S. Adali, K.S. Candan, S.S. Chen, K. Erol, and V.S. Subrahmanian, "The advanced video information system: Data structures and query processing," Multimedia Systems, Vol. 4, No. 4, pp. 172--186, 1996.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....on top of the IFO [1] conceptual data model to describe inner structure and contents of images. OVID [14] is a prototype video object database system. OVID offers schemaless description of database, interval inclusion inheritance, and composition of video objects based on IS A hierarchy. In [2], a way of organizing and structur ing video data to facilitate queries is described. There are two concepts which needs to be retrieved from the queries: entities and video frames. Entities can be video objects which are present in video frames or activities which are subjects in the sequence of ....

....a video clip. Run corresponds to the action (what) in the forest denotes where the run occurs, John represents the object, to be healthy denotes the reason for his run, slowly corresponds to how fast he runs and in the morning represents when he runs. This event structure is similar to the one in [2]. Audio clip is a sequence of audio frames. Each audio frame is a one dimensional signal that has frequency and amplitude as physical properties and title and keywords as description of the signal. The schema of audio clip is depicted in Figure 4 (a) Audio Audial Clip Frames Audio Frame ....

S. Adali, S. Candan, S. Chen, K. Erol, S. Subrahmanian, "The advanced video information systems: data structures and query processing," Multimedia Systems, Vol.4, pp. 172-186, 1996

.... designed and implemented, such as VideoQ, KMED, QBIC, and OVID [6, 8, 15, 34] Furthermore, querying video objects by motion properties has also been studied extensively [16, 25, 27, 33, 39] Some examples on using semantic properties of video data for querying video collections can be found in [2, 19, 21]. Nonetheless, to the best of our knowledge, no proposal has been made This work is supported by the Scientific and Research Council of Turkey (TB ITAK) under Project Code 199E025. thus far for a generic, application independent video database management system that targets to support ....

....a complete deductive system as we do. The authors extend their work defining feature subfeature relationships in [30] When a query cannot be answered, it is relaxed by substituting a subfeature for a feature. This relaxation technique provides some support for reasoning with uncertainty. In [2], a prototype video information system, called Advanced Video Information System (AVIS) is introduced. The authors propose a special kind of segment tree, namely frame segment tree, and a set of arrays to represent objects, events, activities, and their associations. The proposed data model is ....

[Article contains additional citation context not shown here]

S. Adal, K.S. Candan, S. Chen, K. Erol, and V.S. Subrahmanian. Advanced video information systems: Data structures and query processing. ACM Multimedia Systems, 4:172--186, 1996. 29

....levels to enable spec ification of video structure, the annotations, and sharing and reuse in one enhanced ER model. The thematic indexing is achieved by annotations defined for video segments, and by specific annotation entities corresponding to persons, locations, and events. Adali et al. [4] introduce AVIS with a formal video model in terms of interesting video objects. A video object in AVIS refers to a semantic entity that attracts attention in a scene. The model includes events as the instantiation of activity types and the roles of objects in the events. AVIS is supported by an ....

S. Adali, K.S. Candan, S.S. Chen, K. Erol, and V.S. Subrahmanian, "The Advanced Video Information System: Data Structures and Query Processing," Multimedia Systems, vol.4, pp. 172-186, 1996.

....of video. Modeling events by using spatio temporal attributes of objects is performed but this can only be used for specific domains like sports videos. A pass event in a soccer game can be modeled by using spatio temporal attributes but a party event cannot be modeled in this way. In [1], a semantic video model is proposed and the algorithms for handling di#erent types of queries are implemented within a prototype, called Advanced Video Information System (AVIS) In this model, video is divided into fixed time duration frame sequences. Activities, events and objects are related ....

S. Adali, K.S. Candan, S. Chen, K. Erol, and V.S. Subrahmanian. Advanced video information system: Data structures and query processing. ACM-Springer Multimedia Systems Journal, 4(4):172--186, 1996.

....be distinguished, together with some examples, will be given in section 5. Finally, in section 6 we end with a perspective on open research questions. 2 An author s perspective on video documents In contrast to other frameworks, that view video documents from the (visual) data perspective, e.g. [2], we view a video document as a result of an authoring process. Consequence of this approach is that it allows for integration of di#erent modalities more easily. To arrive at our framework for video indexing, we first consider video creation. In this survey we restrict ourselves to video made ....

S. Adali, K.S. Candan, S-S. Chen, K. Erol, and V.S. Subrahmanian. The advanced video information system: Data structures and query processing. Multimedia Systems, 4(4):172--186, 1996.

