Pub Date : 2003-06-16DOI: 10.1109/CVPRW.2003.10036
C. Stauffer
When tracking in a particular environment, objects tend to appear and disappear at certain locations. These locations may correspond to doors, garages, tunnel entrances, or even the edge of a camera view. A tracking system with knowledge of these locations is capable of improved initialization of tracking sequences, reconstitution of broken tracking sequences, and determination of tracking sequence termination. Further, knowledge of these locations is useful for activity-level descriptions of tracking sequences and for understanding relationships between non-overlapping camera views. This paper introduces a method for simultaneously solving these coupled problems: inferring the parameters of a source and sink model for a scene; and fixing broken tracking sequences and other tracking failures. A model selection criterion is also explained which allows determination of the number of sources and sinks in an environment. Results in multiple environments illustrate the effectiveness of this method.
{"title":"Estimating Tracking Sources and Sinks","authors":"C. Stauffer","doi":"10.1109/CVPRW.2003.10036","DOIUrl":"https://doi.org/10.1109/CVPRW.2003.10036","url":null,"abstract":"When tracking in a particular environment, objects tend to appear and disappear at certain locations. These locations may correspond to doors, garages, tunnel entrances, or even the edge of a camera view. A tracking system with knowledge of these locations is capable of improved initialization of tracking sequences, reconstitution of broken tracking sequences, and determination of tracking sequence termination. Further, knowledge of these locations is useful for activity-level descriptions of tracking sequences and for understanding relationships between non-overlapping camera views. This paper introduces a method for simultaneously solving these coupled problems: inferring the parameters of a source and sink model for a scene; and fixing broken tracking sequences and other tracking failures. A model selection criterion is also explained which allows determination of the number of sources and sinks in an environment. Results in multiple environments illustrate the effectiveness of this method.","PeriodicalId":121249,"journal":{"name":"2003 Conference on Computer Vision and Pattern Recognition Workshop","volume":"58 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120848886","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 : 2003-06-16DOI: 10.1109/CVPRW.2003.10087
A. Torii, Akihiro Sugimoto, A. Imiya
We formulate multiple-view geometry for omni-directional and panorama-camera systems. The mathematical formu-lations enable us to derive the geometrical and algebraic constraints for multiple panorama-camera configurations. The constraints permit us to reconstruct three-dimensional objects for a large feasible region.
{"title":"Mathematics of a Multiple Omni-Directional System","authors":"A. Torii, Akihiro Sugimoto, A. Imiya","doi":"10.1109/CVPRW.2003.10087","DOIUrl":"https://doi.org/10.1109/CVPRW.2003.10087","url":null,"abstract":"We formulate multiple-view geometry for omni-directional and panorama-camera systems. The mathematical formu-lations enable us to derive the geometrical and algebraic constraints for multiple panorama-camera configurations. The constraints permit us to reconstruct three-dimensional objects for a large feasible region.","PeriodicalId":121249,"journal":{"name":"2003 Conference on Computer Vision and Pattern Recognition Workshop","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127036992","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 : 2003-06-16DOI: 10.1109/CVPRW.2003.10075
C. Cauchois, E. Brassart, L. Delahoche, A. Clerentin
This paper deals with an absolute mobile robot self-localization algorithm in an indoor environment. Until now, localization methods based on conical omnidirectional vision sensors uniquely used radial segments from vertical environment landmarks projection. The main motivation of this work is to demonstrate that the SYCLOP sensor can be used as a vision sensor rather than a goniometric one. We will show how the calibration allows us to know the omnidirectional image formation process to compute a synthetic image base. Then, we will present the spatial localization method using a base of synthetics images and one real omnidirectional image. Finally, some experimental results obtained with real noisy omnidirectional images are shown.
