precision compared with the random forest-based method that obtained a 53% of precision.","PeriodicalId":446750,"journal":{"name":"2020 IEEE Colombian Conference on Applications of Computational Intelligence (IEEE ColCACI 2020)","volume":"319 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116231781","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 : 2020-08-07DOI: 10.1109/ColCACI50549.2020.9247857
S. Gutiérrez, Noel Pérez, D. Benítez, S. Zapata, D. Augot
Culicoides biting midges are transmission vectors of various diseases affecting humans and animals around the world. An optimal and fast classification method for these and other species have been a challenge and a necessity, especially in areas with limited resources and public health problems. In this work, we developed a mobile application to classify two Culicoides species using the morphological pattern analysis of their wings. The app implemented an automatic classification method based on the calculation of seven morphological features extracted from the wing images and a support vector machine classifier to produce the final classification of Pusillus or Obsoletus class. The proposed app was validated on an experimental dataset with 87 samples, reaching an outstanding mean of AUC score of 0.98 in the classification stage. Besides, we assessed the app feasibility using the mean of time and battery consumption metrics on two different emulators. The obtained scores of 12 and 7 s and 0.11 and 0.03 mAh for the phone and tablet emulators are satisfactory when developing mobile applications. Finally, reducing the feature space using an external wrapper method provided us a considerable improvement in the classification performance, AUC scores from 0.95 to 0.98, and decreasing the volume of information in training stages. Thus, these results enable the proposed app as an excellent approximation to those specialists that need a practical tool to classify Pussillus or Obsoletus species in wildlife environments.
{"title":"MosCla app: An android app to classify Culicoides species","authors":"S. Gutiérrez, Noel Pérez, D. Benítez, S. Zapata, D. Augot","doi":"10.1109/ColCACI50549.2020.9247857","DOIUrl":"https://doi.org/10.1109/ColCACI50549.2020.9247857","url":null,"abstract":"Culicoides biting midges are transmission vectors of various diseases affecting humans and animals around the world. An optimal and fast classification method for these and other species have been a challenge and a necessity, especially in areas with limited resources and public health problems. In this work, we developed a mobile application to classify two Culicoides species using the morphological pattern analysis of their wings. The app implemented an automatic classification method based on the calculation of seven morphological features extracted from the wing images and a support vector machine classifier to produce the final classification of Pusillus or Obsoletus class. The proposed app was validated on an experimental dataset with 87 samples, reaching an outstanding mean of AUC score of 0.98 in the classification stage. Besides, we assessed the app feasibility using the mean of time and battery consumption metrics on two different emulators. The obtained scores of 12 and 7 s and 0.11 and 0.03 mAh for the phone and tablet emulators are satisfactory when developing mobile applications. Finally, reducing the feature space using an external wrapper method provided us a considerable improvement in the classification performance, AUC scores from 0.95 to 0.98, and decreasing the volume of information in training stages. Thus, these results enable the proposed app as an excellent approximation to those specialists that need a practical tool to classify Pussillus or Obsoletus species in wildlife environments.","PeriodicalId":446750,"journal":{"name":"2020 IEEE Colombian Conference on Applications of Computational Intelligence (IEEE ColCACI 2020)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123866912","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 : 2020-08-07DOI: 10.1109/ColCACI50549.2020.9247869
J. Archila, A. Orjuela-Cañón
Nowadays, the human machine interfaces have increased the applications for improving the quality of life in injured people. In spite of the progress in the field, new strategies are important to contribute to solve new problems. This proposal shows the employing of feature extraction in time and frequency domains. Three machine learning techniques as KNN, SVM and Random Forest were used to detect motor imagery from EEG signals. Comparison for feature extraction and the employed detection models were analyzed to find the best election in an application for close-open fist in hands. The results achieved more than 90% in accuracy for both approaches, showing as the frequency domain is preferable for feature extraction and the employment of the KNN classifier as best strategy for the present demand.
{"title":"Machine learning techniques for detecting motor imagery in upper limbs","authors":"J. Archila, A. Orjuela-Cañón","doi":"10.1109/ColCACI50549.2020.9247869","DOIUrl":"https://doi.org/10.1109/ColCACI50549.2020.9247869","url":null,"abstract":"Nowadays, the human machine interfaces have increased the applications for improving the quality of life in injured people. In spite of the progress in the field, new strategies are important to contribute to solve new problems. This proposal shows the employing of feature extraction in time and frequency domains. Three machine learning techniques as KNN, SVM and Random Forest were used to detect motor imagery from EEG signals. Comparison for feature extraction and the employed detection models were analyzed to find the best election in an application for close-open fist in hands. The results achieved more than 90% in accuracy for both approaches, showing as the frequency domain is preferable for feature extraction and the employment of the KNN classifier as best strategy for the present demand.","PeriodicalId":446750,"journal":{"name":"2020 IEEE Colombian Conference on Applications of Computational Intelligence (IEEE ColCACI 2020)","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126749085","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 : 2020-08-07DOI: 10.1109/ColCACI50549.2020.9248717
David Morales, Paula Arguello, M. Márquez, H. Arguello
This work introduces an imaging device that efficiently captures high-speed spectral videos along with a mathematical model that allows reconstructs them from far fewer measurements than those required by conventional scanning devices. This imaging architecture modulates and multiplexes the spectral-temporal information into a single compressed measurement by introducing a Dynamic Vision Sensor (SCAMP5) as a detector in a conventional compressive snapshot spectral image (CASSI) system. SCAMP5 sensor embeds processing and data storage capability into the pixels, which allows developed a high-speed temporal codification. The results of the numerical experiments through high-speed spectral videos shows reliable performance reconstructing spectral videos for a different amount of reconstructed frames. Comparing this proposal approach of snapshot spectral video with the conventional capture of spectral videos with multishot systems, our work arises very close results additionally our system outperforme the temporal spectral compression, more fully, the proposal approach captures a 8 times less samples obtaining a difference of 2.86 in SAM, 0.08 in SSIM, 2.9 in PSNR and 0.03 for RMSE. Therefore, the proposed architecture is an efficient and alternative high-speed spectral video acquisition system.
{"title":"Snapshot compressive spectral video via a monocular optical system","authors":"David Morales, Paula Arguello, M. Márquez, H. Arguello","doi":"10.1109/ColCACI50549.2020.9248717","DOIUrl":"https://doi.org/10.1109/ColCACI50549.2020.9248717","url":null,"abstract":"This work introduces an imaging device that efficiently captures high-speed spectral videos along with a mathematical model that allows reconstructs them from far fewer measurements than those required by conventional scanning devices. This imaging architecture modulates and multiplexes the spectral-temporal information into a single compressed measurement by introducing a Dynamic Vision Sensor (SCAMP5) as a detector in a conventional compressive snapshot spectral image (CASSI) system. SCAMP5 sensor embeds processing and data storage capability into the pixels, which allows developed a high-speed temporal codification. The results of the numerical experiments through high-speed spectral videos shows reliable performance reconstructing spectral videos for a different amount of reconstructed frames. Comparing this proposal approach of snapshot spectral video with the conventional capture of spectral videos with multishot systems, our work arises very close results additionally our system outperforme the temporal spectral compression, more fully, the proposal approach captures a 8 times less samples obtaining a difference of 2.86 in SAM, 0.08 in SSIM, 2.9 in PSNR and 0.03 for RMSE. Therefore, the proposed architecture is an efficient and alternative high-speed spectral video acquisition system.","PeriodicalId":446750,"journal":{"name":"2020 IEEE Colombian Conference on Applications of Computational Intelligence (IEEE ColCACI 2020)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117083756","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：