....be distinguished, together with some examples, will be given in section 5. Finally, in section 6 we end with a perspective on open research questions. 2 An author s perspective on video documents In contrast to other frameworks, that view video documents from the (visual) data perspective, e.g. [2], we view a video document as a result of an authoring process. Consequence of this approach is that it allows for integration of di erent modalities more easily. To arrive at our framework for video indexing, we rst consider video creation. In this survey we restrict ourselves to video made ....

S. Adali, K.S. Candan, S-S. Chen, K. Erol, and V.S. Subrahmanian. The advanced video information system: Data structures and query processing. Multimedia Systems, 4(4):172-186, 1996.

....a complete deductive system as we do. The authors extend their work defining feature subfeature relationships in [33] When a query cannot be answered, it is relaxed by substituting a subfeature for a feature. This relaxation technique provides some support for reasoning with uncertainty. In [1], a special kind of segment tree called frame segment tree and a set of arrays to represent objects, events, activities and their associations are introduced. The proposed model is based on the generic multimedia model described in [34] Additional concepts introduced in the model are activities, ....

....di#erent types of semantic queries were implemented within a prototype, called Advanced Video Information System (AVIS) However, objects have no attributes other than the roles defined for the events. In [19] an SQL like video query language, based on the data model developed by Adal et al. [1], is proposed. Nevertheless, the proposed query language does not provide any support for temporal queries on events. Nor does it have any language construct for spatio temporal querying of video clips since it was designed for semantic queries on video data. In our query model, temporal ....

[Article contains additional citation context not shown here]

S. Adal, K.S. Candan, S. Chen, K. Erol, and V.S. Subrahmanian. Advanced video information systems: Data structures and query processing. ACM Multimedia Systems, 4:172--186, 1996.

....through frames. A large number of artificial insertions is created and thus the index storage requirements increase. A better solution is to store the functions describing how objects move or vary their extents. In animated movies an object s frame evolution is represented by some function [1]. Even though general functions can be used, for simplicity we assume an object can move or grow shrink through a linear function of time. Then a new record is inserted only when the parameters describing an object s (movement or extent) function change. The new record will maintain the object s ....

....and video databases, the approach discussed in this paper is novel. The work in [48] considers only static objects (degenerate case) and uses a 3D R Tree approach to index the objects. Another work that proposes indexing video objects in order to answer mostly temporal queries appears in [1]. In this paper, video movies are pre processed and all entities of interest such as objects, activities, and events, are identified. Subsequently, these entities are associated with specific frames in which they appear. Therefore, every entity is coupled with a set of frames which can be viewed ....

S. Adali, K. Candan, S. Chen, K. Erol, V. S. Subrahmanian. The Advanced Video Information System: Data Structures and Query Processing. In ACM Multimedia Systems, 4 (4):172-186,1996.

....hierarchically as nodes in a tree. As a benefit, nested relationships between the nodes allow user to explore the context in which a node appears. Examples of such an extension can be found in the algebraic video data model [12] and the video object database system OVID [13] Other approaches [14, 15] introduce additional video entities, such as objects and events, as well as their relations, that should be annotated, because they are subjects of interests in video. Another way to model the video entities assumes using spatio temporal relations. The concept of video object can be associated ....

S. Adali, K. S. Candan, S-S. Chen, K. Erol, V. S. Subrahmanian, Advanced Video Information System: Data Structure and Query Processing, Multimedia System 4(4), 1996, 172-86.

....originalities of our approach, is that it allows annotations to be independently attached to any level of video structuration. This feature considerably increases the expressive power of the resulting model. Indeed, in most previous proposals (Oomoto and Tanaka, 1993; Gandhi and Robertson, 1994; Adali et al. 1996; Li et al. 1997) annotations may only be attached to the video frames and not to its shots or its scenes. This withholds the user from expressing properties which are true of a given scene, without being true of each frame in this scene. For instance, stating that in a given scene two ....

....1997) The innovation of our approach, lies on the orthogonality with which the different aspects of video modeling are tackled. Indeed, in our proposal annotations may be independently attached to each level of the video structuration, whereas in most of the existing video data models (e.g. AVIS (Adali et al. 1996) and CVOT (Li et al. 1997) annotations may only be attached to the frames. Moreover, existing approaches in this area ne glect the issue of propagating structuration and annotations when composing videos. Hjelsvold et al. 1995) is perhaps one of the closest works to ours. This paper ....

