Pub Date : 2018-11-01DOI: 10.1109/ICSAI.2018.8599382
Ge Yang, Siping Chen
In dense scenes, a large number of individuals can cause more serious problems such as blurred vision, chaotic scenes, complex behaviors and so on. For low density pedestrian detection algorithm, the accuracy of detection will be greatly reduced, even detection failure when facing these problems in high density scenes. In view of the above problems, the detection algorithm based on human head shoulder model is proposed. Support vector machine is used to train the classifier by machine learning. The detection algorithm proposed in this paper achieves 94% detection by using MIT and INRIA data sets. (Abstract)
{"title":"Pedestrian Detection Under Dense Crowd","authors":"Ge Yang, Siping Chen","doi":"10.1109/ICSAI.2018.8599382","DOIUrl":"https://doi.org/10.1109/ICSAI.2018.8599382","url":null,"abstract":"In dense scenes, a large number of individuals can cause more serious problems such as blurred vision, chaotic scenes, complex behaviors and so on. For low density pedestrian detection algorithm, the accuracy of detection will be greatly reduced, even detection failure when facing these problems in high density scenes. In view of the above problems, the detection algorithm based on human head shoulder model is proposed. Support vector machine is used to train the classifier by machine learning. The detection algorithm proposed in this paper achieves 94% detection by using MIT and INRIA data sets. (Abstract)","PeriodicalId":375852,"journal":{"name":"2018 5th International Conference on Systems and Informatics (ICSAI)","volume":"51 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126019294","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 : 2018-11-01DOI: 10.1109/ICSAI.2018.8599452
Zhanli Li, Junchao Wang, Jiaying Chen
In recent years, wide range of scene surveillance technology, widely used in the field of security monitoring, has become one of the important measure of security monitoring. The paper gives a method, in a single region, first use a low computational complexity of Kalman filter to replace the large computing TLD tracking module to obtain targets trajectories. Thus, in the non-overlapping regions, a Gaussian and mean cross-correlation function method is proposed to estimate the topological nodes between the cameras, which provides a stable camera correlation relationship for the continuous tracking in large scale scenes. The results show that the target tracking under the single region has good effect, and the topological relationship estimation between the cameras also has better anti-interference in the multi-region views, and feasible.
{"title":"Estimating Path in camera network with non-overlapping FOVs","authors":"Zhanli Li, Junchao Wang, Jiaying Chen","doi":"10.1109/ICSAI.2018.8599452","DOIUrl":"https://doi.org/10.1109/ICSAI.2018.8599452","url":null,"abstract":"In recent years, wide range of scene surveillance technology, widely used in the field of security monitoring, has become one of the important measure of security monitoring. The paper gives a method, in a single region, first use a low computational complexity of Kalman filter to replace the large computing TLD tracking module to obtain targets trajectories. Thus, in the non-overlapping regions, a Gaussian and mean cross-correlation function method is proposed to estimate the topological nodes between the cameras, which provides a stable camera correlation relationship for the continuous tracking in large scale scenes. The results show that the target tracking under the single region has good effect, and the topological relationship estimation between the cameras also has better anti-interference in the multi-region views, and feasible.","PeriodicalId":375852,"journal":{"name":"2018 5th International Conference on Systems and Informatics (ICSAI)","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126063656","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 : 2018-11-01DOI: 10.1109/ICSAI.2018.8599414
Yinan Xing, Jianbin Liu
A guided filtering method based on Shen-Castan operator is proposed to solve the problem of a large amount of noise attached to the ultrasound image. In this method, the edge detection result of the Shen-Castan operator was used to reduce the edge halo phenomenon of the guided filter. Apply the Shen-Castan operator to perform edge detection on the noisy image; the edge detection result will be returned to the original image to obtain an edge enhanced guide image; this image will be used as a guided image of the noisy image for guided filtering. Through the simulation experiment, the guided filtering method based on the Shen-Castan operator can not only maintain the original function of smoothing noise, but also improve the signal-to-noise ratio and structural similarity of the passenger peak.
