Iconic Indexing Research Articles

KEYPICS: FREE–HAND DRAWN ICONIC KEYWORDS

We propose an iconic indexing of images to be exposed on the Web. This should be accomplished by Keypics, i.e. auxiliary, simplified pictures referring to the geometrical and/or the semantic content of the indexed image. Keypics should not be rigidly standardized; they should be left free to evolve, to express nuances and to stress details. A mathematical tool for dealing with such freedom, in the retrieval task, already exists: Size Functions. An experiment on 494 Keypics with Size Functions based on three measuring functions (distances, projections and jumps) and their combination is presented.

An efficient iconic indexing strategy for image rotation and reflection in image databases

Spatial relationships are important issues for similarity-based retrieval in many image database applications. With the popularity of digital cameras and the related image processing software, a sequence of images are often rotated or flipped. That is, those images are transformed in the rotation orientation or the reflection direction. However, many iconic indexing strategies based on symbolic projection are sensitive to rotation or reflection. Therefore, these strategies may miss the qualified images, when the query is issued in the orientation different from the orientation of the database images. To solve this problem, some researchers proposed a function to map the spatial relationship to its transformed one. However, this mapping consists of several conditional statements, which is time-consuming. Thus, in this paper, we propose an efficient iconic indexing strategy, in which we carefully assign a unique bit pattern to each spatial relationship and record the spatial information based on the bit patterns in a matrix. Without generating the rotated or flipped image, we can directly derive the index of the rotated or flipped image from the index of the original one by bit operations and matrix manipulation. In our performance study, we analyze the time complexity of our proposed strategy and show the efficiency of our proposed strategy according to the simulation results. Moreover, we implement a prototype to validate our proposed strategy.

Image database design based on 9D-SPA representation for spatial relations

Spatial relationships between objects are important features for designing a content-based image retrieval system. We propose a new scheme, called 9D-SPA representation, for encoding the spatial relations in an image. With this representation, important functions of intelligent image database systems such as visualization, browsing, spatial reasoning, iconic indexing, and similarity retrieval can be easily achieved. The capability of discriminating images based on 9D-SPA representation is much more powerful than any spatial representation method based on minimum bounding rectangles or centroids of objects. The similarity measures using 9D-SPA representation provide a wide range of fuzzy matching capability in similarity retrieval to meet different user's requirements. Experimental results showed that our system is very effective in terms of recall and precision. In addition, the 9D-SPA representation can be incorporated into a two-level index structure to help reduce the search space of each query processing. The experimental results also demonstrated that, on average, only 0.1254 percent /spl sim/ 1.6829 percent of symbolic pictures (depending on various degrees of similarity) were accessed per query in an image database containing 50,000 symbolic pictures.

On Completeness of Reasoning about Planar Spatial Relationships in Pictorial Retrieval Systems

In this paper we consider the completeness problem of reasoning about planar spatial relationships in pictorial retrieval systems. We define a large class of two-dimensional scenes, the extended pseudo-symbolic pictures. The existing rule system R is proved to be complete for (ex- tended) pseudo-symbolic pictures. We also introduce a new iconic indexing, the (extended) pseudo- 2D string representation, for them. The (extended) pseudo-2D string has the good properties of the 2D string. It is unambiguous, like the augmented 2D string, and has a compact form suitable for image retrieval. We then present efficient algorithms to determine whether a given planar picture is (extended) pseudo-symbolic or not, and if it is, these algorithms also return its (extended) pseudo-2D string representation. Picture retrieval by (extended) pseudo-2D strings is also discussed. 1. Introduction. Image database systems have been much studied over the past 20 years. One of the most important problems in the design of image database systems is how images are stored in the image databases (5, 6, 9, 11, 12, 24). While the use of indexing to allow database accessing has been well established in traditional database systems, content-based picture indexing techniques need to be developed for facilitating pictorial information retrieval from a pictorial database. Tanimoto (25) suggested the use of picture icons as picture indexes, thus intro- ducing the concept of iconic indexing. Subsequently, Chang et al. (11) developed the concept of iconic indexing by introducing the 2D string representation of the image. The 2D string approach is based on the idea that the spatial knowledge contained in a real picture can be suitably represented by a symbolic picture (i.e., a matrix of symbols) where every symbol corresponds to a significant element of the image. The position of a symbol in the grid corresponds to the position of the centroid of the rep- resented significant element. Depending on the application, the significant elements of the image can be pixels, lines, regions, and objects, etc. A 2D string representing a symbolic picture is derived from the picture by orthogonally projecting its symbols by columns and by rows. This approach gives an efficient and natural way to con- struct iconic indexes for two-dimensional pictures. With the 2D string approach, the problem of pictorial information retrieval for 2D pictures becomes a problem of 2D

