Pub Date : 2003-05-12DOI: 10.1109/CAMP.2003.1598154
L. Yang, L. Welch, J. Liu, C. Cavanaugh
Dynamic, distributed, real-time control systems must control changing environments in a timely manner despite the fact that the system's load and timing vary in a way that is not characterizable by time-invariant statistical distributions. A quality of service (QoS) manager has been implemented that forecasts timing constraint violations in such systems and corrects them before they occur. The majority of forecasting techniques rely on moving averaging to extrapolate the future values, therefore the existence of outliers frequently impose disastrous effects on the accuracy of prediction. Most existing forecasting methods in literature use thresholding steps to empirically eliminate outliers, whose success heavily depends on the prior knowledge in choosing the initial fit and threshold values. In this paper, we propose a robust algorithm to automatically reject outliers and thus achieve accurate forecasting of host load and path latency. Our algorithm involves minimizing the integral of the squared error (ISE or L2E) between a Gaussian model of the residual and its true density function. The residual here refers to the difference between the path latencies and the trend line. We present the implementation results using L2E as well as other two widely used forecasting methods: least-squares linear regression and Box-Jenkins AR(2) forecasting, with DynBench dynamic, distributed real-time benchmark being employed as the testbed. We experimentally show that our L2 E-based scheme yields higher forecasting accuracy over the other two approaches
{"title":"A robust QoS forecasting technique for a dynamic, distributed real-time testbed","authors":"L. Yang, L. Welch, J. Liu, C. Cavanaugh","doi":"10.1109/CAMP.2003.1598154","DOIUrl":"https://doi.org/10.1109/CAMP.2003.1598154","url":null,"abstract":"Dynamic, distributed, real-time control systems must control changing environments in a timely manner despite the fact that the system's load and timing vary in a way that is not characterizable by time-invariant statistical distributions. A quality of service (QoS) manager has been implemented that forecasts timing constraint violations in such systems and corrects them before they occur. The majority of forecasting techniques rely on moving averaging to extrapolate the future values, therefore the existence of outliers frequently impose disastrous effects on the accuracy of prediction. Most existing forecasting methods in literature use thresholding steps to empirically eliminate outliers, whose success heavily depends on the prior knowledge in choosing the initial fit and threshold values. In this paper, we propose a robust algorithm to automatically reject outliers and thus achieve accurate forecasting of host load and path latency. Our algorithm involves minimizing the integral of the squared error (ISE or L2E) between a Gaussian model of the residual and its true density function. The residual here refers to the difference between the path latencies and the trend line. We present the implementation results using L2E as well as other two widely used forecasting methods: least-squares linear regression and Box-Jenkins AR(2) forecasting, with DynBench dynamic, distributed real-time benchmark being employed as the testbed. We experimentally show that our L2 E-based scheme yields higher forecasting accuracy over the other two approaches","PeriodicalId":443821,"journal":{"name":"2003 IEEE International Workshop on Computer Architectures for Machine Perception","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133444223","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 : 2003-05-12DOI: 10.1109/CAMP.2003.1598162
M. Albanesi, M. Ferretti, A. Dell'Olio, M. De Ponti
In this paper we report on the improvements obtained by using a VLIW approach to handle an image processing application for embedded printing systems. This field of research is gaining significance especially today: the goal in business applications is to achieve high speed in computations while maintaining low cost and low power consumption. For this purpose the VLIW approach, which was originally developed for massive mathematics applications, seems to be a good choice, especially for its flexibility and low cost. The goal of this paper is to demonstrate the effectiveness of VLIW, especially in data parallel computation intensive applications, by studying the behaviour of a print pipeline, which is a collection of four algorithms with different characteristics. For each algorithm we analyse its structure and the implicit and explicit code parallelism, in order to give a complete overview of the control flow involved, then we proceed to perform adjustments on source code to help the compiler exploit the maximum performance