{"title":"A Guided Filtering Method Based on Shen-Castan Operator","authors":"Yinan Xing, Jianbin Liu","doi":"10.1109/ICSAI.2018.8599414","DOIUrl":"https://doi.org/10.1109/ICSAI.2018.8599414","url":null,"abstract":"A guided filtering method based on Shen-Castan operator is proposed to solve the problem of a large amount of noise attached to the ultrasound image. In this method, the edge detection result of the Shen-Castan operator was used to reduce the edge halo phenomenon of the guided filter. Apply the Shen-Castan operator to perform edge detection on the noisy image; the edge detection result will be returned to the original image to obtain an edge enhanced guide image; this image will be used as a guided image of the noisy image for guided filtering. Through the simulation experiment, the guided filtering method based on the Shen-Castan operator can not only maintain the original function of smoothing noise, but also improve the signal-to-noise ratio and structural similarity of the passenger peak.","PeriodicalId":375852,"journal":{"name":"2018 5th International Conference on Systems and Informatics (ICSAI)","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133617342","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 : 2018-11-01DOI: 10.1109/ICSAI.2018.8599425
Pan Li, Rui Zhang, Jing Zhang, Jie Li, G. Zhao, Hua Li
Radar signal sorting in complex electromagnetic environment is the key technology of radar reconnaissance interference, which plays an important role in the measurement, analysis and identification of subsequent radar characteristic parameters. Aiming at the fact that the field experiment can not simulate the complex electromagnetic environment and the huge cost of the real battlefield, a radar electromagnetic environment simulation system based on ADC/DAC+FPGA+ARM architecture is proposed, which realizes the embedded electromagnetic environment The simulation and design of the simulation system are carried out, and the RF noise module, the random pulse module, the linear sweep module, the pulse delay superposition interference module and the single batch false target module of the electromagnetic environment simulation system are simulated.
{"title":"Design of Radar Electromagnetic Environment Simulation System Based on Altera Stratix® III Series FPGA","authors":"Pan Li, Rui Zhang, Jing Zhang, Jie Li, G. Zhao, Hua Li","doi":"10.1109/ICSAI.2018.8599425","DOIUrl":"https://doi.org/10.1109/ICSAI.2018.8599425","url":null,"abstract":"Radar signal sorting in complex electromagnetic environment is the key technology of radar reconnaissance interference, which plays an important role in the measurement, analysis and identification of subsequent radar characteristic parameters. Aiming at the fact that the field experiment can not simulate the complex electromagnetic environment and the huge cost of the real battlefield, a radar electromagnetic environment simulation system based on ADC/DAC+FPGA+ARM architecture is proposed, which realizes the embedded electromagnetic environment The simulation and design of the simulation system are carried out, and the RF noise module, the random pulse module, the linear sweep module, the pulse delay superposition interference module and the single batch false target module of the electromagnetic environment simulation system are simulated.","PeriodicalId":375852,"journal":{"name":"2018 5th International Conference on Systems and Informatics (ICSAI)","volume":"213 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133966078","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 : 2018-11-01DOI: 10.1109/ICSAI.2018.8599440
Feng-Zeng Liu, B. Xiao, Hongbin Jin, Qizeng Zhang
For the problem that closeness is difficult to effectively distinguish the importance of some nodes in complex networks, a new method of node importance measurement is proposed, which fuse the degree and closeness based on node re-ranking in segmentation. According to the network propagation dynamics model and Kendalls Tau coefficient, accuracy indicator and ranking stability indicator for evaluating measurement methods are given. Using the proposed method, simulations are carried out on Barabasi-Albert(BA) scale-free networks and ER random networks with different structures. The results show that compared with degree and closeness, fusion method not only has better measurement accuracy, but also has higher ranking stability.
{"title":"A Fusion Method for Node Importance Measurement in Complex Networks","authors":"Feng-Zeng Liu, B. Xiao, Hongbin Jin, Qizeng Zhang","doi":"10.1109/ICSAI.2018.8599440","DOIUrl":"https://doi.org/10.1109/ICSAI.2018.8599440","url":null,"abstract":"For the problem that closeness is difficult to effectively distinguish the importance of some nodes in complex networks, a new method of node importance measurement is proposed, which fuse the degree and closeness based on node re-ranking in segmentation. According to the network propagation dynamics model and Kendalls Tau coefficient, accuracy indicator and ranking stability indicator for evaluating measurement methods are given. Using the proposed method, simulations are carried out on Barabasi-Albert(BA) scale-free networks and ER random networks with different structures. The results show that compared with degree and closeness, fusion method not only has better measurement accuracy, but also has higher ranking stability.","PeriodicalId":375852,"journal":{"name":"2018 5th International Conference on Systems and Informatics (ICSAI)","volume":"172 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131830759","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 : 2018-11-01DOI: 10.1109/ICSAI.2018.8599336
Yu Jin, Zhongqin Bi
Load curve clustering is a basic task for big data mining in electricity consumption. This paper proposed a clustering algorithm to improve the correct and accurate clustering of the load curve data. Firstly, we introduced the FastDTW as the similarity metric to measure the distance between two time series. Secondly, we used the Affinity Propagation (AP) to cluster. At last, we proposed a novel FastDTW-AP clustering algorithm for load curve clustering. As the similarity measures for clustering, we consider the Euclidean distance, Dynamic Time Warping (DTW), and Fast Dynamic Time Warping (FastDTW), and compare the efficiency of three similarity measures using the labelled dataset SCCTS from UCI. To evaluate the clustering algorithm, the real power load data is analyzed. The results show obvious improvement in evaluation index Adjust Rand Index (ARI) and Adjust Mutual Information (AMI).