A general approach to indexing and retrieval of images in image databases

In this paper we present a very general approach to indexing and retrieval of images in image database systems. We allow dierent ways of indexing and matching for objects in an image, and integrate them with our existing approach of the generalized extended pseudo-2D string and generalized combined 2D string representation of the image to formulate a more general framework for iconic indexing and retrieval of images. Image retrieval is then discussed, and the query processing of image retrieval consists of both matching of objects and matching of spatial relationships. This suggestion for general-purpose image database systems is particularly suitable for building huge, heterogeneous and distributed image databases. 1. Introduction. Image database systems have been very active over the past 20 years. With the explosive interest for the last 10 years in multimedia systems, content-based image retrieval has attracted the attention of researchers across sev- eral disciplines (7), including computer vision, pattern recognition, human-computer interaction, and image databases. Applications that use image databases include oce

A bit-pattern-based matrix strategy for efficient iconic indexing of symbolic pictures

In this paper, we propose an efficient iconic indexing strategy called bit-pattern-based matrix (BP matrix) for symbolic pictures, in which each spatial relationship between any two objects along the x-axis (or y-axis) is represented as a binary-bit pattern, and is recorded in a matrix. There are 12 bits in each bit pattern. When the bits in a certain subset of those 12 bits are set to 1, they denote a certain spatial relationship. Bit-wise-and/bit-wise-or operations are used for query processing; therefore, they are efficient enough as compared to the previous approaches. From our simulation, we show that the proposed BP matrix strategy requires shorter time for query processing than the generalized prime-number-based matrix strategy.

3D-List: a data structure for efficient video query processing

A video query model based on the content of video and iconic indexing is proposed. We extend the notion of two-dimensional strings to three-dimensional strings (3D-Strings) for representing the spatial and temporal relationships among the symbols in both a video and a video query. The problem of video query processing is then transformed into a problem of three-dimensional pattern matching. To efficiently match the 3D-Strings, a data structure, called 3D-List, and its related algorithms are proposed. In this approach, the symbols of a video in the video database are retrieved from the video index and organized as a 3D-List according to the 3D-String of the video query. The related algorithms are then applied on the 3D-List to determine whether this video is an answer to the video query. Based on this approach, we have started a project called Vega. In this project, we have implemented a user friendly interface for specifying video queries, a video index tool for constructing the video index, and a video query processor based on the notion of 3D-List. Some experiments are also performed to show the efficiency and effectiveness of the proposed algorithms.

A generalized prime-number-based matrix strategy for efficient iconic indexing of symbolic pictures

In this paper, we propose an efficient iconic indexing strategy called generalized prime-number-based matrix (GPN Matrix) for symbolic pictures, in which each spatial relationship between any two objects is represented as a product of some prime numbers from a set of 12 prime numbers and is recorded in a matrix. In the proposed strategy, we classify 169 spatial relationships between two objects in 2D space into five spatial categories, and define a generalized category rule (based on module operations) for each of those five spatial categories. As compared to the prime-number-based matrix (PN Matrix) strategy ( Chang and Yang, 1997), in which each spatial relationship between any two objects is represented as a product of some prime numbers from a set of 17 prime numbers, the GPN Matrix strategy has a smaller storage space requirement than the PN Matrix strategy, which also improves the query processing time.