{"title":"Effectiveness of a VLIW architecture in a data parallel image application","authors":"M. Albanesi, M. Ferretti, A. Dell'Olio, M. De Ponti","doi":"10.1109/CAMP.2003.1598162","DOIUrl":"https://doi.org/10.1109/CAMP.2003.1598162","url":null,"abstract":"In this paper we report on the improvements obtained by using a VLIW approach to handle an image processing application for embedded printing systems. This field of research is gaining significance especially today: the goal in business applications is to achieve high speed in computations while maintaining low cost and low power consumption. For this purpose the VLIW approach, which was originally developed for massive mathematics applications, seems to be a good choice, especially for its flexibility and low cost. The goal of this paper is to demonstrate the effectiveness of VLIW, especially in data parallel computation intensive applications, by studying the behaviour of a print pipeline, which is a collection of four algorithms with different characteristics. For each algorithm we analyse its structure and the implicit and explicit code parallelism, in order to give a complete overview of the control flow involved, then we proceed to perform adjustments on source code to help the compiler exploit the maximum performance","PeriodicalId":443821,"journal":{"name":"2003 IEEE International Workshop on Computer Architectures for Machine Perception","volume":"37 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132213062","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 : 2003-05-12DOI: 10.1109/CAMP.2003.1598173
T. Darwish, G. Seetharaman, M. Bayoumi
Given a video sequence characterized by correlations in space and time, the objective of motion estimation is to find and accurately represent such correlations. Block matching algorithms are mainly used as the core of the motion estimation unit. Search algorithms include FSBMA, TSSMA, logarithmic etc.. At least 3 isointensity centroids in two frames are required to compute 2-D affine motion. Access to up to 8 isointensity centroids allow for two independent affine motions. Virtual displacement method is used to form linear constraints
{"title":"A fast isointensity maps based approach to optimal motion field computation for MPEG coding","authors":"T. Darwish, G. Seetharaman, M. Bayoumi","doi":"10.1109/CAMP.2003.1598173","DOIUrl":"https://doi.org/10.1109/CAMP.2003.1598173","url":null,"abstract":"Given a video sequence characterized by correlations in space and time, the objective of motion estimation is to find and accurately represent such correlations. Block matching algorithms are mainly used as the core of the motion estimation unit. Search algorithms include FSBMA, TSSMA, logarithmic etc.. At least 3 isointensity centroids in two frames are required to compute 2-D affine motion. Access to up to 8 isointensity centroids allow for two independent affine motions. Virtual displacement method is used to form linear constraints","PeriodicalId":443821,"journal":{"name":"2003 IEEE International Workshop on Computer Architectures for Machine Perception","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127810991","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 : 2003-05-12DOI: 10.1109/CAMP.2003.1598171
J. Hernandez, B. Romero
Gabor method of multiscale image representation is very useful in many computer vision applications and to simulate the human being visual system. However, this transform is characterised by repetitive and iterative operations on large amounts of data, which imply a high computational cost. In this point is when the hardware/software codesign technique makes sense. In this work we have implemented the Gabor transform using a reconfigurable FPGA following a pipeline scheme in order to speed up the process
{"title":"Multiscale image representation based on Gabor transform using reconfigurable FPGA","authors":"J. Hernandez, B. Romero","doi":"10.1109/CAMP.2003.1598171","DOIUrl":"https://doi.org/10.1109/CAMP.2003.1598171","url":null,"abstract":"Gabor method of multiscale image representation is very useful in many computer vision applications and to simulate the human being visual system. However, this transform is characterised by repetitive and iterative operations on large amounts of data, which imply a high computational cost. In this point is when the hardware/software codesign technique makes sense. In this work we have implemented the Gabor transform using a reconfigurable FPGA following a pipeline scheme in order to speed up the process","PeriodicalId":443821,"journal":{"name":"2003 IEEE International Workshop on Computer Architectures for Machine Perception","volume":"58 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122997562","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 : 2003-05-12DOI: 10.1109/CAMP.2003.1598147
A. Bevilacqua
Motion detection systems for visual surveillance and monitoring purposes have aroused interest in the computer video community for many years. The main task of these applications is to identify (and track) moving targets. Usually, these applications requires that a large number of parameters is tuned in order to work properly. In the traffic monitoring application we have developed about thirty parameters concerning the detection algorithm have been considered as to be optimized. Genetic algorithms (GAs) are an optimization technique which involves a search from a population of solutions rather than from a single point. Although they usually are very time-consuming, they owe a high intrinsic parallelism. Accordingly, this paper shows how a distributed implementation of a GA over a network of workstations can successfully accomplish the parameter optimization task within a motion detection system and achieve excellent performance within a reduced amount of time