{"title":"Power Load Curve Clustering Algorithm Using Fast Dynamic Time Warping and Affinity Propagation","authors":"Yu Jin, Zhongqin Bi","doi":"10.1109/ICSAI.2018.8599336","DOIUrl":"https://doi.org/10.1109/ICSAI.2018.8599336","url":null,"abstract":"Load curve clustering is a basic task for big data mining in electricity consumption. This paper proposed a clustering algorithm to improve the correct and accurate clustering of the load curve data. Firstly, we introduced the FastDTW as the similarity metric to measure the distance between two time series. Secondly, we used the Affinity Propagation (AP) to cluster. At last, we proposed a novel FastDTW-AP clustering algorithm for load curve clustering. As the similarity measures for clustering, we consider the Euclidean distance, Dynamic Time Warping (DTW), and Fast Dynamic Time Warping (FastDTW), and compare the efficiency of three similarity measures using the labelled dataset SCCTS from UCI. To evaluate the clustering algorithm, the real power load data is analyzed. The results show obvious improvement in evaluation index Adjust Rand Index (ARI) and Adjust Mutual Information (AMI).","PeriodicalId":375852,"journal":{"name":"2018 5th International Conference on Systems and Informatics (ICSAI)","volume":"80 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134358360","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 : 2018-11-01DOI: 10.1109/ICSAI.2018.8599420
Zhongqiu Liu, Jianchao Li, Y. Shu, Dongping Zhang
In this paper, a convolutional neural network model based on YOLO9000 is introduced to meet the need of real-time engineering computing. This network model can study and classify the targets in depth, aiming at the characteristics of scissors and aerosols. The characteristics have various kinds such as overlap, cover and multiscale. At the present stage, the average speed is 68 FPS on the windows platform with GPU (Geforce GTX Titan X) acceleration. In addition, the average precision and recall rate are 94. 5%, 92. 6%, respectively.
{"title":"Detection and Recognition of Security Detection Object Based on Yolo9000","authors":"Zhongqiu Liu, Jianchao Li, Y. Shu, Dongping Zhang","doi":"10.1109/ICSAI.2018.8599420","DOIUrl":"https://doi.org/10.1109/ICSAI.2018.8599420","url":null,"abstract":"In this paper, a convolutional neural network model based on YOLO9000 is introduced to meet the need of real-time engineering computing. This network model can study and classify the targets in depth, aiming at the characteristics of scissors and aerosols. The characteristics have various kinds such as overlap, cover and multiscale. At the present stage, the average speed is 68 FPS on the windows platform with GPU (Geforce GTX Titan X) acceleration. In addition, the average precision and recall rate are 94. 5%, 92. 6%, respectively.","PeriodicalId":375852,"journal":{"name":"2018 5th International Conference on Systems and Informatics (ICSAI)","volume":"306 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134193511","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 : 2018-11-01DOI: 10.1109/ICSAI.2018.8599368
Shimian Zhang, Dexin Yang
Generative adversarial networks (GANs) have shown great performance on image-to-image translation tasks. Many approaches have been proposed for translation of human face images, scene pictures and artful paintings, but few works considered about translating a pet image. In this paper, we propose a method based on cycle-consistent adversarial network (CycleGAN) to solve pet hair color transfer problem. Given a pet image, our model can translate its hair color into a desired one while keeping its other features unchanged, which makes our generated images seem quite realistic. We do several improvements on CycleGAN including doing segmentation to avoid the influence of background, and using spectral normalization to improve the quality of generated images. We build a large pet image dataset consisting of a total number of 7. 5K images, categorized by different hair colors. Our proposed method is trained and tested on this data set and the results show the promising performance on translating between white and orange hair color of dog images.
{"title":"Pet Hair Color Transfer Based On CycleGAN","authors":"Shimian Zhang, Dexin Yang","doi":"10.1109/ICSAI.2018.8599368","DOIUrl":"https://doi.org/10.1109/ICSAI.2018.8599368","url":null,"abstract":"Generative adversarial networks (GANs) have shown great performance on image-to-image translation tasks. Many approaches have been proposed for translation of human face images, scene pictures and artful paintings, but few works considered about translating a pet image. In this paper, we propose a method based on cycle-consistent adversarial network (CycleGAN) to solve pet hair color transfer problem. Given a pet image, our model can translate its hair color into a desired one while keeping its other features unchanged, which makes our generated images seem quite realistic. We do several improvements on CycleGAN including doing segmentation to avoid the influence of background, and using spectral normalization to improve the quality of generated images. We build a large pet image dataset consisting of a total number of 7. 5K images, categorized by different hair colors. Our proposed method is trained and tested on this data set and the results show the promising performance on translating between white and orange hair color of dog images.","PeriodicalId":375852,"journal":{"name":"2018 5th International Conference on Systems and Informatics (ICSAI)","volume":"230 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114473233","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}
In this paper, variants of Getz-Marsden dynamic system (GMDS) and Zhang neural network (ZNN), i.e., GMDS-ZNN variants, are proposed and discretized by different discretization formulas, i.e., discretized by Euler forward formula, Taylor-Zhang discretization formula and ZD5i (Zhang discretization with 5 instants) formula. In order to investigate the proposed GMDS-ZNN variants, we conduct numerical experiments, As comparisons, conventional dynamic systems GMDSI and GMDS2 (which are proved to have higher precision) are presented. Numerical results show that these discrete GMDS-ZNN variants have fixed error pattern when computing time-dependent complex matrix inverse. The error pattern is confirmed as being proportional to sampling gap.
