Relevance feedback can effectively improve the performance of content-based multimedia retrieval systems. To be effective, a relevance feedback approach must be able to efficiently capture the user's query concept...
详细信息
ISBN:
(纸本)0780379659
Relevance feedback can effectively improve the performance of content-based multimedia retrieval systems. To be effective, a relevance feedback approach must be able to efficiently capture the user's query concept from a very limited number of training samples. To address this issue, we propose a novel adaptive classification method using random forests, which is a machine learning algorithm with proven good performance on many traditional classification problems. With random forests, our method reduces the relevance feedback to a two-class classification problem and classifies database objects as relevant or irrelevant. From the relevant object set, our approach returns the top k nearest neighbors of the query to the user. Briefly speaking, our relevance feedback method has the following dominant features. First, our method is able to address the multimodal distribution of relevant points, because it trains a nonparametric and nonlinear classifier, i.e., random forests, for relevance feedback. Second, it does not overfit training data because it uses an ensemble of tree classifiers to classify multimedia objects. Experiments on a Corel image set (with 31,438 images) show that our method significantly outperforms the state-of-the-art relevance feedback approaches.
In this paper we present an audio thumbnailing technique based on audio segmentation by similarity search. The segmentation is performed on MPEG-7 low level audio feature descriptors as a growing source of multimedia ...
详细信息
ISBN:
(纸本)0819451258
In this paper we present an audio thumbnailing technique based on audio segmentation by similarity search. The segmentation is performed on MPEG-7 low level audio feature descriptors as a growing source of multimedia meta data. Especially for database applications or audio-on-demand services this technique could be very helpful, because there is no need to have access to the probably copyright protected original audio material. The result of the similarity search is a matrix which contains off-diagonal stripes representing similar regions, which are usually the refrains of a song and thus a very suitable segment to be used as audio thumbnail. Using the a priori knowledge that we search off-diagonal stripes which must represent several seconds of audio data and that the adjustment of the stripes must be characteristically, we implemented a filter to enhance the structure of the similarity matrix and to extract a relevant segment as a audio thumbnail.
Digital multimedia data (i.e audio, video, and text) are difficult to store and replay due to their real time nature and large size. The requirement of continuous retrieval, in the presence of multiple media streams w...
详细信息
ISBN:
(纸本)0889863806
Digital multimedia data (i.e audio, video, and text) are difficult to store and replay due to their real time nature and large size. The requirement of continuous retrieval, in the presence of multiple media streams whose display must proceed in a mutually synchronized manner are the distinguishing features that are unique to digital multimedia. In this study, the problem of integration of compressed media segments is addressed, which are sequences of continuously recorded video frames, audio samples, and text on disk, for storage and retrieval. Motion based JPEG has been used for the compression of video frames as it allows on-line processing and random access of video frames. Companding scheme has been used for audio compression, whereas uncompressed text is used. The schemes used for compression of audio and video data both yield a fixed length code to save on the buffering requirements. A mechanism for integration is proposed that can be applied to arrange video along with the associated audio and text, into fixed length media segments before storing them on disk. All the algorithms have been implemented in the C language.
In this paper, we study the resource management problem in Direct Sequence - Wideband Code Division Multiple access (DS-WCDMA) systems of mobile communication network. The transmission power and transmission rate cont...
详细信息
ISBN:
(纸本)0780379063
In this paper, we study the resource management problem in Direct Sequence - Wideband Code Division Multiple access (DS-WCDMA) systems of mobile communication network. The transmission power and transmission rate control, in resource management are to be optimized while without violating the Quality of Service. Both single objective approach and multiobjective genetic algorithms have been investigated. Not only comparison has been conducted between the two cases, but their performance has also been compared with other currently available searching techniques.
Future Department of Defense warfighting concepts leverage information superiority and will require vast improvements in information transfer in terms of higher bandwidth, Quality of Service (QoS) support and connecti...
详细信息
ISBN:
(纸本)0780381408
Future Department of Defense warfighting concepts leverage information superiority and will require vast improvements in information transfer in terms of higher bandwidth, Quality of Service (QoS) support and connection to a high speed backbone. The new IEEE 802.16 broadband wireless access system is a viable alternative that can meet such requirements. In addition, this network can be swiftly deployed to interconnect the military theater, emergency response, and disaster relief operations to the backbone. Due to the diverse multimedia traffic with different priorities and QoS requirements, it is a well know fact that it is imperative to provide QoS support in military networks. However, the IEEE 802.16 provides only signaling mechanisms, but does not specify any scheduling or admission control algorithms that ultimately provide QoS support. In this paper we introduce a new scheduling algorithm for IEEE 802.16 broadband wireless access standard. The proposed solution which is practical and compatible to the IEEE 802.16 standard, provides QoS support to different traffic classes. To the best of our knowledge this is the first such algorithm. The simulation studies show that the proposed solution includes QoS support for all types of traffic classes as defined by the standard. We have shown the relationship between traffic characteristics and its QoS requirements and the network performance. This study will help network architects to decide the system parameters as well as the kind of traffic characteristics for which the network can provide QoS support.
An important application of multicasting is group-oriented multimedia communication, such as video-on-demand and video conferencing. Before these services can be successfully deployed, security infrastructures must be...
详细信息
Semantic Concept Detection from multimedia features enables high-level access to multimediacontent. While constructing robust detectors is feasible for concepts with sufficient training samples, concepts with fewer t...
详细信息
Semantic Concept Detection from multimedia features enables high-level access to multimediacontent. While constructing robust detectors is feasible for concepts with sufficient training samples, concepts with fewer training samples are hard to train efficiently. Comparable performance may be possible if the dependence of these concepts on the ones that can be robustly modeled is exploited. In this paper we show this phenomenon using the TREC Video 2002 Corpus as a test bed. Using a basic set of 12 semantic concepts modeled with Support Vector Machines, we predict presence of 4 other concepts. We then compare the performance of these predictors with direct SVM models for these 4 concepts and observe improvements of up to 150% in average precision.
The proceedings contains 75 papers. Topics discussed include self-stabilizing systems, sensor networks and energy management, peer to peer services, distributed algorithms, multimedia streaming, secure data sharing, q...
详细信息
The proceedings contains 75 papers. Topics discussed include self-stabilizing systems, sensor networks and energy management, peer to peer services, distributed algorithms, multimedia streaming, secure data sharing, quality of service, internet protocols, middleware communication services, fault tolerant systems, ubiquitous computing, network security, real time systems, web servers, middleware and interoperability, formal methods, Ad hoc network protocols, peer to peer infrastructure, synchronization, wide area distributed computing, mobile agents, program/software analysis, wireless access control and scalable data services.
Traditional cryptography systems treat every portion of a message (a video, an image, or an audio) equally and encrypt the entire message as a whole. As a result, those security systems often have only two states: acc...
详细信息
Traditional cryptography systems treat every portion of a message (a video, an image, or an audio) equally and encrypt the entire message as a whole. As a result, those security systems often have only two states: access authorization and access denial. To facilitate multi-level access control with interoperability, scalable security mechanism is needed. Further, the availability of varying network bandwidth and diverse receiver device capabilities demand scalable and flexible approaches that are capable of adapting to changing network conditions as well as device capabilities. In this paper, we describe a multimedia encryption scheme that supports scalability. One desirable feature of the proposed scheme is its simplicity and flexibility in supporting scalable contentaccess control.
Optimal multiuser detection is a 0-1 quadratic optimization problem with binary constraints which demands the search of the whole signal space to achieve the optimal solution. In the presence of such hardship, some so...
详细信息
暂无评论