Pub Date : 2016-10-01DOI: 10.1109/CISP-BMEI.2016.7852775
Changyu Jiang, Bailing Zhang
Vehicle detection and vehicle type/make classification have been attracting more research in recent years. Previous methods for vehicle detection typically rely on large number of annotated training images by object bounding boxes, which is expensive and often subjective. In this paper, we propose a vehicle detection and recognition system by applying weakly-supervised convolutional neural network (CNN), with training relying only on image-level labels. Experiments were conducted on a datasets acquired from field-captured traffic surveillance cameras, with vehicle classification performance mAP 98.79% and accuracy 98.28%, and vehicle detection performance mAP 85.26%.
{"title":"Weakly-supervised vehicle detection and classification by convolutional neural network","authors":"Changyu Jiang, Bailing Zhang","doi":"10.1109/CISP-BMEI.2016.7852775","DOIUrl":"https://doi.org/10.1109/CISP-BMEI.2016.7852775","url":null,"abstract":"Vehicle detection and vehicle type/make classification have been attracting more research in recent years. Previous methods for vehicle detection typically rely on large number of annotated training images by object bounding boxes, which is expensive and often subjective. In this paper, we propose a vehicle detection and recognition system by applying weakly-supervised convolutional neural network (CNN), with training relying only on image-level labels. Experiments were conducted on a datasets acquired from field-captured traffic surveillance cameras, with vehicle classification performance mAP 98.79% and accuracy 98.28%, and vehicle detection performance mAP 85.26%.","PeriodicalId":275095,"journal":{"name":"2016 9th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132368428","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 : 2016-10-01DOI: 10.1109/CISP-BMEI.2016.7852916
Sangwoo Kang, W. Park
The MUltiple SIgnal Classification - MUSIC - algorithm and the Linear Sampling - LS - method are fast, stable, and effective non-iterative imaging techniques in inverse scattering problem. In fact, some previous studies indicated that the linear sampling method is an extended version of MUSIC. However, numerical results in support of this assertion have not been provided. In this contribution, we compare the imaging performance of MUSIC with that of the LS method with noisy data and underpin the above assertion.
多信号分类- MUSIC -算法和线性采样- LS -方法是快速、稳定、有效的反散射成像技术。事实上，之前的一些研究表明，线性抽样方法是MUSIC的扩展版本。但是，没有提供支持这一断言的数值结果。在这篇文章中，我们比较了MUSIC与LS方法在噪声数据下的成像性能，并支持了上述断言。
{"title":"Comparing the imaging performance of MUSIC and Linear Sampling method","authors":"Sangwoo Kang, W. Park","doi":"10.1109/CISP-BMEI.2016.7852916","DOIUrl":"https://doi.org/10.1109/CISP-BMEI.2016.7852916","url":null,"abstract":"The MUltiple SIgnal Classification - MUSIC - algorithm and the Linear Sampling - LS - method are fast, stable, and effective non-iterative imaging techniques in inverse scattering problem. In fact, some previous studies indicated that the linear sampling method is an extended version of MUSIC. However, numerical results in support of this assertion have not been provided. In this contribution, we compare the imaging performance of MUSIC with that of the LS method with noisy data and underpin the above assertion.","PeriodicalId":275095,"journal":{"name":"2016 9th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122396019","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 : 2016-10-01DOI: 10.1109/CISP-BMEI.2016.7852795
Dongming Li, Meijuan Wang, Jing Gao
There are many problems in the process of agricultural production, such as high labor costs, lack of processional management, delayed disaster warning, waste of agricultural information. In order to solve these problems, we developed the Argo-meteorological system. On the base of achieving real-time monitoring of farmland and disaster warning, it focused on the implementation of the comprehensive analysis and the storage of data on cloud platform which simplified the system structure and improved the efficiency of agricultural management. To help managers understand the exact situation like the growth of crops, pests and diseases, weather, and environment, sensors and binocular imaging array were used by the low-power sensing devices to obtain data. Then the data was converged to the data center on the cloud platform to be classified and processed. Meanwhile, warning feedback was given after analyzing the collected data and the standard indicators of agricultural production. The results of processing were pushed to monitoring system on PC and then showed in real-time. The test results showed that the system could achieve stable data transmission, efficient data processing and provide massive data for data mining. The cost of system maintenance and upgrade was reduced.