{"title":"GMDS-ZNN Variants Having Errors Proportional to Sampling Gap as Compared with Models 1 and 2 Having Higher Precision","authors":"Jian Li, Guofu Wu, Chuming Li, Mengling Xiao, Yunong Zhang","doi":"10.1109/ICSAI.2018.8599354","DOIUrl":"https://doi.org/10.1109/ICSAI.2018.8599354","url":null,"abstract":"In this paper, variants of Getz-Marsden dynamic system (GMDS) and Zhang neural network (ZNN), i.e., GMDS-ZNN variants, are proposed and discretized by different discretization formulas, i.e., discretized by Euler forward formula, Taylor-Zhang discretization formula and ZD5i (Zhang discretization with 5 instants) formula. In order to investigate the proposed GMDS-ZNN variants, we conduct numerical experiments, As comparisons, conventional dynamic systems GMDSI and GMDS2 (which are proved to have higher precision) are presented. Numerical results show that these discrete GMDS-ZNN variants have fixed error pattern when computing time-dependent complex matrix inverse. The error pattern is confirmed as being proportional to sampling gap.","PeriodicalId":375852,"journal":{"name":"2018 5th International Conference on Systems and Informatics (ICSAI)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131729492","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 : 2018-11-01DOI: 10.1109/ICSAI.2018.8599451
Yunong Zhang, Jinjin Guo, Liu He, Yang Shi, Chaowei Hu
In recent years, Zhang et al. discretization (ZeaD) as a new class of time-discretization methods has been proposed, named and applied by Zhang et al. Note that ZeaD formulas can accurately discretize Zhang neural networks $(mathrm {i}.mathrm {e}.$, ZNN, or say, Zhang dynamics) models as well as ordinary differential equation systems. In previous work, various ZeaD formulas have been presented and unified, including Euler forward formula as 2-instant ZeaD formula that is convergent with a truncation error being proportional to the first power of sampling period and Taylor-type discretization formula as 4-instant ZeaD formula that is convergent with a truncation error being proportional to the second power of sampling period. During our pursuit of ZeaD formulas that are convergent with a higher precision, we discover that there exists no 6-instant ZeaD formula that is convergent with a quartic (ie, biquadratic, of degree 4) or higher precision. The truncation error of any 6-instant ZeaD formula is proportional to the third power of sampling period or bigger. The contributions are theoretically proved in this paper as well.
{"title":"Any ZeaD Formula of Six Instants Having No Quartic or Higher Precision with Proof","authors":"Yunong Zhang, Jinjin Guo, Liu He, Yang Shi, Chaowei Hu","doi":"10.1109/ICSAI.2018.8599451","DOIUrl":"https://doi.org/10.1109/ICSAI.2018.8599451","url":null,"abstract":"In recent years, Zhang et al. discretization (ZeaD) as a new class of time-discretization methods has been proposed, named and applied by Zhang et al. Note that ZeaD formulas can accurately discretize Zhang neural networks $(mathrm {i}.mathrm {e}.$, ZNN, or say, Zhang dynamics) models as well as ordinary differential equation systems. In previous work, various ZeaD formulas have been presented and unified, including Euler forward formula as 2-instant ZeaD formula that is convergent with a truncation error being proportional to the first power of sampling period and Taylor-type discretization formula as 4-instant ZeaD formula that is convergent with a truncation error being proportional to the second power of sampling period. During our pursuit of ZeaD formulas that are convergent with a higher precision, we discover that there exists no 6-instant ZeaD formula that is convergent with a quartic (ie, biquadratic, of degree 4) or higher precision. The truncation error of any 6-instant ZeaD formula is proportional to the third power of sampling period or bigger. The contributions are theoretically proved in this paper as well.","PeriodicalId":375852,"journal":{"name":"2018 5th International Conference on Systems and Informatics (ICSAI)","volume":"69 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134253018","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：