{"title":"3D Localization with Conical Vision","authors":"C. Cauchois, E. Brassart, L. Delahoche, A. Clerentin","doi":"10.1109/CVPRW.2003.10075","DOIUrl":"https://doi.org/10.1109/CVPRW.2003.10075","url":null,"abstract":"This paper deals with an absolute mobile robot self-localization algorithm in an indoor environment. Until now, localization methods based on conical omnidirectional vision sensors uniquely used radial segments from vertical environment landmarks projection. The main motivation of this work is to demonstrate that the SYCLOP sensor can be used as a vision sensor rather than a goniometric one. We will show how the calibration allows us to know the omnidirectional image formation process to compute a synthetic image base. Then, we will present the spatial localization method using a base of synthetics images and one real omnidirectional image. Finally, some experimental results obtained with real noisy omnidirectional images are shown.","PeriodicalId":121249,"journal":{"name":"2003 Conference on Computer Vision and Pattern Recognition Workshop","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126799100","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 : 2003-06-16DOI: 10.1109/CVPRW.2003.10103
J. Marques, P. Jorge, A. Abrantes, J. M. Lemos
This paper describes an algorithm for tracking groups of objects in video sequences. The main difficulties addressed in this work concern total occlusions of the objects to be tracked as well as group merging and splitting. A two layer solution is proposed to overcome these difficulties. The first layer produces a set of spatio temporal strokes based on low level operations which manage to track the active regions most of the time. The second layer performs a consistent labeling of the detected segments using a statistical model based on Bayesian networks. The Bayesian network is recursively computed during the tracking operation and allows the update of the tracker results everytime new information is available. Experimental tests are included to show the performance of the algorithm in ambiguous situations.
{"title":"Tracking Groups of Pedestrians in Video Sequences","authors":"J. Marques, P. Jorge, A. Abrantes, J. M. Lemos","doi":"10.1109/CVPRW.2003.10103","DOIUrl":"https://doi.org/10.1109/CVPRW.2003.10103","url":null,"abstract":"This paper describes an algorithm for tracking groups of objects in video sequences. The main difficulties addressed in this work concern total occlusions of the objects to be tracked as well as group merging and splitting. A two layer solution is proposed to overcome these difficulties. The first layer produces a set of spatio temporal strokes based on low level operations which manage to track the active regions most of the time. The second layer performs a consistent labeling of the detected segments using a statistical model based on Bayesian networks. The Bayesian network is recursively computed during the tracking operation and allows the update of the tracker results everytime new information is available. Experimental tests are included to show the performance of the algorithm in ambiguous situations.","PeriodicalId":121249,"journal":{"name":"2003 Conference on Computer Vision and Pattern Recognition Workshop","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126816662","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 : 2003-06-16DOI: 10.1109/CVPRW.2003.10001
K. Ikeuchi, A. Nakazawa, K. Nishino, Takeshi Oishi
This paper overviews our research on digital preservation of cultural assets and digital restoration of their original appearance. Geometric models are digitally achieved through a pipeline consisting of scanning, registering and merging multiple range images. We have developed a robust simultaneous registration method and an efficient and robust voxel-based integration method. On the geometric models created, we have to align texture images acquired from a color camera. We have developed two texture mapping methods. In an attempt to restore the original appearance of historical heritage objects, we have synthesized several buildings and statues using scanned data and literature survey with advice from experts.
{"title":"Creating Virtual Buddha Statues through Observation","authors":"K. Ikeuchi, A. Nakazawa, K. Nishino, Takeshi Oishi","doi":"10.1109/CVPRW.2003.10001","DOIUrl":"https://doi.org/10.1109/CVPRW.2003.10001","url":null,"abstract":"This paper overviews our research on digital preservation of cultural assets and digital restoration of their original appearance. Geometric models are digitally achieved through a pipeline consisting of scanning, registering and merging multiple range images. We have developed a robust simultaneous registration method and an efficient and robust voxel-based integration method. On the geometric models created, we have to align texture images acquired from a color camera. We have developed two texture mapping methods. In an attempt to restore the original appearance of historical heritage objects, we have synthesized several buildings and statues using scanned data and literature survey with advice from experts.","PeriodicalId":121249,"journal":{"name":"2003 Conference on Computer Vision and Pattern Recognition Workshop","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122371631","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 : 2003-06-16DOI: 10.1109/CVPRW.2003.10097
H. Kjellström, S. Wirkander
This paper presents a particle filtering formulation for tracking an unknown and varying number of vehicles in terrain. The vehicles are modeled as a random set, i.e. a set of random variables, for which the cardinality is itself a random variable. The particle filter formulation is here extended according to finite set statistics (FISST) which is an extension of Bayesian theory to define operations on random sets. The filter was successfully tested on a simulated scenario with three vehicles moving in terrain, observed by humans in the terrain.