Adali, S., Candan, K. S., Chen, S., Erol, K., and Subrahmanian, V. (1996). The Advanced Video Information System: data structures and query processing. Multimedia Systems, 4:172 -- 186.

....The low level schemes, however, are very limited in expressing queries. For example, it would be difficult to ask for a video clip showing the sinking of the ship in the movie Titanic using only color, texture, and audio information. In contrast, video data models based on semantic content ( [6, 15, 14, 10, 1, 9, 5, 17, 12]) are capable of supporting more natural queries. They, however, must rely partially on manual annotation. A limitation of this approach is that semantic content can be ambiguous and context dependent. This problem can be controlled by limiting the context and providing multiple semantic ....

....graph. Algorithm 1 [Refinement Algorithm] Given a video graph G = V , e, f , g, h) for a video. The corresponding refined graph is built as the following steps. 1. V 2 is initialized to the set of all non key objects in the source video graph. 2. For each node v 2 V 2 , add a new node storing [0,1] and add a link labeled TIME from v to the new node. 3. Initialize UpdateCounter and UpdateCounter 1 to 0. 4. For any two internal nodes v and v 0 that are connected by an r link, adjust the value of their TIME node so as to satisfy h(v, v 0 ) If a node is a key object with a complete TIME ....

S. Adali, K. S. Candan, S.-S. Chen, K. Erol, and V. S. Subrahmanian. Advanced video information system: Data structures and query processing. ACM-Springer Multimedia Systems Journal, 1996.

....is closest to and which complements our approach in the context of data modeling. Table 1 summarizes the key efforts in the area of data modeling techniques from a multimedia information selection perspective. 13 Key related work in the video domain for the selection of video segments includes [1, 38, 68]. Of these, Omoto et al. uses a knowledge hierarchy to facilitate annotation, while others use simple keyword based techniques without a hierarchy (see Table 1) The model of Omoto et al. fails to provide a mechanism that automatically converts generalized descriptions into specialized ones. ....

....from existing objects, and some attributes values are inherited from these existing objects based on the principle of interval inclusion. Omoto et al. also proposes a query language, VideoSQL, to retrieve video objects through the specification of specific attributes and values. 14 Adali et al. [1] develop a video data model that exploits spatial data structures (e.g. characters in a movie) rather than temporal objects as is the case in our data model. In Adali et al. objects, activities, and roles are identified in the video frames. Each object and event is associated with a set of frame ....

[Article contains additional citation context not shown here]

S. Adali, K. S. Candan, S. Chen, K. Erol, and V. S. Subrahmanian, "Advanced Video Information System: Data Structures and Query Processing," ACMSpringer Multimedia Systems Journal, vol. 4, pp. 172-186, 1996.

....through frames. A large number of arti cial insertions is created and thus the index storage requirements increase. A better solution is to store the functions describing how objects move or vary their extents. In animated movies an object s frame evolution is represented by some function [1]. Even though general functions can be used, for simplicity we assume an object can move or grow shrink through a linear function of time. Then a new record is inserted only when the parameters describing an object s (movement or extent) function change. The new record will maintain the object s ....

....and video databases, the approach discussed in this paper is novel. The work in [48] considers only static objects (degenerate case) and uses a 3D R Tree approach to index the objects. Another work that proposes indexing video objects in order to answer mostly temporal queries appears in [1]. 28 0 5000 10000 15000 20000 25000 30000 35000 40000 45000 500 1000 1500 2000 2500 Avg. # of Objects per frame 3D R tree maxMBR PP R tree Greedyx1.5 3D R tree Greedyx1.5 PP R tree Figure 30: Storage consumption for the GN datasets. 0 5 10 15 20 25 30 35 40 45 50 ....

S. Adali, K. Candan, S. Chen, K. Erol, V. S. Subrahmanian. The Advanced Video Information System: Data Structures and Query Processing. In ACM Multimedia Systems, 4 (4):172-186,1996.

....into a set of temporally ordered shots and then build a multilevel abstraction upon them. The drawback of this approach, as pointed out in [14] is the lack of flexibility and the incapability of representing semantics residing in overlapped segments. In a contrary direction, stratification models [14, 9, 1, 16, 12] segment contextual information of the video instead of simply partitioning. The video units, each called a stratum, can overlap and encompass each other and are associated with a time interval corresponding to a physical segment in the video stream. Most recently is [7] which extends the ....