A spatio-temporal semantic model for multimedia database systems and multimedia information systems

As more information sources become available in multimedia systems, the development of abstract semantic models for video, audio, text, and image data is becoming very important. An abstract semantic model has two requirements: it should be rich enough to provide a friendly interface of multimedia presentation synchronization schedules to the users and it should be a good programming data structure for implementation in order to control multimedia playback. An abstract semantic model based on an augmented transition network (ATN) is presented. The inputs for ATNs are modeled by multimedia input strings. Multimedia input strings provide an efficient means for iconic indexing of the temporal/spatial relations of media streams and semantic objects. An ATN and its subnetworks are used to represent the appearing sequence of media streams and semantic objects. The arc label is a substring of a multimedia input string. In this design, a presentation is driven by a multimedia input string. Each subnetwork has its own multimedia input string. Database queries relative to text, image, and video can be answered via substring matching at subnetworks. Multimedia browsing allows users the flexibility to select any part of the presentation they prefer to see. This means that the ATN and its subnetworks can be included in multimedia database systems which are controlled by a database management system (DBMS). User interactions and loops are also provided in an ATN. Therefore, ATNs provide three major capabilities: multimedia presentations, temporal/spatial multimedia database searching, and multimedia browsing.

A unique-ID-based matrix strategy for efficient iconic indexing of symbolic pictures

In this paper, we propose an efficient iconic indexing strategy called unique-ID-based matrix (UID matrix) for symbolic pictures, in which each spatial relationship between any two objects is assigned with a unique identifier (ID) and is recorded in a matrix. Basically, the proposed strategy can represent those complex relationships that are represented in 2D C-strings in a matrix, and an efficient range checking operation can be used to support pictorial query, spatial reasoning and similarity retrieval; therefore, they are efficient enough as compared to the previous approaches. From our simulation, we show that the proposed UID matrix strategy requires shorter time to convert the input data into the corresponding representation than the 2D C-string strategy, so is the case with query processing. Moreover, our proposed UID matrix strategy may require lesser storage cost than the 2D C-string strategy in some cases.

Symbolic picture retrieval by relative-metric spatial relations

Symbolic pictures can be used for iconic indexing, spatial reasoning, and similarity retrieval in the design of intelligent image database systems. [S. K. Chang, C. W. Yan, Donald C. Dimitroff, and Timothy Arndt, IEEE Trans. Software Engineering 1988, 14, 681–688; S.-K. Chang, Principles of Pictorial Information Systems Design, Prentice-Hall, New York, 1989.] However, previous approaches to designing such systems usually ignore relative-metric information on symbolic pictures and cause several deficiencies in indexing, spatial reasoning, and retrieval. In our approach, we extract relative-metric information from symbolic pictures and use such information to help establish indexes based on an improvement from a minimal perfect hashing scheme. As a result, more accurate picture retrieval can be achieved through our indexing mechanism. Capabilities in spatial reasoning and query representation/processing are also improved. By utilizing relative-metric spatial relations, an image database system becomes more flexible and intelligent. ©2000 John Wiley & Sons, Inc.

Breadth-first search oriented symbolic picture representation for spatial match retrieval

To provide sufficient information for percepting symbolic objects among pictures from pictorial information systems, an efficient spatial data structure is needed to identify objects and to specify the spatial relationships among objects in a picture. In this paper, a new symbolic picture representation is proposed through the traversal of a quadtree corresponding to a given picture using breadth-first search and thus is called BFS string. An iconic indexing scheme for retrieving symbolic pictures based on the proposed BFS strings is presented and the algorithms introduced in the bin-tree oriented iconic indexing scheme by Chang and Lee (1998b) [Journal of Data and Knowledge Engineering 26 (2)]. is applied to the scheme for manipulating spatial relationships among objects. Compared with the bin-tree oriented scheme in storage complexity, experimental results show that the proposed BFS string requires less storage space when the picture has a great number of objects.

2D C-Tree Spatial Representation for Iconic Image

The 2D string approaches provide a natural way of constructing iconic indexing for images. The 2D C-string representation with an efficient cutting mechanism is more characteristic of spatial knowledge and efficient in the representation of iconic images. However, the computation of object ranks in a 2D C-string might make the inference of spatial reasoning somewhat complicated. This shortcoming is overcome by the 2D C-tree representation. The 2D C-tree not only keeps the comprehensive spatial knowledge in the original 2D C-string, but also the ordered labeled tree is more suitable for spatial reasoning and image retrieval. The spatial knowledge can be derived directly from the inference rules embedded in the characteristic structure of the 2D C-tree representation.

