The Experts below are selected from a list of 2835 Experts worldwide ranked by ideXlab platform
Gerhard Sagerer - One of the best experts on this subject based on the ideXlab platform.
-
Integration of Vision and Speech Understanding Using Bayesian Networks
2000Co-Authors: Sven Wachsmuth, Gudrun Socher, Hans Brandt-pook, Franz Kummert, Gerhard SagererAbstract:The interaction of image and speech processing is a crucial property of multimedia systems. Classical systems using inferences on pure qualitative high-level descriptions miss much information when concerned with erroneous, vague, or incomplete data. We propose a new architecture that integrates various levels of processing by using multiple representations of the visually Observed Scene. The representations are vertically connected by Bayesian networks in order to find the most plausible interpretation of the Scene. The interpretation of a spoken utterance naming an object in the visually Observed Scene is modeled as another partial representation of the Scene. Using this concept, the key problem is the identification of the verbally specified object instances in the visually Observed Scene. Therefore, a Bayesian network is generated dynamically from the spoken utterance and the visual Scene representation.
-
ICVS - Multilevel Integration of Vision and Speech Understanding Using Bayesian Networks
Lecture Notes in Computer Science, 1999Co-Authors: Sven Wachsmuth, Gudrun Socher, Hans Brandt-pook, Franz Kummert, Gerhard SagererAbstract:The interaction of image and speech processing is a crucial property of multimedia systems. Classical systems using inferences on pure qualitative high level descriptions miss a lot of information when concerned with erroneous, vague, or incomplete data. We propose a new architecture that integrates various levels of processing by using multiple representations of the visually Observed Scene. They are vertically connected by Bayesian networks in order to find the most plausible interpretation of the Scene. The interpretation of a spoken utterance naming an object in the visually Observed Scene is modeled as another partial representation of the Scene. Using this concept, the key problem is the identification of the verbally specified object instances in the visually Observed Scene. Therefore, a Bayesian network is generated dynamically from the spoken utterance and the visual Scene representation. In this network spatial knowledge as well as knowledge extracted from psycholinguistic experiments is coded. First results show the robustness of our approach.
-
Multilevel Integration of Vision and Speech Understanding Using Bayseian Networks
1999Co-Authors: Sven Wachsmuth, Gudrun Socher, Hans Brandt-pook, Franz Kummert, Gerhard SagererAbstract:The interaction of image and speech processing is a crucial property of multimedia systems. Classical systems using inferences on pure qualitative high level descriptions miss a lot of information when concerned with erroneous, vague, or incomplete data. We propose a new architecture that integrates various levels of processing by using multiple representations of the visually Observed Scene. They are vertically connected by Bayesian networks in order to find the most plausible interpretation of the Scene. The interpretation of a spoken utterance naming an object in the visually Observed Scene is modeled as another partial representation of the Scene. Using this concept, the key problem is the identification of the verbally specified object instances in the visually Observed Scene. Therefore, a Bayesian network is generated dynamically from the spoken utterance and the visual Scene representation. In this network spatial knowledge as well as knowledge extracted from psycholinguistic experiments is coded. First results show the robustness of our approach.
Zheng Bao - One of the best experts on this subject based on the ideXlab platform.
-
A Knowledge-Based Target Relocation Method for Wide-Area GMTI Mode
IEEE Geoscience and Remote Sensing Letters, 2014Co-Authors: Ruixian Hu, Dongdong Liu, Baochang Liu, Tong Wang, Zheng BaoAbstract:This letter deals with the issue of moving-target relocation in wide-area ground moving-target indication for dual-channel radar systems. Due to channel mismatch, along-track baseline error, the existence of across-track baseline, etc., it is difficult to accurately relocate the detected targets. In this letter, we propose a new knowledge-based (KB) method for target relocation. The key idea of the proposed method is to use such knowledge that a moving target and its neighboring clutter, which is situated at the same azimuth angular position and the same range cell as this target in the Observed Scene, have the same interferometric phase. This new KB method, as an indirect one, does not employ the conventional relocation formula and therefore is not influenced by most of (if not all) interferometric phase errors. Experimental results from real radar data demonstrate a fairly high degree of target relocation accuracy.
Sven Wachsmuth - One of the best experts on this subject based on the ideXlab platform.