....SUBJECT, ACTION, TIMEg, f Gamma (oe) Gamma , fDOM (SUBJECT) fDOM (ACTION) 2 = f Delta (oe) fffi i g for i = 1. 15. ffi 1 = 1, SUBJECT: froof, town squareg, TIME: 0, 3] ffi 2 = 2, SUBJECT: fbank, black iron barsg) ffi 3 = 3, ACTION: fcarryg, SUBJECT: fbriefcase, handg, TIME: [4, 1]) ffi 4 = 4, ACTION: fassembleg, SUBJECT: frifle, murdererg, TIME: 5, 1] ffi 5 = 5, ACTION: fplaceg, SUBJECT: fmoney, briefcaseg, TIME: 0, 7] ffi 6 = 6, ACTION: fsmileg, SUBJECT: fteeth, murdererg) ffi 7 = 7, ACTION: fshutg, SUBJECT: fbriefcaseg, TIME: 7, 1] ffi 8 = 8, ACTION: ....

[Article contains additional citation context not shown here]

S. Adali, K. S. Candan, S.-S. Chen, K. Erol, and V. S. Subrahmanian. Advanced video information system: Data structures and query processing. ACM-Springer Multimedia Systems Journal, 1996.

....6 Although we use audio, here we show related work in the video domain which is closest to and which complements our approach in the context of data modeling for the facilitation of information selection requests. Key related work in the video domain for selection of video segments includes [1, 11, 16]. Of these, Omoto et al. 16] use a knowledge hierarchy to facilitate annotation, while others use simple keyword based techniques without a hierarchy. The model of Omoto et al. fails to provide a mechanism that automatically converts a generalized description into a specialized one(s) Further, ....

....also be retrieved In order to evaluate the retrieval performance of two systems, we can employ an F score [88] The F score is the harmonic mean of recall and precision, a single measure that combines recall and precision. The function ensures that an F score will have values within the interval [0, 1]. The F score is 0 when no relevant documents have been retrieved, and it is 1 when all retrieved documents are relevant. Furthermore, the harmonic mean F assumes a high value only when both precision and recall are high. Therefore, determination of the maximum value for F can be interpreted as an ....

S. Adali, K. S. Candan, S. Chen, K. Erol, and V. S. Subrahmanian, "Advanced Video Information System: Data Structures and Query Processing," ACM-Springer Multimedia Systems Journal, vol. 4, pp. 172-186, 1996.

....hidden or suppressed meaning, which the same video content may have, we make the problem even more complex. The simplest way to model the video content is by using free text manual annotation. An example is stratification approach [8] with a few extensions [9, 10] Some other approaches [11, 12] introduce additional video entities, such as objects and events, as well as their relations, that should be annotated, because they are subjects of interests in video. Another way to model the video entities assumes using spatio temporal relations. The concept of video object can be associated ....

S. Adali, K. S. Candan, S-S. Chen, K. Erol, V. S. Subrahmanian, "Advanced Video Information System: Data Structure and Query Processing", Multimedia System Vol. 4, No. 4, Aug. 1996, pp. 172-86.

.... Attempts to structure and index multimedia, and thereby provide random access, have depended on simplification either by restriction to a narrow and inherently well structured domain (e.g. news broadcasts [12] or by restriction to a single information channel (e.g. image recognition [1], 10] Multimedia retrieval is a hard problem. The information is diverse and challenges even confounds any single indexing technique. In our opinion, only a collection of complementary techniques will truly suffice, with the strength of one compensating for the weakness of another. Our ....

S. Adali, S. Candan, S. Chen, K. Erol, V. Subrahmanian, "The Advanced Video Information System: data structures and query processing", Multimedia Systems, 4(4): 173-186, 1996.

....structures and software packages. These currently include (or have included in the past) relational database management systems (Oracle, Ingres, Dbase, Paradox) an object oriented system (ObjectStore) a multimedia system called MACS [Brink et al. 1995] a video information system called AVIS [Adali et al. 1996], a geographic data structure called a PRquadtree, arbitrary flat files (as long as their schemas are specified) a US Army route planner over free terrain [Benton and Subrahmanian, 1994] a variety of US Army logistics data including specialized Oracle and nested multirecord TAADS data [Schafer ....

Adali, S., Candan, K. S., Chen, S.-S., Erol, K., and Subrahmanian, V. S. (1996). Advanced Video Information Systems:Data Structures and Query Processing. Multimedia Systems, 4(4):172--186.