{"title":"Tracking Random Sets of Vehicles in Terrain","authors":"H. Kjellström, S. Wirkander","doi":"10.1109/CVPRW.2003.10097","DOIUrl":"https://doi.org/10.1109/CVPRW.2003.10097","url":null,"abstract":"This paper presents a particle filtering formulation for tracking an unknown and varying number of vehicles in terrain. The vehicles are modeled as a random set, i.e. a set of random variables, for which the cardinality is itself a random variable. The particle filter formulation is here extended according to finite set statistics (FISST) which is an extension of Bayesian theory to define operations on random sets. The filter was successfully tested on a simulated scenario with three vehicles moving in terrain, observed by humans in the terrain.","PeriodicalId":121249,"journal":{"name":"2003 Conference on Computer Vision and Pattern Recognition Workshop","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115935647","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 : 2003-06-16DOI: 10.1109/CVPRW.2003.10002
G. Godin, F. Blais, L. Cournoyer, J. Beraldin, J. Domey, John Taylor, M. Rioux, S. El-Hakim
Archaeology is emerging as one of the key areas of applications for laser range imaging. This particular context imposes a number of specific constraints on the design and operations of range sensors. In this paper, we discuss some of the issues in designing and using laser range sensor systems for archaeology. Results obtained on remote archaeological sites will serve to illustrate these considerations.
{"title":"Laser range imaging in archaeology: issues and results","authors":"G. Godin, F. Blais, L. Cournoyer, J. Beraldin, J. Domey, John Taylor, M. Rioux, S. El-Hakim","doi":"10.1109/CVPRW.2003.10002","DOIUrl":"https://doi.org/10.1109/CVPRW.2003.10002","url":null,"abstract":"Archaeology is emerging as one of the key areas of applications for laser range imaging. This particular context imposes a number of specific constraints on the design and operations of range sensors. In this paper, we discuss some of the issues in designing and using laser range sensor systems for archaeology. Results obtained on remote archaeological sites will serve to illustrate these considerations.","PeriodicalId":121249,"journal":{"name":"2003 Conference on Computer Vision and Pattern Recognition Workshop","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115396721","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 : 2003-06-16DOI: 10.1109/CVPRW.2003.10095
Qiang Zhang, Xiuwen Liu, Anuj Srivastava
Although linear representations of images are widely used in appearance-based recognition of objects, the frequently used ideas, such as PCA, ICA, and FDA, are often found to be suboptimal. A stochastic search algorithm has been proposed recently [4] for finding representations that are optimal for specific tasks and datasets. However, this search algorithm is computationally efficient only when the image size is relatively small. Here we propose a hierarchical learning algorithm to speed up the search. The proposed approach decomposes the original optimization problem into several stages according to a hierarchical organization. In particular, the following idea is applied recursively: (i) reduce the image dimension using a shrinkage matrix, (ii) optimize the recognition performance in the reduced space, and (iii)expand the optimal subspace to the bigger space in a pre-specified way. We show that the optimal performance is maintained in the last step. By applying this decomposition procedure recursively, a hierarchy of layers is formed. This speeds up the original algorithm significantly since the search is performed mainly in reduced spaces. The effectiveness of hierarchical learning is illustrated on a popular database, where the computation time is reduced by a large factor compared to the original algorithm.
{"title":"Statistical Search for Hierarchical Linear Optimal Representations of Images","authors":"Qiang Zhang, Xiuwen Liu, Anuj Srivastava","doi":"10.1109/CVPRW.2003.10095","DOIUrl":"https://doi.org/10.1109/CVPRW.2003.10095","url":null,"abstract":"Although linear representations of images are widely used in appearance-based recognition of objects, the frequently used ideas, such as PCA, ICA, and FDA, are often found to be suboptimal. A stochastic search algorithm has been proposed recently [4] for finding representations that are optimal for specific tasks and datasets. However, this search algorithm is computationally efficient only when the image size is relatively small. Here we propose a hierarchical learning algorithm to speed up the search. The proposed approach decomposes the original optimization problem into several stages according to a hierarchical organization. In particular, the following idea is applied recursively: (i) reduce the image dimension using a shrinkage matrix, (ii) optimize the recognition performance in the reduced space, and (iii)expand the optimal subspace to the bigger space in a pre-specified way. We show that the optimal performance is maintained in the last step. By applying this decomposition procedure recursively, a hierarchy of layers is formed. This speeds up the original algorithm significantly since the search is performed mainly in reduced spaces. The effectiveness of hierarchical learning is illustrated on a popular database, where the computation time is reduced by a large factor compared to the original algorithm.","PeriodicalId":121249,"journal":{"name":"2003 Conference on Computer Vision and Pattern Recognition Workshop","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114829977","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 : 2003-06-16DOI: 10.1109/CVPRW.2003.10034
A. Ekin, A. Tekalp
This paper presents real-time, or near real-time, probabilistic event detection methods for broadcast sports video using cinematic features, such as shot classes and slow-motion replays. Novel algorithms have been developed for probabilistic detection of soccer goal events and basketball play-break events in a generic framework. The proposed framework includes generic algorithms for automatic dominant (field) color region detection and shot boundary detection, and domain-specific shot classification algorithms for soccer and basketball. Finally, the detected goal events in soccer and play events in basketball are employed to generate summaries of long games. The efficiency and effectiveness of the proposed system and the algorithms have been shown over more than 13 hours of sports video captured by the broadcasters from different regions around the world.