{"title":"Design and implementation of the agricultural meteorological system based on machine vision and cloud platform","authors":"Dongming Li, Meijuan Wang, Jing Gao","doi":"10.1109/CISP-BMEI.2016.7852795","DOIUrl":"https://doi.org/10.1109/CISP-BMEI.2016.7852795","url":null,"abstract":"There are many problems in the process of agricultural production, such as high labor costs, lack of processional management, delayed disaster warning, waste of agricultural information. In order to solve these problems, we developed the Argo-meteorological system. On the base of achieving real-time monitoring of farmland and disaster warning, it focused on the implementation of the comprehensive analysis and the storage of data on cloud platform which simplified the system structure and improved the efficiency of agricultural management. To help managers understand the exact situation like the growth of crops, pests and diseases, weather, and environment, sensors and binocular imaging array were used by the low-power sensing devices to obtain data. Then the data was converged to the data center on the cloud platform to be classified and processed. Meanwhile, warning feedback was given after analyzing the collected data and the standard indicators of agricultural production. The results of processing were pushed to monitoring system on PC and then showed in real-time. The test results showed that the system could achieve stable data transmission, efficient data processing and provide massive data for data mining. The cost of system maintenance and upgrade was reduced.","PeriodicalId":275095,"journal":{"name":"2016 9th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131174008","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 : 2016-10-01DOI: 10.1109/CISP-BMEI.2016.7852864
Wenjian Chen, Hui Sun, Yizhe Wang, Liang Lv, Tiansi An
A method of measuring the CW pulse signal phase difference is proposed, which is able to measure the phase difference between the difference frequency wave of parametric array and the small amplitude wave with the same frequency. The method is verified by simulation experiment. The factors influencing the measurement results ware discussed, such as the sampling rate, noise and sensor position. The numerical analysis results show that the proposed method can effectively measure the phase difference between the difference frequency wave of parametric array and the small amplitude wave with the same frequency.
{"title":"A method of measuring the phase difference between two pulse signals","authors":"Wenjian Chen, Hui Sun, Yizhe Wang, Liang Lv, Tiansi An","doi":"10.1109/CISP-BMEI.2016.7852864","DOIUrl":"https://doi.org/10.1109/CISP-BMEI.2016.7852864","url":null,"abstract":"A method of measuring the CW pulse signal phase difference is proposed, which is able to measure the phase difference between the difference frequency wave of parametric array and the small amplitude wave with the same frequency. The method is verified by simulation experiment. The factors influencing the measurement results ware discussed, such as the sampling rate, noise and sensor position. The numerical analysis results show that the proposed method can effectively measure the phase difference between the difference frequency wave of parametric array and the small amplitude wave with the same frequency.","PeriodicalId":275095,"journal":{"name":"2016 9th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131510241","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 : 2016-10-01DOI: 10.1109/CISP-BMEI.2016.7852752
Ping Yang, Li Chen, Jing Tian
The existing video jitter detection algorithms only calculate parameters such as displacement, rotation angle, zoom scale. The lack of extent calculation of jitter leads to wrong judgments while faced with complications. To tackle the problem, the longest path based algorithm for extent estimation of jitter was proposed in this paper. First, global motion estimation was conducted by gray projection algorithm introduced local consistency of jitter. Then, ratio, frequency and amplitude of jitter were calculated according to the longest path algorithm combined with global motion parameters. Finally, the extent estimation of jitter was realized via machine learning approach on the basis of jitter parameters obtained in the last step. Experimental results demonstrate the validity and feasibility of the proposed algorithm.
{"title":"Extent estimation of jitter based on the longest path for surveillance videos","authors":"Ping Yang, Li Chen, Jing Tian","doi":"10.1109/CISP-BMEI.2016.7852752","DOIUrl":"https://doi.org/10.1109/CISP-BMEI.2016.7852752","url":null,"abstract":"The existing video jitter detection algorithms only calculate parameters such as displacement, rotation angle, zoom scale. The lack of extent calculation of jitter leads to wrong judgments while faced with complications. To tackle the problem, the longest path based algorithm for extent estimation of jitter was proposed in this paper. First, global motion estimation was conducted by gray projection algorithm introduced local consistency of jitter. Then, ratio, frequency and amplitude of jitter were calculated according to the longest path algorithm combined with global motion parameters. Finally, the extent estimation of jitter was realized via machine learning approach on the basis of jitter parameters obtained in the last step. Experimental results demonstrate the validity and feasibility of the proposed algorithm.","PeriodicalId":275095,"journal":{"name":"2016 9th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"1376 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132540211","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 : 2016-10-01DOI: 10.1109/CISP-BMEI.2016.7852729
Tao Xie, Chunming Xia, Fei Chen, S. Zhang, Yue Zhang
Glossoscopy is an important part of Traditional Chinese Medicine (TCM). To analyze the tongue properties objectively, we need extract the tongue region from images. This paper presents a method to segment the tongue images based on kernel FCM (Fuzzy Cluster means). Firstly we pre-processed the tongue images by gray-level integral projection. Secondly the features were extracted to form a feature vector which contained texture, color, location and other information. Then, according to the feature vectors, pixels were clustering by kernel FCM whose parameters were decided by the proposed method. Finally, according to the pixels' neighbor connection theory, the tongue region was extracted. The results show that this method to segment tongue images is effective as its average accuracy reached upto 96.42%.