....addition to that, if we consider multiple semantic meaning such as metaphorical, associative, hidden or suppressed meaning, which the same video content may have, we make a problem more complex. The simplest way to model the video content is by using free text manual annotation. Some approaches [8, 9] introduce additional video entities, such as objects and events, as well as their relations, that should be annotated, because they are subjects of interests in video. One of the major limitations of these approaches is that search process is based mainly on the attribute information, which are ....

S. Adali, K. S. Candan, S-S. Chen, K. Erol, V. S. Subrahmanian, "Advanced Video Information System: Data Structure and Query Processing", Multimedia System Vol. 4, No. 4, Aug. 1996, pp. 172-86.

....number of artificial insertions and thus it increases the index storage space. A better way is to store the functions describing how objects move or vary their extents. This is particularly the case in animated movies where an object s frame evolution is represented by some function (see also [1]) Even though general functions can be used, for simplicity we assume an object can move or grow shrink through a linear function of time. Then a new record is inserted only when the parameters describing an object s movement or 3 extent functions change. The new record will maintain the ....

....and video databases, the approach discussed in this paper is novel. The work in [45] considers only static objects (degenerate case) and uses a 3D R Tree approach to index the objects. Another work that proposes indexing video objects in order to answer mostly temporal queries appears in [1]. In this, video movies are preprocessed and all entities of interest such as objects, activities, and events, are identified. Subsequently, these entities are associated with specific frames in which they appear. Therefore, every entity is coupled with a set of frames which can be viewed as a set ....

S. Adali, K. Seljuk Candan, S. Chen, K. Erol, V. S. Subrahmanian. The Advanced Video Information System: Data Structures and Query Processing. In ACM Multimedia Systems, 4 (4):172-186,1996.

....organised hierarchically as nodes in a tree. As a benefit, nested relationships between the nodes allow user to explore the context in which a node appears. Examples of such a extension can be found in algebraic video data model [12] and video object database system OVID [13] Other approaches [14, 15] introduce additional video entities, such as objects and events, as well as their relations, that should be annotated, because they are subjects of interests in video. Another way to model the video entities assumes using spatio temporal relations. The concept of video object can be associated ....

S. Adali, K. S. Candan, S-S. Chen, K. Erol, V. S. Subrahmanian, Advanced Video Information System: Data Structure and Query Processing, Multimedia System 4(4), 1996, 172-86.

....originalities of our approach, is that it allows annotations to be independently attached to any level of video structuration. This feature considerably increases the expressive power of the resulting model. Indeed, in most previous proposals (Oomoto and Tanaka, 1993; Gandhi and Robertson, 1994; Adali et al. 1996; Li et al. 1997) annotations may only be attached to the video frames and not to its shots or its scenes. This withholds the user from expressing facts which are true of a given scene, without being true of each frame in this scene. For instance, stating that in a given scene two characters ....

....Indeed, in our proposal annotations may be independently attached to each level of the video structuration, whereas in most of the existing video data models, Published in Proc. of the AAAI 2000 Workshop on Spatial and Temporal Granularities 6 REFERENCES e.g. OVID (Oomoto and Tanaka, 1993) AVIS (Adali et al. 1996) and CVOT (Li et al. 1997) annotations may only be attached to the frames 2 . This approach withholds the user from expressing facts which are true of a given scene, without being true of each frame in that scene. For instance, suppose that the user wishes to encode into a set of annotations, ....

Adali, S., Candan, K. S., Chen, S., Erol, K., and Subrahmanian, V. (1996). The Advanced Video Information System: data structures and query processing.

....a framework for video modelling with emphasis on representation of semantic content of video data. We describe our formal video model in detail and show that the proposed model facilitates the efficient execution of different types of contentbased queries. The work presented in the paper builds on [2,3,4,5,6]. Fig. 1. Layer hierarchy of video data model Static features Temporary extended features colours shapes textures spatial relations motion spatio temporal relations Feature layer Object layer Event layer Video raw data Roughly, our approach distinguishes four layers in ....

S. Adali, K. S. Candan, S-S. Chen, K. Erol, V. S. Subrahmanian, "Advanced Video Information System: Data Structure and Query Processing", Multimedia System Vol. 4, No. 4, Aug. 1996, pp. 172-86.

....there is little research work on finding semantic foundations for representing and querying video information. This paper is a contribution in this direction. The framework presented here integrates formalisms developed in constraint, object and sequence databases. The paper builds on the works of [32, 34, 23, 1, 30, 37, 8, 31, 18] to propose a hybrid data model for video data and a declarative, rule based, constraint query language, that has a clear declarative and operational semantics. We make the following contributions: 1. We develop a simple video data model on the basis of relation, object and constraint paradigms. ....