{"title":"Generic Event Detection in Sports Video using Cinematic Features","authors":"A. Ekin, A. Tekalp","doi":"10.1109/CVPRW.2003.10034","DOIUrl":"https://doi.org/10.1109/CVPRW.2003.10034","url":null,"abstract":"This paper presents real-time, or near real-time, probabilistic event detection methods for broadcast sports video using cinematic features, such as shot classes and slow-motion replays. Novel algorithms have been developed for probabilistic detection of soccer goal events and basketball play-break events in a generic framework. The proposed framework includes generic algorithms for automatic dominant (field) color region detection and shot boundary detection, and domain-specific shot classification algorithms for soccer and basketball. Finally, the detected goal events in soccer and play events in basketball are employed to generate summaries of long games. The efficiency and effectiveness of the proposed system and the algorithms have been shown over more than 13 hours of sports video captured by the broadcasters from different regions around the world.","PeriodicalId":121249,"journal":{"name":"2003 Conference on Computer Vision and Pattern Recognition Workshop","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127374583","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 : 2003-06-16DOI: 10.1109/CVPRW.2003.10019
J. Ryu, Berthold K. P. Horn, M. S. Mermelstein, S. Hong, D. M. Freeman
We describe how structured illumination patterns can be used to increase the resolution of an imaging system for optical microscopy. A target is illuminated by a sequence of finely textured light patterns formed by the interference of multiple coherent beams. The sequence of brightness values reported from a single pixel of a CCD imager encodes the target contrast pattern with sub-pixel resolution. Fourier domain components at spatial frequencies contained in the probing illumination patterns can be recovered from the pixel brightness sequence by solving a set of over-determined linear equations. We show that uniform angular spacing of the beams generating the illumination patterns leads to less than ideal sampling of the transform space and we propose alternative geometric arrangements. We describe an image reconstruction algorithm based on the Voronoi diagram that applies when the transform domain is not sampled uniformly. Finally, the contrast patterns within individual pixels can be spliced together to forman image encompassing multiple pixels.
{"title":"Application of Structured Illumination in Nano-Scale Vision","authors":"J. Ryu, Berthold K. P. Horn, M. S. Mermelstein, S. Hong, D. M. Freeman","doi":"10.1109/CVPRW.2003.10019","DOIUrl":"https://doi.org/10.1109/CVPRW.2003.10019","url":null,"abstract":"We describe how structured illumination patterns can be used to increase the resolution of an imaging system for optical microscopy. A target is illuminated by a sequence of finely textured light patterns formed by the interference of multiple coherent beams. The sequence of brightness values reported from a single pixel of a CCD imager encodes the target contrast pattern with sub-pixel resolution. Fourier domain components at spatial frequencies contained in the probing illumination patterns can be recovered from the pixel brightness sequence by solving a set of over-determined linear equations. We show that uniform angular spacing of the beams generating the illumination patterns leads to less than ideal sampling of the transform space and we propose alternative geometric arrangements. We describe an image reconstruction algorithm based on the Voronoi diagram that applies when the transform domain is not sampled uniformly. Finally, the contrast patterns within individual pixels can be spliced together to forman image encompassing multiple pixels.","PeriodicalId":121249,"journal":{"name":"2003 Conference on Computer Vision and Pattern Recognition Workshop","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121804588","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：