-
Integration of Vision and Speech Understanding Using Bayesian Networks
2000Co-Authors: Sven Wachsmuth, Gudrun Socher, Hans Brandt-pook, Franz Kummert, Gerhard SagererAbstract:The interaction of image and speech processing is a crucial property of multimedia systems. Classical systems using inferences on pure qualitative high-level descriptions miss much information when concerned with erroneous, vague, or incomplete data. We propose a new architecture that integrates various levels of processing by using multiple representations of the visually Observed Scene. The representations are vertically connected by Bayesian networks in order to find the most plausible interpretation of the Scene. The interpretation of a spoken utterance naming an object in the visually Observed Scene is modeled as another partial representation of the Scene. Using this concept, the key problem is the identification of the verbally specified object instances in the visually Observed Scene. Therefore, a Bayesian network is generated dynamically from the spoken utterance and the visual Scene representation.
-
ICVS - Multilevel Integration of Vision and Speech Understanding Using Bayesian Networks
Lecture Notes in Computer Science, 1999Co-Authors: Sven Wachsmuth, Gudrun Socher, Hans Brandt-pook, Franz Kummert, Gerhard SagererAbstract:The interaction of image and speech processing is a crucial property of multimedia systems. Classical systems using inferences on pure qualitative high level descriptions miss a lot of information when concerned with erroneous, vague, or incomplete data. We propose a new architecture that integrates various levels of processing by using multiple representations of the visually Observed Scene. They are vertically connected by Bayesian networks in order to find the most plausible interpretation of the Scene. The interpretation of a spoken utterance naming an object in the visually Observed Scene is modeled as another partial representation of the Scene. Using this concept, the key problem is the identification of the verbally specified object instances in the visually Observed Scene. Therefore, a Bayesian network is generated dynamically from the spoken utterance and the visual Scene representation. In this network spatial knowledge as well as knowledge extracted from psycholinguistic experiments is coded. First results show the robustness of our approach.
-
Multilevel Integration of Vision and Speech Understanding Using Bayseian Networks
1999Co-Authors: Sven Wachsmuth, Gudrun Socher, Hans Brandt-pook, Franz Kummert, Gerhard SagererAbstract:The interaction of image and speech processing is a crucial property of multimedia systems. Classical systems using inferences on pure qualitative high level descriptions miss a lot of information when concerned with erroneous, vague, or incomplete data. We propose a new architecture that integrates various levels of processing by using multiple representations of the visually Observed Scene. They are vertically connected by Bayesian networks in order to find the most plausible interpretation of the Scene. The interpretation of a spoken utterance naming an object in the visually Observed Scene is modeled as another partial representation of the Scene. Using this concept, the key problem is the identification of the verbally specified object instances in the visually Observed Scene. Therefore, a Bayesian network is generated dynamically from the spoken utterance and the visual Scene representation. In this network spatial knowledge as well as knowledge extracted from psycholinguistic experiments is coded. First results show the robustness of our approach.
Bruno Siciliano - One of the best experts on this subject based on the ideXlab platform.
-
Vision-based and IMU-aided scale factor-free linear velocity estimator
Autonomous Robots, 2017Co-Authors: Rafik Mebarki, Vincenzo Lippiello, Bruno SicilianoAbstract:This paper presents a new linear velocity estimator based on the unscented Kalman filter and making use of image information aided with inertial measurements. The proposed technique is independent of the scale factor in case of planar Observed Scene and does not require a priori knowledge of the Scene. Image moments of virtual objects, i.e. sets of classical image features such as corners collected online, are employed as the sole correcting information to be fed back to the estimator. Experimental results performed with a quadrotor equipped with a fisheye camera highlight the potential of the proposed approach.
Albert Rizzo - One of the best experts on this subject based on the ideXlab platform.
-
Immersive panoramic video
Proceedings of the eighth ACM international conference on Multimedia - MULTIMEDIA '00, 2000Co-Authors: Ulrich Neumann, Thomas Pintaric, Albert RizzoAbstract:Television and video images pervade our professional and home environments. For over fifty years video images have provided a virtual eye" into distant times and locations. Over the same period, video technology has matured from gray-scale images to big-screen color and digitally processed imagery. One aspect of both the delivery technology and the content creation has remained largely unchanged however the view is controlled at the source and identical for all observers. Panoramic video overcomes the passive and structured limitations of how video imagery is presented and perceived. The recent convergence of camera, processing, and display technologies make it possible to consider providing each viewer with individual control of their viewing direction. Viewers of panoramic video become virtual participants immersed in the Observed Scene, creating a new interactive dimension in the way people perceive video imagery within a virtual environment.