....systems have been developed, among others, VIOLONE [41] and JACOB [28] However, because of the weakness of content analysis algorithms, they focus on a very specific exploitation. On the other hand, much more aided video content indexing systems have been designed, among others, OVID [34] AVIS [1], or VideoStar [23] Some fascinating database issues in the context of video data and multimedia in general (see, among others, 14, 21, 6] are note presented here. In the context of image and video data, queries can be formulated using several techniques, which fall broadly into two ....

[Article contains additional citation context not shown here]

Sibel Adali, Kasim S. Candan, Su-Shing Chen, Kutluhan Erol, and V. S. Subrahmanian. Advanced video information system: Data structures and query processing. ACM-Springer Multimedia System, 4:172--186, 1996.

....systems have been developed, among others, VIOLONE [29] and JACOB [18] However, because of the weakness of content analysis algorithms, they focus on a very specific exploitation. On the other hand, much more aided video content indexing systems have been designed, among others, OVID [22] AVIS [1], or VideoStar [14] In the context of image and video data, queries can be formulated using several techniques, which fall broadly into two categories: textual and visual. Several systems have been developed to retrieve visual data based on color, shape, size, texture, image segments, keyword, ....

....to retrieve visual data based on color, shape, size, texture, image segments, keyword, relational operators, objects, and bibliographic data (see, among others, 6, 7, 13, 16, 21] In this paper, we focus on textual languages. The work presented here is closest to and complements the ones in [20, 1, 22, 14]. Meghini [20] proposed a retrieval model for images based on first order logical language which spans along four main dimensions: visual, spatial, mapping and content. Queries on images can address anyone of these dimensions or any combination of them. In the proposed model, objects cannot be ....

[Article contains additional citation context not shown here]

S. Adali, K. S. Candan, S.-S. Chen, K. Erol, and V. S. Subrahmanian. Advanced video information system: Data structures and query processing. ACM-Springer Multimedia System, 4:172--186, 1996.

....Modeling and Querying Video Data. Oomoto and Tanaka [8] proposed a schema less video object data model. They focus on the capabilities of object oriented database features (their extension) for supporting schema evolution and to provide a mechanism for sharing some descriptive data. Adali et al. [1] have developed a formal video data model, and they exploit spatial data structures for storing such data. Hjelsvold and Midtstraum [7] proposed a generic video data model. Their proposal combines ideas from the stratification and the segmentation approaches. Decleir et al. 5] proposed a data ....

Sibel Adali, Kasim S. Candan, Su-Shing Chen, Kutluhan Erol, and V. S. Subrahmanian. Advanced Video Information System: Data Structures and Query Processing. ACM-Springer Multimedia System, 4:172-- 186, 1996. 3 Quality implies accessibility, optimization, validation, etc.

....to store and manage video content (e.g. raw features, annotations) Recently, this field has received a particular attention. Two approaches have been considered: 1) systems that use only automated feature extraction, like Jacob [4] and (2) more elaborated systems for concrete applications [3] [1] [5] that integrate user specifications in their video data model. In this paper, we develop a generic data model to represent user annotations as well as computer extracted features, and a rule based query language. The model is based on the notion of objects of interest that can be annotated ....

S. Adali, K.S. Candan, S. Chen, K. Erol, and V.S. Subrahmanian. Advanced video information system : Data structures and query processing. Multimedia Systems, 1996.

....of the weakness of content analysis algorithms, they focus on a very specific exploitation. 2. Computer aided content indexing approach (i.e. indices are provided by users ( supported by some tools) by analyzing video content) Systems based on this approach are, among others, OVID [17] AVIS [1], and VideoStar [10] A database support for video information will help sharing information among applications and make it available for analysis. Therefore, new technologies and techniques are required for organizing, storing, manipulating and retrieving by content video data. Although various ....

....is little research work on finding semantic foundations for representing and querying video information. This paper is a contribution in this direction. The framework presented here integrates formalisms developed in video, constraint object and sequence databases. The paper builds on the works of [1, 15, 10, 17, 20, 5, 16, 8] to propose a data model for video databases and a declarative, rule based, constraint query language, that has a clear declarative and operational semantics. We make the following contributions: 1. We develop a simple and useful video data model on the basis of relation, object and constraint ....

S. Adali, K. S. Candan, S.-S. Chen, K. Erol, and V. S. Subrahmanian. Advanced video information system: Data structures and query processing. ACM-Springer Multimedia System, 1995.