Comments on “mesh and pyramid algorithms for iconic indexing”

Comments on “mesh and pyramid algorithms for iconic indexing”

Comments on “Mesh and Pyramid Algorithms for Iconic indexing”: authors’ reply

Comments on “Mesh and Pyramid Algorithms for Iconic indexing”: authors’ reply

A Prime-Number-based Matrix strategy for efficient iconic indexing of symbolic pictures

In the previous approaches to represent pictorial data, as the complexity of the representation strategy is increased, the more spatial relationships can be represented, which also results in a more complex strategy for query processing and a limited types of queries that can be answered. In this paper, we propose an efficient iconic indexing scheme called Prime-Number-based Matrix (PN Matrix) for symbolic pictures, which combines the advantages of the 2D C-string and the 9DLT matrix. Basically, the proposed strategy can represent those complex relationships which are represented in 2D C-strings in a matrix, and an efficient module-based operation can be used to support pictorial query, spatial reasoning and similarity retrieval. In the proposed scheme, we classify 169 spatial relationships between two objects in 2D space into five spatial categories, and define a category rule for each of those five spatial categories. Those category rules are module-operation-based; therefore, they are efficient enough as compared to the previous approaches. Following those category rules, we propose algorithms to efficiently support spatial reasoning, picture queries and similarity retrieval based on a data structure of a Prime-Number-based Matrix (PN Matrix).

A Unified Approach to Iconic Indexing, Retrieval, and Maintenance of Spatial Relationships in Image Databases

In this paper we propose a unified iconic indexing, the generalized combined 2D string representation, for images in image databases. Each 2D image is modelled as a generalized extended pseudo-symbolic picture, which has the GEP-2D string representation. We present an efficient algorithm to generate the GEP-2D string representation for each 2D image. We also show how to maintain the complete information about the absolute spatial relationships in the image. Our proposed iconic indexing combines both the GEP-2D string representation and the usual 2D string representation to capture absolute and relative spatial relationships in the image. The result is better representation of spatial relationships in image databases. These results extend our earlier work on a large class of 2D scenes, the extended pseudo-symbolic pictures. Picture retrieval by generalized combined 2D strings is discussed. Our approach can also be easily formulated for 3D images.

Using 2D C +-strings as spatial knowledge representation for image database systems

The pictorial data structure based on 2D strings provides an efficient method for iconic indexing in image database systems. Previous approaches ignore the sizes of and distances between objects. These representations have deficiencies in picture reconstruction, similarity retrieval, and spatial reasoning. In this paper, we propose a new spatial knowledge representation scheme called 2D C +-strings as well as a picture reconstruction algorithm. As a result, our scheme offers the following major advantages: (1) more accuracy in picture reconstruction; (2) less ambiguity in similarity retrieval; (3) reasoning about relative sizes, locations, and distances for a symbolic picture is possible.

Retrieving the most similar symbolic pictures from pictorial databases

In this article, we suggest an iconic indexing mechanism for spatial similarity retrieval on iconic image databases based upon the spatial relationships among the objects in a picture. The iconic objects we deal with are some kinds of gross panorama of simple objects. We also assume that any one iconic object is not distinguished from any other object of the same kind. For our mechanism, we first transform each iconic picture into a set of ordered triples ( O i , O j , R ij ) where O i and O j are objects and R ij is the predefined spatial relationship codes between O i and O j . Then we construct a set of hashing functions for all spatial relationship codes R ij , separately, associated with all ordered pairs ( O i , O j ) extracted from the ordered triples ( O i , O j , R ij ). Thereafter, an iconic index table can be established according to the constructed hashing functions for all predefined spatial relationship codes. By applying the constructed hashing functions, the most similar pictures in the database satisfying a specified query can be fast determined. We can easily extend our mechanism for handling the case when some new spatial relationship codes are defined later for the considerations of refined spatial similarity retrieval under the maximum-likelihood measure criterion.

Mesh and pyramid algorithms for iconic indexing

Parallel algorithms on meshes and pyramids for iconic indexing are presented. The algorithms are asymptotically superior to previously known parallel algorithms. Also presented are experimental results for these algorithms on the Connection Machine (CM-2).