{"title":"Calibrating a motion detection system by means of a distributed genetic algorithm","authors":"A. Bevilacqua","doi":"10.1109/CAMP.2003.1598147","DOIUrl":"https://doi.org/10.1109/CAMP.2003.1598147","url":null,"abstract":"Motion detection systems for visual surveillance and monitoring purposes have aroused interest in the computer video community for many years. The main task of these applications is to identify (and track) moving targets. Usually, these applications requires that a large number of parameters is tuned in order to work properly. In the traffic monitoring application we have developed about thirty parameters concerning the detection algorithm have been considered as to be optimized. Genetic algorithms (GAs) are an optimization technique which involves a search from a population of solutions rather than from a single point. Although they usually are very time-consuming, they owe a high intrinsic parallelism. Accordingly, this paper shows how a distributed implementation of a GA over a network of workstations can successfully accomplish the parameter optimization task within a motion detection system and achieve excellent performance within a reduced amount of time","PeriodicalId":443821,"journal":{"name":"2003 IEEE International Workshop on Computer Architectures for Machine Perception","volume":"70 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131780523","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 : 2003-05-12DOI: 10.1109/CAMP.2003.1598148
I. Ulusoy, U. Halici, K. Leblebicioğlu
In this paper we describe an algorithm for object recognition and cognitive map formation using stereo image data in a 3D virtual world where 3D objects and a robot with stereo imaging system are simulated. Stereo imaging system is simulated so that the actual human visual system properties such as focusing, accommodation, field of view are parameterized. Only the stereo images obtained from this world are supplied to the virtual robot (agent). By applying our disparity algorithm on stereo image pairs, depth map for the current view is obtained. Using the depth information for the current view, a cognitive map of the environment is updated gradually while the virtual agent is exploring the environment. The agent explores its environment in an intelligent way using the current view and environmental map information obtained up to date. Also, during exploration if a new object is observed, from its view from different directions, it is labeled with its shape such as sphere, cylinder, cone
{"title":"Object recognition and cognitive map formation using active stereo vision in a virtual world","authors":"I. Ulusoy, U. Halici, K. Leblebicioğlu","doi":"10.1109/CAMP.2003.1598148","DOIUrl":"https://doi.org/10.1109/CAMP.2003.1598148","url":null,"abstract":"In this paper we describe an algorithm for object recognition and cognitive map formation using stereo image data in a 3D virtual world where 3D objects and a robot with stereo imaging system are simulated. Stereo imaging system is simulated so that the actual human visual system properties such as focusing, accommodation, field of view are parameterized. Only the stereo images obtained from this world are supplied to the virtual robot (agent). By applying our disparity algorithm on stereo image pairs, depth map for the current view is obtained. Using the depth information for the current view, a cognitive map of the environment is updated gradually while the virtual agent is exploring the environment. The agent explores its environment in an intelligent way using the current view and environmental map information obtained up to date. Also, during exploration if a new object is observed, from its view from different directions, it is labeled with its shape such as sphere, cylinder, cone","PeriodicalId":443821,"journal":{"name":"2003 IEEE International Workshop on Computer Architectures for Machine Perception","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131305706","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 : 2003-05-12DOI: 10.1109/CAMP.2003.1598166
S. Maya-Rueda, C. Torres-Huitzil, M. Arias-Estrada
Motion estimation of a scene is an interesting problem in computer vision since it is the basis for the dynamic analysis of a scene. However this task is computational intensive for conventional processors. In this work, a FPGA-based hardware architecture for real-time motion estimation is proposed. The technique used for motion estimation is a variation of the optical flow algorithm where the problem is reformulated as a sum of overlapped basis functions, and solved as a linear system. The proposed architecture is based on a systolic approach and is composed of parallel modules organized in a regular structure. The systolic processor accelerates the matrix operations required to achieve real-time performance. The architecture design is presented. Preliminary results are shown and discussed
{"title":"A real-time FPGA-based architecture for optical flow computation","authors":"S. Maya-Rueda, C. Torres-Huitzil, M. Arias-Estrada","doi":"10.1109/CAMP.2003.1598166","DOIUrl":"https://doi.org/10.1109/CAMP.2003.1598166","url":null,"abstract":"Motion estimation of a scene is an interesting problem in computer vision since it is the basis for the dynamic analysis of a scene. However this task is computational intensive for conventional processors. In this work, a FPGA-based hardware architecture for real-time motion estimation is proposed. The technique used for motion estimation is a variation of the optical flow algorithm where the problem is reformulated as a sum of overlapped basis functions, and solved as a linear system. The proposed architecture is based on a systolic approach and is composed of parallel modules organized in a regular structure. The systolic processor accelerates the matrix operations required to achieve real-time performance. The architecture design is presented. Preliminary results are shown and discussed","PeriodicalId":443821,"journal":{"name":"2003 IEEE International Workshop on Computer Architectures for Machine Perception","volume":"212 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133177590","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 : 2003-05-12DOI: 10.1109/CAMP.2003.1598151
V. di Gesú, G. Lo Bosco, D. Tegolo
The paper describes an application of image retrieval based on DAISY architecture (distributed architecture for intelligent system). The creation of pictorial indexes may require a number of hours depending on the size of the pictorial data base. The problem can become more complex in the case of distributed database systems. In both cases a distributed architecture can be the natural and more efficient solution. DAISY architecture is based on the concept of co-operating behavioral agents supervised by a central engagement module. Preliminary experiments, to evaluate the performance of the system, have been performed on a astronomical database and coral image
{"title":"Distributed image retrieval on DAISY","authors":"V. di Gesú, G. Lo Bosco, D. Tegolo","doi":"10.1109/CAMP.2003.1598151","DOIUrl":"https://doi.org/10.1109/CAMP.2003.1598151","url":null,"abstract":"The paper describes an application of image retrieval based on DAISY architecture (distributed architecture for intelligent system). The creation of pictorial indexes may require a number of hours depending on the size of the pictorial data base. The problem can become more complex in the case of distributed database systems. In both cases a distributed architecture can be the natural and more efficient solution. DAISY architecture is based on the concept of co-operating behavioral agents supervised by a central engagement module. Preliminary experiments, to evaluate the performance of the system, have been performed on a astronomical database and coral image","PeriodicalId":443821,"journal":{"name":"2003 IEEE International Workshop on Computer Architectures for Machine Perception","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124238584","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 : 2003-05-12DOI: 10.1109/CAMP.2003.1598145
G. Iannizzotto, F. La Rosa, A. Rizzo, M. Xibilia
This paper introduces a still image segmentation technique based on an active contour obtained via single-layer CNNs. The contour initially laid on the frame of the image shrinks, deforms and multiplies until it matches the edges of each of the objects present in the scene. The shape of each object in the image is accurately extracted and nested objects, if any, are correctly detected. Experimental measures of the accuracy of the segmentation were carried out using the Hausdorff distance
{"title":"2D still-image segmentation with CNN-Amoeba","authors":"G. Iannizzotto, F. La Rosa, A. Rizzo, M. Xibilia","doi":"10.1109/CAMP.2003.1598145","DOIUrl":"https://doi.org/10.1109/CAMP.2003.1598145","url":null,"abstract":"This paper introduces a still image segmentation technique based on an active contour obtained via single-layer CNNs. The contour initially laid on the frame of the image shrinks, deforms and multiplies until it matches the edges of each of the objects present in the scene. The shape of each object in the image is accurately extracted and nested objects, if any, are correctly detected. Experimental measures of the accuracy of the segmentation were carried out using the Hausdorff distance","PeriodicalId":443821,"journal":{"name":"2003 IEEE International Workshop on Computer Architectures for Machine Perception","volume":"67 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121378559","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 : 2003-05-12DOI: 10.1109/CAMP.2003.1598152
L. Cinque, S. Levialdi, A. Malizia
The automatic document recognition is fundamental for office automation becoming every day a more powerful tool in those fields where information is still on paper. Document recognition follows from data acquisition, from both journals, and entire books in order to transform them in digital objects. We present a new architecture for document recognition that follows the open source methodologies for documents segmentation and classification, which turns to be beneficial in terms of computation efficiency, general-purpose availability and cost
{"title":"eXEDRA: a complete open source architecture for paper document recognition","authors":"L. Cinque, S. Levialdi, A. Malizia","doi":"10.1109/CAMP.2003.1598152","DOIUrl":"https://doi.org/10.1109/CAMP.2003.1598152","url":null,"abstract":"The automatic document recognition is fundamental for office automation becoming every day a more powerful tool in those fields where information is still on paper. Document recognition follows from data acquisition, from both journals, and entire books in order to transform them in digital objects. We present a new architecture for document recognition that follows the open source methodologies for documents segmentation and classification, which turns to be beneficial in terms of computation efficiency, general-purpose availability and cost","PeriodicalId":443821,"journal":{"name":"2003 IEEE International Workshop on Computer Architectures for Machine Perception","volume":"119 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116199403","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：