....work on finding a common terminology and semantic foundations for representing and querying video information. This paper is a contribution in this direction. The framework presented here integrates formalisms developed in video, constraint, and object databases. The paper builds on the works of [1, 9, 7, 10, 11] to propose a data model for video databases and a declarative, rule based, constraint query language, that has a clear declarative and operational semantics. We make the following contributions: 1. We develop a simple video data model on the basis of relation, object and constraint paradigms. ....

....are values. If o = oid; v) with v = A 1 : v 1 ; An : vn ] then attr(o) denotes the set of all attributes in v (i.e. fA 1 ; An g) and value(o) denotes the value v, that is, v = value(o) A value v i is denoted by o:A i . 3.1. Example Let us see how the example given in [1] can be modeled in our framework. First let us recall the example: It concerns the movie: The Rope by Alfred Hitchcock. In the movie, two friends, Philip and Brandon decide to commit the perfect crime. They want to prove they are of the privileged group of people who are allowed to kill just for ....

S. Adali, K. S. Candan, S.-S. Chen, K. Erol, and V. S. Subrahmanian. Advanced video information system: Data structures and query processing. ACM-Springer Multimedia System.

....This work was supported in part by the Army Research Laboratory under Cooperative Agreement No. DAAL01 96 2 0003 and in part by NSF DARPA NASA Digital Library Initiative Program under Cooperative Agreement No. 94 11318. Also in part by 863 High Tech No. 863 306 04 03 3 and NSF of China. [2, 3]. For example, in [3] the system manually annotated detailed information from video, thus the system can deal with very complicated queries, such as find all people who appear in frames in which Gene Kelly and Giner Rogers are getting married . Apparently it is impractical to do the annotation ....

....in part by the Army Research Laboratory under Cooperative Agreement No. DAAL01 96 2 0003 and in part by NSF DARPA NASA Digital Library Initiative Program under Cooperative Agreement No. 94 11318. Also in part by 863 High Tech No. 863 306 04 03 3 and NSF of China. 2, 3] For example, in [3], the system manually annotated detailed information from video, thus the system can deal with very complicated queries, such as find all people who appear in frames in which Gene Kelly and Giner Rogers are getting married . Apparently it is impractical to do the annotation manually since there ....

S. Adali, K. Candan, S.-S. Chen, K. Erol, and V. Subrahmanian, "Advanced video information system: Data structures and query processing," To appear in ACM-Springer Multimedia Systems Journal.

....we assume that production rates for disk servers vary depending on the number of streams they serve. 1. 3 Related Work Recent research on multimedia presentations concentrated on synchronization models for presentations [Stein90, RanRVK93, BlaS96, Haindl96] presentation querying languages [LSBBOO97, ACCKS96], systems for authoring, retrieval and scheduling of presentations [CandPS96, Dalal96] and automated construction of presentations [HKO97, HO97] The area of continuous media storage servers is currently a very active research area [OBRS94, GarOS98, GhKS95] 2. Unlimited level of production: ....

Adali, S. et al, "The Advanced Video Information System: Data Structures and Query Processing", ACM Multimedia Systems Journal, 1996.

....domain, we assume that production rates for disk servers vary depending on the number of streams they serve. 1. 3 Related Work Recent research on multimedia presentations concentrated on synchronization models for presentations [Stein90, RanRVK93, BlaS96, Haindl96] presentation querying languages [LSBBOO97, 4 ACCKS96], systems for authoring, retrieval and scheduling of presentations [CandPS96, Dalal96] and automated construction of presentations [HKO97, HO97] The area of continuous media storage servers is currently a very active research area [OBRS94, GarOS98, GhKS95] Multimedia storage servers deal with ....

Adali, S. et al, "The Advanced Video Information System: Data Structures and Query Processing", ACM Multimedia Systems Journal, 1996.

....measure. To overcome the above problem, MARS uses the Gaussian normalization to map the feature components within the feature vector to the N(0,1) range [7] ffl Inter feature Normalization: It is used to map the distance values of the image from the query based on a feature vector into the range [0,1] so that they can be interpreted as the degree of membership in the fuzzy model or relevance probability in the probability model. Similar to intra feature normalization, Gaussian normalization is also used for inter feature normalization [7] 2.5 Feature Weighting Techniques used in MARS While ....