{"title":"A method of tongue image segmentation based on kernel FCM","authors":"Tao Xie, Chunming Xia, Fei Chen, S. Zhang, Yue Zhang","doi":"10.1109/CISP-BMEI.2016.7852729","DOIUrl":"https://doi.org/10.1109/CISP-BMEI.2016.7852729","url":null,"abstract":"Glossoscopy is an important part of Traditional Chinese Medicine (TCM). To analyze the tongue properties objectively, we need extract the tongue region from images. This paper presents a method to segment the tongue images based on kernel FCM (Fuzzy Cluster means). Firstly we pre-processed the tongue images by gray-level integral projection. Secondly the features were extracted to form a feature vector which contained texture, color, location and other information. Then, according to the feature vectors, pixels were clustering by kernel FCM whose parameters were decided by the proposed method. Finally, according to the pixels' neighbor connection theory, the tongue region was extracted. The results show that this method to segment tongue images is effective as its average accuracy reached upto 96.42%.","PeriodicalId":275095,"journal":{"name":"2016 9th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"72 3","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132782199","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 : 2016-10-01DOI: 10.1109/CISP-BMEI.2016.7852806
Mengjiao Dong, Lin Cao, Dong-Ming Zhang, Ru Guo
Microsoft Kinect is a motion sensing device that provides users friendly interfaces through natural postures and gestures. In this paper, based on Kinect v2, a system is proposed to control UAV flight, without remote control. In this system, more postures and gestures are defined, thus to easily control UAV flight, than the existing works. Furthermore, the false recognition on some postures are analyzed, meanwhile MPRA (many parameters restriction algorithm) is presented to improve the precision by imposing multiple restriction on posture recognition. The evaluation platform is built based on Kinect v2 and Ar. Drone2.0, and then the experiments are carried out, which show that the proposed system outperforms the existing works.
{"title":"UAV flight controlling based on Kinect for Windows v2","authors":"Mengjiao Dong, Lin Cao, Dong-Ming Zhang, Ru Guo","doi":"10.1109/CISP-BMEI.2016.7852806","DOIUrl":"https://doi.org/10.1109/CISP-BMEI.2016.7852806","url":null,"abstract":"Microsoft Kinect is a motion sensing device that provides users friendly interfaces through natural postures and gestures. In this paper, based on Kinect v2, a system is proposed to control UAV flight, without remote control. In this system, more postures and gestures are defined, thus to easily control UAV flight, than the existing works. Furthermore, the false recognition on some postures are analyzed, meanwhile MPRA (many parameters restriction algorithm) is presented to improve the precision by imposing multiple restriction on posture recognition. The evaluation platform is built based on Kinect v2 and Ar. Drone2.0, and then the experiments are carried out, which show that the proposed system outperforms the existing works.","PeriodicalId":275095,"journal":{"name":"2016 9th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133503249","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 : 2016-10-01DOI: 10.1109/CISP-BMEI.2016.7852783
Zhaorong Lin, Na Yao, Gaode Qin, Hongxia Cui
A quad-head combined aerial camera is first introduced in this paper. Due to its multiple center projection at each exposure station, an image post-processing procedure including camera calibration, matching, self-calibration stitching and color balancing is proposed to generate the stitched virtual image, in which the mathematical models of geometric calibration and self-calibration algorithms are primarily derived. The experiments show that the post-processing procedure has been capable of obtaining satisfying stitched virtual images. Besides, the related precisions of a 1:500 scale photogrammetry project has totally meet the requirements of national specifications of surveying and mapping.
{"title":"Image post-processing of a quad-head combined aerial camera and its surveying application","authors":"Zhaorong Lin, Na Yao, Gaode Qin, Hongxia Cui","doi":"10.1109/CISP-BMEI.2016.7852783","DOIUrl":"https://doi.org/10.1109/CISP-BMEI.2016.7852783","url":null,"abstract":"A quad-head combined aerial camera is first introduced in this paper. Due to its multiple center projection at each exposure station, an image post-processing procedure including camera calibration, matching, self-calibration stitching and color balancing is proposed to generate the stitched virtual image, in which the mathematical models of geometric calibration and self-calibration algorithms are primarily derived. The experiments show that the post-processing procedure has been capable of obtaining satisfying stitched virtual images. Besides, the related precisions of a 1:500 scale photogrammetry project has totally meet the requirements of national specifications of surveying and mapping.","PeriodicalId":275095,"journal":{"name":"2016 9th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132346431","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 : 2016-10-01DOI: 10.1109/CISP-BMEI.2016.7852736
Fengquan Zhang, Yujie Zhao, Zhaowei Wang, T. Lei
Consistent illumination is an important research target in the augmented reality system. In this paper, an efficient illumination method is implemented to solve the consistent illumination problem for mixed reality application. We design and implement an all-frequency environment rendering methods based on wavelet transform, and proposed an accelerated rendering method on GPU. The environment rendering method consists of two steps, one is precomputation and the other real-time rendering. An improved shadow algorithm based on shadow map is propose, which can solve least area where light source view frustum should surround according to the relationship of view frustum in eye space and terrain. In the end, we develop a mixed reality system for aircraft design based on the research results of all the work above which realize the illumination and shadow effects on the virtual aircraft. It can be used to assess how the weather impact on the operating.
{"title":"An efficient all-frequency environment rendering method for mixed reality","authors":"Fengquan Zhang, Yujie Zhao, Zhaowei Wang, T. Lei","doi":"10.1109/CISP-BMEI.2016.7852736","DOIUrl":"https://doi.org/10.1109/CISP-BMEI.2016.7852736","url":null,"abstract":"Consistent illumination is an important research target in the augmented reality system. In this paper, an efficient illumination method is implemented to solve the consistent illumination problem for mixed reality application. We design and implement an all-frequency environment rendering methods based on wavelet transform, and proposed an accelerated rendering method on GPU. The environment rendering method consists of two steps, one is precomputation and the other real-time rendering. An improved shadow algorithm based on shadow map is propose, which can solve least area where light source view frustum should surround according to the relationship of view frustum in eye space and terrain. In the end, we develop a mixed reality system for aircraft design based on the research results of all the work above which realize the illumination and shadow effects on the virtual aircraft. It can be used to assess how the weather impact on the operating.","PeriodicalId":275095,"journal":{"name":"2016 9th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115179465","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 : 2016-10-01DOI: 10.1109/CISP-BMEI.2016.7852687
Zhen Gao, Guoliang Lu, Peng Yan
Sequence matching/alignment based scheme has been common for action recognition. Such a typical scheme, however, requires tremendous amounts of computation when the volume of prototypical action videos is large and easily causes mismatching for border-isolated samples in action categories. In this paper, we propose a framework of averaging video sequences based on multi-dimensional dynamic time warping (MD-DTW) and propose to use the resulting average actions, instead of prototypical action videos, for action recognition. Experimental results show that 1) average actions were shown to be more discriminative than prototypical video sequences for action modeling, and 2) action recognition using average actions rather than using prototypical action videos is much more efficient and advanced.
{"title":"Averaging video sequences to improve action recognition","authors":"Zhen Gao, Guoliang Lu, Peng Yan","doi":"10.1109/CISP-BMEI.2016.7852687","DOIUrl":"https://doi.org/10.1109/CISP-BMEI.2016.7852687","url":null,"abstract":"Sequence matching/alignment based scheme has been common for action recognition. Such a typical scheme, however, requires tremendous amounts of computation when the volume of prototypical action videos is large and easily causes mismatching for border-isolated samples in action categories. In this paper, we propose a framework of averaging video sequences based on multi-dimensional dynamic time warping (MD-DTW) and propose to use the resulting average actions, instead of prototypical action videos, for action recognition. Experimental results show that 1) average actions were shown to be more discriminative than prototypical video sequences for action modeling, and 2) action recognition using average actions rather than using prototypical action videos is much more efficient and advanced.","PeriodicalId":275095,"journal":{"name":"2016 9th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114474469","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：