....color, textures, shapes, etc. provide a powerful representation for content based retrieval, such low level descriptors may not be powerful enough to capture the rich content of a video. Instead, the representation will need to be augmented with a semantic description of video content (as in [1, 5]) The information retrieval techniques will then need to be generalized for content based retrieval over the semantically augmented representation. Such retrieval techniques are being currently explored in MARS. ....

Sibel Adali, K.S. Candan, Su-Shing Chen, Kutluhan Erol, and V.S. Subrahmanian. Advanced video information system: Data structures and query processing. To appear in ACM-Springer Multimedia Systems Journal.

.... query initialization wait for response display the results times. Executing Remote Calls with Caching and or Invariants: Figure 5 shows a small representative sample of the times obtained when running queries that required accessing data operations in a video retrieval package called AVIS [3]. It is easy to see from these figures that using caches always leads to savings in time when the software data is located at remote sites. Furthermore, using invariants is useful when the query is not explicitly cached in such cases both partial invariants and equality invariants lead to ....

S. Adali, K.S. Candan, Su-Shing Chen, K. Erol and V.S. Subrahmanian. (1995) Advanced Video Information System: Data Structures and Query Processing., Accepted for publication in the ACM-Springer Multimedia Systems Journal.

No context found.

S. Adali, K.S. Candan, S.S. Chen, K. Erol, and V.S. Subrahmanian, "The advanced video information system: Data structures and query processing," Multimedia Systems, Vol. 4, No. 4, pp. 172--186, 1996.

No context found.

. Adali, S., Candan, K.S., Chen, S., Erol, K., Subrahmanian, V.: The advanced video information system: Data structures and query processing. Multimedia Systems, Vol.4, pages 172-186, 1996

No context found.

Adal S, Candan KS, Chen S, Erol K, Subrahmanian VS (1996) Advanced video information systems: data structures and query processing. ACM Multimedia Sys 4:172--186

No context found.

S. Adali, K. Seljuk Candan, S. Chen, K. Erol, V. S. Subrahmanian. The Advanced Video Information System: Data Structures and Query Processing. In ACM Multimedia Systems, 4 (4):172-186,1996.

No context found.

S. Adal, K.S. Candan, S. Chen, K. Erol, and V.S. Subrahmanian. Advanced video information systems: Data structures and query processing. ACM Multimedia Systems, 4:172--186, 1996.

No context found.

Adali, S., K. Candan, S. Chen, K. Erol, and V. Subrahmanian: 1996, `Advanced Video Information Systems: Data Structures and Query Processing'. ACM Multimedia Systems 4, 172--186.

No context found.

S. Adali, K.S. Candan, S.S. Chen, K. Erol, and V.S. Subrahmanian, "The Advanced Video Information System: Data Structures and Query Processing," Multimedia Systems, vol. 4, pp. 172-186, 1996.

No context found.

S. Adali, K. S. Candan, S.-S. Chen, K. Erol, and V. S. Subrahmanian. The Advanced Video Information System: Data Structures and Query Processing. Multimedia Systems, 4(4):172--186, 1996.

No context found.

S. Adali, K. S. Candan, S.-S. Chen, K. Erol, and V. S. Subrahmanian. Advanced Video Information Systems:Data Structures and Query Processing. Multimedia Systems, 4(4):172-186, 1996.

No context found.

Adali, S., Candan, K.S., Chen, S-S., Erol, K., Subrahmanian, S., "Advanced Video Information System: Data Structures and Query processing", ACM Multimedia Systems Journal, Aug. 1996.

No context found.

S. Adali, K. Selcuk Candan, S.S. Chen, K. Erol, and V.S. Subrahmanian. The Advanced Video Information System: data structures and query processing. Multimedia Systems, 4:172 -- 186, 1996.

No context found.

S. Adali, K. S. Candan, S. S. Chen, K. Erol, and V. S. Subrahmanian. Advanced video information system: Data structures and query processing. In ACM-Springer Multimedia Systems Journal (in press), 1995.

No context found.

S. Adali, K.S. Candan, S. Chen, K. Erol, and Subrahmanian V.S. The advanced video information system: data structures and query processing. Multimedia Systems, 4,pages=172-186, August 1996.

No context found.

S. Adali, K. S. Candan, S. S. Chen, K. Erol, and V. S. Subrahmanian. Advanced video information system: Data structures and query processing. accepted for publication by ACM Multimedia Journal, 1995.

No context found.

S. Adali, K. S. Candan, S. S. Chen, K. Erol, and V. S. Subrahmanian. The advanced video information system: Data structures and query processing. ACM-Springer Multimedia Systems Journal, 4(August):172---186, 1996.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC