No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Quadtree representations for storage and manipulation of clusters of images
Abstract
This article gives an overview of different approaches proposed for the storage and manipulation of clusters of images. Clustering images consists of grouping together images having a defined relationship. In this article, images are represented by quadtrees implemented in a hierarchical or linear way. The discussion, presented at the end of the article, allows selecting a quadtree-based representation well adapted to a specific area of application or to the characteristics of the manipulated images. oui
... However, using the approximate values of internal nodes appearing at level 2, with δ = L 1 : S (2) (1, 2) = |1/4 − 1/2| * 1/9 = 1/36. Several approaches, presented and compared in Manouvrier et al. (2002), propose to store similar images organized in quadtrees. Their main goal is to reduce the memory space used by image quadtrees by sharing common parts between quadtrees. ...
... As a consequence, the S-distance can be used in these approaches to organize the images in a hierarchy: an image i is a descendant of an image i , in the hierarchy, if for all image j of the hierarchy: S(i, i ) ≤ S(i, j). The smaller the S-distance between image trees is, the greater the sharing of node values between their trees is-see in Manouvrier et al. (2002) for more details. However, two images i and j with two very different multi-level feature vectors, i.e. ...
... Each weight w n corresponds to the proportion of feature points in the sub-quadtree rooted by n in comparison with the total number of feature points in the entire quadtree. In Jomier et al. (2000) and Manouvrier et al. (2002), the S-distance is used to measure the sharing ratio between quadtrees in order to optimize the storage of similar image clusters. Lu et al. (1994), Lin et al. (2001) and Jomier et al. (2005) represent each image of the database by a full fixed-depth balanced quadtree. ...
This article presents the Δ-distance, a family of distances between images recursively decomposed into segments and represented by multi-level feature vectors. Such a structure is a quad, a quin or a nona-tree resulting from a fixed and arbitrary image partition or from an image segmentation process. It handles positional information of image features (e.g. color, texture or shape). Δ-distance is the generalized form of dissimilarity measures between multi-level feature vectors. Using different weights on tree nodes and different distances between nodes, distances between trees or visual similarity between images can be computed based on the general definition of Δ. In this article, we present three Δ-based distance families: two families of distances between tree structures, called
-distance( for Tree) and
-distance ( for Segment), and a family of visual distances between images, called ( for Visual). The -distance visually compares two images using their tree representation and the other two distances compare the tree structures resulting from image segmentation. Moreover, we show how existing distances between multi-level feature vectors appear to be particular cases of the Δ-distance
... Previous research on quadtree based data structures and query processing focused on storage and manipulation of clusters of (overlapped) images. We refer readers to [[55] and the review article by Manouvrier[56] for more details. These data structures are not tailored for read-only applications and they may not be suitable for supporting visual explorations of large-scale raster data, especially in a Web environment. ...
... Instead of reporting the cells individually from the query result, using a collection of quadrants may significantly save memory consumption and improve query results rendering for visualization purposes. Assuming that a quadtree for each environmental variable has been constructed, it is not difficult to see that evaluating the compound condition can be achieved by synchronized traversal of relevant quadtrees as described in (Manouvrier et al 2002) and will be omitted here due to space constraints. We next focus on the construction of a single quadtree and evaluating a single condition on a quadtree. ...
Traditionally environmental scientists are limited to simple display and animation of large-scale raster geospatial data derived
from remote sensing instrumentation and model simulation outputs. Identifying regions that satisfy certain range criteria,
e.g., temperature between [t1,t2) and precipitation between [p1,p2), plays an important role in query-driven visualization
and visual exploration in general. In this study, we have proposed a Binned Min-Max Quadtree (BMMQ-Tree) to index large-scale
numeric raster geospatial data and efficiently process queries on identifying regions of interests by taking advantages of
the approximate nature of visualization related queries. We have also developed an end-to-end system that allows users visually
and interactively explore large-scale raster geospatial data in a Web-based environment by integrating our query processing
backend and a commercial Web-based Geographical Information System (Web-GIS). Experiments using real global environmental
data have demonstrated the efficiency of the proposed BMMQ-Tree. Both experiences and lessons learnt from the development
of the prototype system and experiments on the real dataset are reported.
... Oracle GeoRaster 19 allows storing the bounding boxes and derived attributes of tile images as vector geospatial data, which subsequently can be indexed and queried so that only selected tile images need to be retrieved for display. A few of existing works have addressed the issue of managing a set of similar/related rasters for efficient query processing based on the concept of overlapping quadtrees (Tzouramanis et al 1998, Manolopoulos et al 2001, Manouvrier et al 2002). The techniques are similar to indexing spatial-temporal vector geospatial data such as Historical R-Tree (Nascimento et al 1998), MV3R-Tree (Tao and Papadias 2001) and TPR-Tree (Saltenis 2000) from a methodological perspective. ...
The increasingly available Graphics Processing Units (GPU) hardware resources and the emerging General Purpose computing on GPU (GPGPU) technologies provide an alternative and complementary solution to existing cluster based high-performance geospatial computing. However, the complexities of the unique GPGPU hardware architectures and the steep learning curve of GPGPU programming have imposed signficant technical challenges on the geospatial computing community to develop efficient parallel geospatial data structures and algorithms that can make full use of the hardware capabilities to solve ever growing large and complex real world geospatial problems. In this study, we propose a practical approach to simplifying high-performance geospatial computing on GPGPUs by using parallel primitives. We take a case study of quadtree construction on large-scale geospatial rasters to demonstrate the effectiveness of the proposed approach. Comparing the proposed parallel primitives based implementation with a naïve CUDA implementation, a signficant reduction on coding complexity and a 10X speedup have been achieved. We believe that GPGPU based software development using generic parallel primitives can be a first step towards developing geospatial-specific and more efficient parallel primitives for high-performance geospatial computing in both personal and cluster computing environments and boost the performance of geospatial cyberinfrastructure.
... Oracle GeoRaster [43]allows storing the bounding boxes and derived attributes of tile images as vector geospatial data, which subsequently can be indexed and queried so that only selected tile images need to be retrieved for display. A few of existing works have addressed the issue of managing a set of similar/related rasters based on overlapping quadtrees [62][34] [35]. The techniques are similar to indexing spatial-temporal vector geospatial data such as Historical R-Tree [38], MV3R-Tree [59], TPR-Tree [51] from a methodology perspective. ...
Advances in geospatial technologies have generated large amounts of raster geospatial data. Massively parallel General Purpose Graphics Processing Unit (GPGPU) computing technologies have provided personal computers with tremendous computing capabilities. In this paper, we report our work on fast indexing of large-scale raster geospatial data using GPGPU computing. We have designed a cache conscious quadtree data structure (CCQ-Tree) that is suitable for GPU indexing. A set of algorithms have been developed and integrated to construct CCQ-Trees on GPUs by utilizing multiple pyramid data structures and Z-order based prefix sum. Experiments on multiple 4096*4096 blocks of a global precipitation raster data have shown that CCQ-Tree indexing using a 112-core Nvidia Quadro FX3700 GPU device reduces construction times from around 9.83 seconds to 0.42 seconds (23X speedup).
... If an image is not homogeneous (according to a particular criterion), the quadtree root has four descendant nodes representing the four first level image quadrants. A node is a leaf when its corresponding image quadrant is homogeneous; otherwise the node is internal [3]. In general, the quadtree is unbalanced. ...
Application areas such as medical imaging or satellite imaging often store large collections of similar images. Lossless compression techniques are usually needed in such critical applications. Previous researches have introduced the centroid method, which gets benefit from the inter-image redundancy ìthe set redundancyî. In this paper a new algorithm is proposed as an extension of the centroid method. Experimental results with two sets of CT and MRI brain images demonstrate the efficiency and superiority of the proposed algorithm in respect to compression ratio.
... The Quadtree structure has been widely used in the fields of image processing [Fazekas and Santa, 2000;Yang et al., 2000;Manouvrier et al., 2002], computer graphics [Samet and Webber, 1988], GIS [Shaffer et al., 1990;Wang and Armstrong, 2003], image databases [El-Qawasmeh, 2003], and image compression [ Wang et al., 2000;Chung and Tseng, 2001]. A Quadtree is a member of a class of hierarchical structure based on the recursive division of the image in four disjoint quadrants, and is well suited to applications that involve searches [Samet, 1990]. ...
The author presents in this paper a new approach for indexing and Content-based image retrieval based on the Quad-tree structure. The 3D objects are represented by their silhouettes and codified following the filling rate of each quadrant at different levels of the quad-tree subdivision. The author proposes a modified linear codification for silhouettes, this method improves the processing time because, in opposite to the traditional algorithms, the author’s algorithm has not a processing time that is proportional to the number of pixels in the image. As the same descriptor may characterize a set of different shapes, the author proposes also, efficient similarity measures to distinguish different objects having the same index in order to apply the approach to the retrieval process.
L’optimisation des bases de données d’images est un domaine très vaste et qui nécessite beaucoup de travaux de recherche car le problème majeur de ce type de donnée est le volume d’information qu’occupe une image, Le volume de données nécessaire pour le stockage de ce type d‘informations est généralement très élevé ce qui nécessite l’utilisation d’algorithmes spécifiques de compression et de représentation de l’information de la manière la plus compacte possible. Parmi les bases de données d’images qu’on propose d’optimiser dans ce projet, ce sont les bases de données d’images des applications de la reconnaissance biométrique de l’iris de l’oeil humain, les caractéristiques de l’iris de l’oeil humain permettent d’identifier l’individu car chaque être humain possède son propre code d’iris, nous proposons dans ce projet d’utiliser un algorithme de compression de données qui permet de réduire au maximum la taille de la base de données tout en réduisant la perte d’informations, cet algorithme est la compression par ondelette ou wavelet (JPEG2000), en utilisant trois bases de données différentes et chaque base de données est défini par des caractéristiques et des spécificités particulières, ensuite, on évaluera les résultats obtenus lors de l’indexation de la base de données en terme de performances et en prenant en compte les critères, tel que l’espace mémoire occupé par la base de données avant et après traitements, le temps d’exécution des requêtes...etc.
The approach developed in this paper is about objects recognition. The method consists to recognize objects that are partially occulted. A modified alignment approach is also considered to make our approach invariant to rotation. The proposed technique consists to create first, an index containing codes associated to each image of the database. The index is created using two techniques; the Quad-tree technique and also a technique using the curvature information of the object. The curvature information is based on an alignment method. Our System deals with the occlusion. It is, also, robust to the rotation of the objects.
We propose a novel binary image representation algorithm using the non-symmetry and anti-packing model and the coordinate encoding procedure (NAMCEP). By taking some idiomatic standard binary images in the field of image processing as typical test objects, and by comparing our proposed NAMCEP representation with linear quadtree (LQT), binary tree (Bintree), non-symmetry and anti-packing model (NAM) with K-lines (NAMK), and NAM representations, we show that NAMCEP can not only reduce the average node, but also simultaneously improve the average compression. We also present a novel NAMCEP-based algorithm for area calculation and show experimentally that our algorithm offers significant improvements.
The increasingly available graphics processing units (GPU) hardware and the emerging general purpose computing on GPU (GPGPU) technologies provide an attractive solution to high-performance geospatial computing. In this study, we have proposed a parallel, primitive-based approach to quadtree construction by transforming a multidimensional geospatial computing problem into chaining a set of generic parallel primitives that are designed for one-dimensional (1D) arrays. The proposed approach is largely data-independent and can be efficiently implemented on GPGPUs. Experiments on 4096*4096 and 16384*16384 raster tiles have shown that the implementation can complete the quadtree constructions in 13.33 ms and 250.75 ms, respectively, on average on an NVidia GPU device. Compared with an optimized serial CPU implementation based on the traditional recursive depth-first search (DFS) tree traversal schema that requires 1191.87 ms on 4096*4096 raster tiles, a significant speedup of nearly 90X has been observed. The performance of the GPU-based implementation also suggests that an indexing rate in the order of more than one billion raster cells per second can be achieved on commodity GPU devices.
In this paper, we are going to develop a system of indexation and research for the anomalies of red blood cells. This system uses a local approach which takes into account the spatial localization of globules in the images. The method is based on the image represented by generalized quaternary tree. In this approach we have given to the expert the possibility of choosing the adequate descriptor (color histogram, color texture), for every level, by a simple click on this last one. Currently, our system, called diagnosis, is capable of determining two anomalies of color type which can affected blood cells. He also allows to show the results via an effective graphic interface, which is simple to use and allows for a certain interaction with the user.
In this paper, we propose a non-symmetry and anti-packing image representation model (NAM). NAM is a hierarchical image representation method and its aim is less data amount and faster operation. By taking a rectangle sub-pattern for example, we describe the idea of NAM. In this work, based on block-wise LS linear predictor method, a gray-coded bit-plane binary image NAM representation method is presented to compress the error image. The compression performance has been further improved. Experimental results show that our compression scheme is clearly better from data compression point of view than other methods.
The representation methods of the hierarchical data structures have been widely applied in many important fields, such as computer visualization, robotics, computer graphics, image processing, and pattern recognition. Since these methods put too much emphasis upon the symmetry of segmentation, they are not the optimal representation methods. In this paper, we present a Direct Non-symmetry and Anti-packing Model (DNAM) for color images. Also, we propose a novel algorithm of the DNAM for color images and analyze the data amount of the proposed algorithm. By taking a rectangle subpattern for example, we implement the proposed algorithm and make a comparison with the algorithm of the popular linear quadtree. The theoretical and experimental results presented in this paper show that our proposed algorithm can reduce the data storage much more effectively than that of the linear quadtree and it is a better method to represent color images.
The main attempt of this paper is to present a new methodology to model a generic low-level flight close to terrain, which guarantees terrain collision avoidance. Benefiting the advantages of high-speed computer technology, this method uses satellite elevation maps to generate so called 'quad-tree forms'. The latter is then used to find the optimal trajectories for low-level flights. The novelty of this approach, entitled the 'cost map', lies in the integration of aircraft dynamics into the segmented map. This procedure results in some near-optimal trajectories with respect to aircraft dynamics that could easily be used for minimization of flight path together with pilot effort. Different simulations show the method is applicable wherever there is no need for sharp changes in flight level. Nevertheless, the current method could be expanded to three dimensional terrain avoidance/terrain following flights
The main attempt of this paper is to present a new methodology to model a generic low-level flight close to terrain, which guarantees terrain collision avoidance. Benefiting the advantages of high-speed computer technology, this method uses satellite elevation maps to generate so-called "quad-tree forms." The latter is then used to find the optimal trajectories for low-level flights. The novelty of this approach, entitled the "cost map," lies in the integration of aircraft dynamics into the segmented map. This procedure results in some near-optimal trajectories with respect to aircraft dynamics that could easily be used for minimization of flight path together with pilot effort. Different simulations show the method is applicable wherever there is no need for sharp changes in flight level. Nevertheless, the current method could be expanded to three-dimensional terrain avoidance/terrain following flights.
In this paper, we propose a non-symmetry and anti-packing image representation model (NAM). NAM is a hierarchical image representation method and its aim is less data amount and faster operation. By taking a rectangle sub-pattern for example, we describe the idea of NAM. In addition, an approach for adaptive area histogram equalization image contrast enhancement based on a NAM image is presented. The contrast enhancement approach is designed to meet the NAM image representation and can be fast implemented. In this work, the contrast enhancement method combines dynamic range modification and adaptive area histogram equalization to improve the visualization of images. Complexity analysis and experimental results show that the NAM based algorithm for image contrast enhancement is faster and more effective than that based on matrix image.
Cet article présente une structure d’index pour la recherche d’images par le contenu, l’arbre QUIP (acronyme anglais pour Quadtree-based Index for Image Retrieval and Image Pattern search). Dans notre approche, chaque image de la base est représentée par un descrip- teur dit multi-niveau, qui stocke les descripteurs des quadrants de l’image, obtenus par une décomposition de l’image en arbre quaternaire. L’arbre QUIP permet de regrouper les images en clusters, en fonction de la similarité de leurs quadrants. Cette structure d’index permet non seulement des recherches globales d’images par le contenu, en appliquant un filtrage multi- niveau via l’arbre quaternaire, mais aussi des recherches d’images similaires par région. This article presents a quadtree-based data structure for effective indexing of images. An image is represented by a multi-level feature vector , computed by a recursive decomposition of the image into four quadrants and stored as a full fixed-depth balanced quadtree. A node of the quadtree stores a feature vector of the corresponding image quadrant. A more general quadtree-based structure called QUIP-tree (QUadtree-based Index for image retrieval and Pat- tern search) is used to index the multi-level feature vectors of the images and their quadrants. A QUIP-tree node is an entry to a set of clusters that groups similar quadrants according to some pre-defined distances. The QUIP-tree allows a multi-level filtering in content-based image retrieval as well as partial queries on images. oui
Although the image representation methods of the hierarchical data structures have many merits and applications, they put too much emphasis upon the symmetry of segmentation. Therefore, they are not the optimal representation methods. In this paper, we propose a novel gray image representation method by using the direct non-symmetry and anti-packing model with K-lines (DNAMK). Also, we present an algorithm of the DNAMK for gray images and analyze the storage structure and the total data amount of the algorithm. By comparing the representation algorithm of the DNAMK with those of the triangle non-symmetry and anti-packing model (TNAM) and the classic linear quadtree, the experimental results presented in this paper show that the former can greatly reduce the numbers of subpatterns or nodes and simultaneously save the data storage much more effectively than the latter, and therefore it is a better method to represent gray images.
In this paper, a non-symmetry and anti-packing image representation model (NAM) has been proposed. NAM is a hierarchical image representation method and it aims to provide faster operations and less storage requirement. By taking a rectangle sub-pattern, for example, we describe the idea of NAM and its encoding algorithm.In addition, an approach for adaptive area histogram equalization for image contrast enhancement based on a NAM image is presented. The contrast enhancement approach is designed to meet the NAM image representation and it can be duplicated with faster operation. The complexity analysis and the experimental results show that the NAM based algorithm for image contrast enhancement is faster and more effective than that based on matrix image.
Cet article présente une définition générale de distance de similarité, appelée Delta, entre images représentées par des arbres quaternaires. L’arbre quaternaire est une structure hiérarchique qui, lorsqu’elle est utilisée pour la recherche d’images par le contenu, permet de tenir compte de la localisation spatiale des caractéristiques d’image (couleur , texture, etc.). En fonction de la pondération des nœuds d’arbre quaternaire et de la distance choisie entre ces nœuds, lors du calcul de Delta, plusieurs distances entre arbres quaternaires ou plusieurs distances de similarité visuelle entre images peuvent être définies. Certaines mesures de similarité définies dans des articles scientifiques apparaissent comme des cas particuliers de la distance Delta. De nouvelles distances entre images peuvent également être définies à partir de la distance Delta. This article presents Delta-distance, a generalization of image similarity distances de- fined on images represented with a quadtree. A quadtree is a recursive partitioning structure which can be used to handle positional information of image features in content-based image retrieval. Using different weights on quadtree nodes and different distances between nodes, distances between quadtrees or surface similarity between images can be computed based on the general definition of Delta. We show how existing quadtree-based image distances appear to be particular cases of the Delta-distance. Moreover , new distances between images can be defined based on the Delta-distance definition. oui
Less data and faster operation are the aims of image representation. The popular matrix image representation needs huge storage space and its processing speed is usually slow because of its operation pattern of pixel to pixel. Some traditional image representations with hiberarchy (e.g. quadtree) have operation pattern of block to block and can enhance some processing speed, but they can’t get the best segment of an image because of its symmetry. With the concept of packing problem, a non-symmetry anti-packing pattern representation model (NAM) is presented. NAM has less limitation when representing an image, so it can acquire higher compression ratio and do some image processing directly and more quickly. Comparison between rectangle NAM and other image representations proves that the rectangle NAM is an efficient representation at the aspect of data storage, and the experimental result of contour extraction of binary image represented by rectangle NAM proves that the rectangle NAM has the advantage of some image processing speed.
Although the representation methods of the quadtree for the multi-valued image have many important applications in different fields, they put too much emphasis upon the symmetry of image segmentation. Therefore, they are not the optimal representation methods. In this paper, inspired by the concept of the packing problem, we present a novel algorithm of binary-bit plane decomposition based on the non-symmetry and anti-packing pattern representation model (BPD-based NAM) for the multi-valued image representation. The theoretical analyses and experimental results presented in this paper show that when the algorithm of BPD-based NAM is compared with that of the popular quadtree, the former can reduce the data storage much more effectively than the latter and it is a better method to represent the multi-valued image. This novel algorithm of BPD-based NAM opens a new research area with respect to the multi-valued image representation and is valuable for the further theoretical research and potential business foreground.
With the extension of spatial database applications, field oriented systems emerge as an important research issue in order to deal with continuous natural phenomena during the last years. It however has a large volume of data and efficient indexing methods for field data are necessary to overcome the performance obstacle. In special, we introduce indexing methods for field value queries (i.e. searching some regions where the temperature is more 20 degrees). We introduce the concept of subfield and show how we make use of this concept to index field values in field oriented systems. We present two implementation methods based on Quadtree space subdivision. We modify traditional linear quadtree implementation method for field value query processing using subfields. We analyze the performance of our methods. Experimentation with real terrain data shows that proposed indexing methods improve the query processing time of field value queries in comparison with the case of no indexing method.
Overlapping is a technique used in access methods to com- biie consecutive structure instances into a single structure by not storing identical sub-stmctnres. This way, space is saved without sacrificing time performance. Here, we present the structure of Overlapping Linear Quadtrees -which is used to store consecutive rader images according to trans- action tim~ Experimentation with synthetic region data shows that considerable storage is saved in comparison to independent linear quadtre~~, in the case of similar consecu- tive images. Therefore, this stmcture can be used in spati~ temporal databases to support. query processing of evolv- ing images. Besides, an efficient aIgorithm that uses the new structure and answers spatio-temporal window queries is presented.
In this paper a system called CLIMS (CLausthal Image Management System) for content based image retrieval as an important subsystem of a general multimedia database is presented. It offers querying
by sketch and image example and uses colour and wavelet based features for the comparison of images. Each image in the database
is represented by a set of wavelet coefficients and colour attributes, which form the fundament for the retrieval.
In order to enable efficient similarity search two index structures, VP-Trees and Lq metric, are introduced and discussed. With the extension of the original VP-tree algorithm a ranking of the n most similar images is possible. The efficiency of the proposed retrieval methods is evaluated on a sample, general image
catalogue.
In this paper, indexing in spatio-temporal databases by using the technique of overlapping is investigated. Overlapping has been previously applied in various access methods to combine consecutive structure instances into a single structure, without storing identical sub-structures. In this way, space is saved without sacrificing time performance. A new access method, overlapping linear quadtrees is introduced. This structure is able to store consecutive historical raster images, a database of evolving images. Moreover, it can be used to support query processing in such a database. Five such spatio-temporal queries along with the respective algorithms that take advantage of the properties of the new structure are introduced. The new access method was implemented and extensive experimental studies for space efficiency and query processing performance were conducted. A number of results of these experiments are presented. As far as space is concerned, these results indicate that, in the case of similar consecutive images, considerable storage is saved in comparison to independent linear quadtrees. In the case of query processing, the results indicate that the proposed algorithmic approaches outperform the respective straightforward algorithms, in most cases. The region data sets used in experiments were real images of meteorological satellite views and synthetic random images with specified aggregation.
This paper describes QUILT, a prototype geographic information system (GIS) that uses the quadtree data structure as the underlying representation for cartographic data. While QUILT contains many features typically available in a GIS, its primary purpose is to serve as a testbed for the design and testing of new data structures and algorithms for use in computer cartography. Quadtree variants for region, point and line data are implemented using the linear quadtree, organized on disk by a B-tree. QUILT provides a simple attribute attachment system which associates non-spatial data with geographic objects. The user views QUILT as an augmented LISP environment. QUILT's geographic functions include conversion of rasters to and from quadtrees; subset operations to select specified geographic objects; map editing, display, windowing, intersection and union operations; polygon expansion; and computation of geographic object properties such as the centroid, area, perimeter and bounding rectangle for sets of geographic objects.
A quadtree may be represented without pointers by encoding each black node with a quaternary integer whose digits reflect successive quadrant subdivisions. We refer to the sorted array of black nodes as the “linear quadtree” and show that it introduces a saving of at least 66 percent of the computer storage required by regular quadtrees. Some algorithms using linear quadtrees are presented, namely, (i) encoding a pixel from a 2n × 2>n array (or screen) into its quaternary code; (ii) finding adjacent nodes; (iii) determining the color of a node; (iv) superposing two images. It is shown that algorithms (i)-(iii) can be executed in logarithmic time, while superposition can be carried out in linear time with respect to the total number of black nodes. The paper also shows that the dynamic capability of a quadtree can be effectively simulated.
The quad tree is a data structure appropriate for storing information to be retrieved on composite keys. We discuss the specific case of two-dimensional retrieval, although the structure is easily generalised to arbitrary dimensions. Algorithms are given both for staightforward insertion and for a type of balanced insertion into quad trees. Empirical analyses show that the average time for insertion is logarithmic with the tree size. An algorithm for retrieval within regions is presented along with data from empirical studies which imply that searching is reasonably efficient. We define an optimized tree and present an algorithm to accomplish optimization in n log n time. Searching is guaranteed to be fast in optimized trees. Remaining problems include those of deletion from quad trees and merging of quad trees, which seem to be inherently difficult operations.
We propose multi-precision similarity matching where the image is divided into a number of subblocks, each with its associated color histogram. We present experimental results showing that the spatial distribution information recorded by multiprecision color histograms helps to make similarity matching more precise. We also show that sub-image queries are much better supported with multi-precision color histograms. To minimize the overhead, we employ a filtering scheme based on the 3-dimensional average color vectors. We provide a formal result proving that filtering with multi-precision color histograms is complete. Finally, we develop a novel extendible hashing structure for indexing the average color vectors. We give experimental results showing that the proposed structure significantly outperforms the SR-tree.
We propose a method to handle approximate searching by image
content in medical image databases. Image content is represented by
attributed relational graphs holding features of objects and
relationships between objects. The method relies on the assumption that
a fixed number of “labeled” or “expected”
objects (e.g., “heart”, “lungs”, etc.) are
common in all images of a given application domain in addition to a
variable number of “unexpected” or “unlabeled”
objects (e.g., “tumor”, “hematoma”, etc.). The
method can answer queries by example, such as “find all X-rays
that are similar to Smith's X-ray”. The stored images are mapped
to points in a multidimensional space and are indexed using
state-of-the-art database methods (R-trees). The proposed method has
several desirable properties: (a) Database search is approximate, so
that all images up to a prespecified degree of similarity (tolerance)
are retrieved. (b) It has no “false dismissals” (i.e., all
images qualifying query selection criteria are retrieved). (c) It is
much faster than sequential scanning for searching in the main memory
and on the disk (i.e., by up to an order of magnitude), thus scaling-up
well for large databases
Research on ways to extend and improve query methods for image
databases is widespread. We have developed the QBIC (Query by Image
Content) system to explore content-based retrieval methods. QBIC allows
queries on large image and video databases based on example images,
user-constructed sketches and drawings, selected color and texture
patterns, camera and object motion, and other graphical information. Two
key properties of QBIC are (1) its use of image and video
content-computable properties of color, texture, shape and motion of
images, videos and their objects-in the queries, and (2) its graphical
query language, in which queries are posed by drawing, selecting and
other graphical means. This article describes the QBIC system and
demonstrates its query capabilities. QBIC technology is part of several
IBM products
Current retrieval methods in geographic image databases use only pixel-by-pixel spectral information. Texture is an important property of geographical images that can improve retrieval effectiveness and efficiency. In this paper, we present a content-based retrieval approach that utilizes the texture features of geographical images. Various texture features are extracted using wavelet transforms. Based on the texture features, we design a hierarchical approach to cluster geographical images for effective and efficient retrieval, measuring distances between feature vectors in the feature space. Using wavelet-based multi-resolution decomposition, two different sets of texture features are formulated for clustering. For each feature set, different distance measurement techniques are designed and experimented for clustering images in a database. The experimental results demonstrate that the retrieval efficiency and effectiveness improve when our clustering approach is used. Keywords: geogr...
Search operations in databases require some special support at the physical level. This is true for conventional databases as well as for spatial databases, where typical search operations include the point query (find all objects that contain a given search point) and the region query (find all objects that overlap a given search region). More than ten years of spatial database research have resulted in a great variety of multidimensional access methods to support such operations. This paper gives an overview of that work. After a brief survey of spatial data management in general, we first present the class of point access methods, which are used to search sets of points in two or more dimensions. The second part of the paper is devoted to spatial access methods, which are able to manage extended objects (such as rectangles or polyhedra). We conclude with a discussion of theoretical and experimental results concerning the relative performance of the various approaches. Ke...
Numerical images are becoming more and more important and an increasing emphasis on multimedia applications has resulted in large volumes of images. However, images need a large memory space to be stored, so their efficient storage and retrieval generate challenges to the database community. This paper proposes a new algorithm for an efficient storage of sets of images. It is based on a version approach used in databases. It shows how to store and operate on similar images; two images are defined as similar if the quad-trees encoding them have only few different nodes. A data structure called Generic Quad-Tree (GQT) is proposed. It optimizes the memory space required to store similar images and allows an efficient navigation among them. An Image Tree stores the ancestors and descendants of an image, like a version hierarchy. Using the Image Tree, the Generic Quad-Tree allows an image to share common parts with its ancestors and descendants. The GQT approach and some algorithms for reading, modifying or removing images from the Generic Quad-Tree are described. Examples using black and white images and gray scale images are presented. Keywords: Generic Quad-Tree, operations on quad-trees, image representation, image comparison, optimization of memory space, image processing application.
Processing a spatial query leads to the execution of complex and costly geometric operations. For common operations such as point queries, sequentially scanning and checking whether each object of a large collection contains a point involve a large number of disk accesses and the repeated expensive evaluation of geometric predicates. Hence, both the time-consuming geometric algorithms and the large volume of spatial collections stored on secondary storage motivate the design of efficient spatial access methods (SAMs), which reduce the set of objects to be processed. This chapter discusses the design and usage of SAMs. With a SAM, one expects search time that is logarithmic in the collection size, or even smaller. In most cases, a SAM uses an explicit structure called a spatial index. The collection of objects for which an index is built is said to be indexed. The chapter focuses on objects whose spatial component is defined in the 2-D plane; thus, it describes the indexing of lines and polygons. The chapter presents two families of indices representative of two major trends: the grid and the R-tree. These structures are simple, robust, and efficient, at least on well-behaved data. For each family, the chapter illustrates the process of index construction and search operations, where the search space is assumed to be a rectangular subset of the 2-D plane with sides parallel to the x and y axes.
The fundamentals of hierarchical data structures are reviewed and it is shown how they are used in the implementation of some basic operations in computer graphics. The properties of hierarchical structures are discussed, focusing on quadtrees and octrees. The latter are defined, some of the more common ways in which they are implemented are examined, and an explanation of the quadtree/octree complexity theorem is provided. Vector quadtrees and vector octrees are discussed. The performance of basic operations using quadtrees is considered.<
></ETX
A B+-tree structure for large quadtrees is presented. This is based on a form of key for a node derived from the ordered list of its ancestors, which facilitate tree traversal. The structure is a list of key values for black nodes which may be accessed in ascending order by key and directly by key. Algorithms for tree traversal and for examination of adjacent nodes are given. Analysis of space requirements and solution times suggests the structure offers advantages in applications involving large quadtrees where use of auxiliary storage is required.
In this paper, we focus on the problem of content-oriented retrieval in an image database. This problem can be interpreted as the search and the selection of images containing whichever pattern introduced beforehand by the user. We propose a data structure valid for this kind of manipulation, the Full Inverted Quadtree (FI-Quadtree). The structure represents a set of images within a single quadtree. We analyze the distribution of the data in the base, the operations of insertion, selection and we report some experimental results.
A tutorial survey is presented of the quadtree and related hierarchical data structures. They are based on the principle of recursive decomposition. The emphasis is on the representation of data used in applications in image processing, computer graphics, geographic information systems, and robotics. There is a greater emphasis on region data (ie, two-dimensional shapes) and to a lesser extent on point, curvilinear, and three-dimensional data. A number of operations in which such data structures find use are examined in greater detail.-Author
A method of representing a binary pictorial pattern is developed. Its original idea comes from a sequence of terminal symbols of a context-free grammar. It is a promising technique of data compression for ordinary binary-valued pictures such as texts, documents, charts, etc. Fundamental notions like complexity, primitives, simplifications, and other items about binary-valued pictures are introduced at the beginning. A simple context-free grammar G is also introduced. It is shown that every binary-valued picture is interpretable as a terminal sequence of that G. The DF-expression is defined as the reduced terminal sequence of G. It represents the original picture in every detail and contains no surplus data for reproducing it. A quantitative discussion about the total data of a DF-expression leads to the conclusion that any binary-valued picture with complexity less than 0.47 is expressed by the DF-expression with fewer data than the original ones. The coding algorithm of original data into the DF-expression is developed. It is very simple and recursively executable. Experiments were carried out using a PDS (photo digitizing system), where test pictures were texts, charts, diagrams, etc. with 20 cm à 20 cm size. Data compression techniques in facsimile were also simulated on the same test pictures. Throughout these studies it was made clear that the DF-expression is a very effective technique as a data compression for binary pictorial patterns not only because it yields high data compression but also because its coding and decoding algorithms are very feasible.
In this paper, a new and compact improved quadtree (compact-IQ) representation is presented for representing binary images. Then, efficient set operations on compact-IQs are also presented. Some experimentations are carried out to demonstrate the memory-efficiency and computational advantages of the proposed method. The experimental results reveal that the proposed image representation has 11.04–18.80% compression improvement and 24.39–36.94% computation-time improvement for set operations when compared to the recently published method on the constant bit-length linear quadtrees (CBLQ) (T.W. Lin, Set operations on constant bit-length linear quadtrees, Pattern Recognition 30(7) (1997) 1239–1249). In addition, geometric operations (area and centroid) on the compact-IQ and the performance comparison with JBIG are also investigated.
Based on linear quadtree structures and overlapping concepts, a new approach is proposed for storing a sequence of similar binary images. The space reduction of the method from linear quadtrees to overlapped linear quadtrees is equal to the overlapping percentage of images. We make a modification to the representation to provide bidirectional image retrieval. A simple, systematic and efficient method is devised for forward and backward retrieval to obtain the desired images quickly. The new representations are efficient and effective for a practical sstem in the fields of animation and geographic information systems.
The constant Bit-length Linear Quadtree (CBLQ) is a new encoding scheme for representing binary images. The representation keeps the level of detail property, and it makes the CBLQ suitable for interactive application systems. Based on the CBLQ representation, action tables for set operations are derived. A new algorithm using these derived tables is proposed. The time complexity of the algorithm is linear. A control scheme is also presented to keep the level of detail property in the algorithm. Therefore the algorithm can be used in the interactive application systems
This paper is a study of persistence in data structures. Ordinary data structures are ephemeral in the sense that a change to the structure destroys the old version, leaving only the new version available for use. In contrast, a persistent structure allows access to any version, old or new, at any time. We develop simple, systematic, and efficient techniques for making linked data structures persistent. We use our techniques to devise persistent forms of binary search trees with logarithmic access, insertion, and deletion times and O(1) space bounds for insertion and deletion.
We present a new structure, able to support and index a pictorial database. It is a variation of the region quadtree, termed Dynamic Inverted Quadtree, which is suitable for answering queries based on image content (exact and fuzzy image pattern searching). It exhibits certain advantages over the Fully Inverted Quadtree, an analogous structure that can be used for the same purposes. On the average it requires far less disk space; image pattern searching is performed efficiently; it is dynamic; reorganization is not obligatory, it can take place at any status of the structure and can be done on-line. Apart from the presentation of the new structure, we describe the pattern searching algorithms and we analyze the average space of the structure. Moreover, we compare analytically the space needs of Fully and Dynamic Inverted Quadtrees, as well as two different storage order methods that may be used in these structures. Analysis is based on a popular model of image randomness, which we express formally as a branching process.
In this paper we propose a technique for segmenting images by texture content with application to indexing images in a large image database. Using quad-tree decomposition, texture features are extracted from spatial blocks at a hierarchy of scales in each image. The quad-tree is grown by iteratively testing conditions for splitting parent blocks based on texture content of children blocks. While this approach does not achieve smooth identification of texture region borders, homogeneous blocks of texture are extracted which can be used in a database index. Furthermore, this technique performs the segmentation directly using image spatial-frequency data. In the segmentation reported here, texture features are extracted from the wavelet representation of the image. This method however, can use other subband decompositions including Discrete Cosine Transform (DCT), which has been adopted by the JPEG standard for image coding. This makes our segmentation method extremely applicable to databases containing compressed image data. We show application of the texture segmentation towards providing a new method for searching for images in large image databases using “Query-by-texture.”
Content-based image retrieval is today ubiquitous in computer vision. Most systems use the query-by-example approach, performing queries such as "show me more images that look like this one". Most often, the user is more specifically interested in specifying an object (or region) and in retrieving more images with similar objects (or regions), as opposed to similar images as a whole. This paper deals with that problem, called region querying. We suggest a method that uses a multiresolution quadtree representation of the images and thus avoids the hard problem of region segmentation. Several experimental results are presented in real-world databases.
Images are becoming an important asset and managing them for efficient retrieval poses challenges to the database community. In this paper, we proposed a novel three-tier color index that supports efficient image retrieval by color contents which is important to certain applications, especially when shapes and semantic objects cannot be easily recognized. A prototype painting database system is designed and implemented to demonstrate the effectiveness of the proposed indexing technique. Besides the color index, two other indexes, B
+-trees for structured attributes and a signature file for free-text descriptions, were also implemented. As a result, a wide range of queries, both text-based and content-based can be processed efficiently. We also look at existing image database systems based on their query retrieval capabilities.
The S-tree is an efficient spatial data structure for representing binary images. However, for some image operations, such as set, query, and neighbour-finding operations, a linear scan through the entire representation is required. This paper first presents a modified compact S-tree (MCST) for representing binary images. Then, some efficient set, query, and neighbour-finding operations on the MCST are presented. Some experimental results are carried out to demonstrate the computational advantage of the proposed method.
Overlapping has been proposed for several tree structures. The adaptation of this technique to region quadtrees, along with the necessary algorithms, are presented. The resulting structure is a form of a partial persistent quadtree and can be used to represent sequences of similar binary raster images emerging in various applications such as computer graphics, image processing, geographic information systems, or even satellite pictures. Our motive is to save considerable space, while access time of any of the similar images remains unaffected. Experimentation on random image forms exhibits very promising results in the space reduction achieved.
Encoding digital binary images by using bincodes is very simple and storage-saving. Many efficient algorithms have been developed for manipulating images represented by bincodes. Among these algorithms, however, some special treatments on individual bits within bincodes are often needed. This strongly increases computational complexity. In this paper, we first transform the bincode into a new logical representation. Then, we apply some logical operations on these logical expressions to manipulate images efficiently. Finally, nine basic image operations are implemented in hardware which can be viewed as a special-purpose processor to speed up these operations in order to meet the real-time imaging demand. Experimental results are illustrated to demonstrate the good performance of our hardware implementation.
When repeatedly editing a text file, one is often faced with a choice of keeping previous generations for backup or deleting previous generations to reduce storage requirements. Since one generation of a text file is often very similar to the previous generation, the above conflict can often be resolved by sharing much of the common information. We propose using a tree structure to represent a text file. Common subtrees can be shared. Results of an experiment with one file are reported.
The need to represent a collection of files having similar content in an efficient way leads to the concept of overlapping B-trees. An algorithm for the manipulation of overlapping B-trees is presented.
An efficient storage structure is proposed for quadtree-encoded images, and this structure is used in algorithms that develop efficient and practical quadtree-to-raster conversion methods.
Quadtrees appear attractive as an alternative to fixed-cell-size representations for areal entities in geographic information systems. This paper considers some aspects of use of linear quadtrees in such applications. Compaction (reduction of a quadtree to its maximal leaves) is desirable to minimise space requirements and to simplify some basic operations. An algorithm for compaction as an adjunct to quadtree-generating processes is presented, with analysis showing it to be linear with the sizes of the input and compacted quadtrees. Some simple forms of set operations are also described.
We describe a numbering scheme for versions with alternatives that has a useful lexicographical ordering. The version hierarchy is a tree. By inspection of the version numbers, we can easily determine whether one version is an ancestor of another. If so, we can determine the version sequence between these two versions. If not, we can determine the most recent common ancestor to these two versions (i.e., the least upper bound, lub). Sorting the version numbers lexicographically results in a version being followed by all descendants and preceded by all its ancestors. We use a representation of nonnegative integers that is self delimiting and whose lexicographical ordering matches the ordering by value
A structure based on B+trees is proposed, and some procedures for
manipulating it are given. Analytical estimates of the extra space
needed each time some records are updated are obtained. Experimental
results that agree with the analytical results are reported. Overlapping
structures can be used for backup and recovery, for storing variations
of a standard file or text files, and for temporal databases
this paper, we restrict our attention to geographic information systems (GIS). Such systems use as a kernel a geographic database management system (DBMS). Data handled by geographic DBMS has structured attributes as well as spatial attributes that describe the geometry and the topology of geographic objects. A fast access to such data, the volume of which is usually extremely large, requires the use of indices. Indices as provided by regular DBMS, e.g., B-tree like structures [BM72] are inefficient when the query is related to the location (geometry) of the objects. One major issue related to the choice of such indices, central to the design of geographic DBMS is the variety of geometric queries: one might argue that there is no index which performs best whatever the operation is. However we believe that all commercial systems will eventually provide a single simple-to-be-implemented data structure, although it might lead to uneven results depending on the query. This is the case with relational database technology: most of the commercial relational DBMS provide a unique B-tree-like index data structure. The objective of this paper is to contribute to the choice of such a single index. The problem is quite challenging for two reasons: 1. there is no standard yet concerning the query language and therefore there is no agreement on which are the most important spatial operations which require the use of a spatial index for performance reasons. Geometric selection, clipping and map overlay are among the most common operations. Examples of geometric selection are point queries (find the objects that contain a given point) and window queries (objects overlapping, i.e., intersecting or containing a given rectangle). In this paper, we restrict our attention to point queries. ...
Objets Similaires de Grande Taille dans les Bases de Donne Âes
- M Manouvrier
M. Manouvrier, Objets Similaires de Grande Taille dans les Bases de Donne Âes, PhD Thesis, Paris IX Dauphine Univ. (France), January, 2000, http://www.lamsade.dauphine.fr/~manouvri/thesis_ subject.html.
Indexing ®eld values in ®eld oriented systems: interval quadtree
- R Li
- Laurini
Li, R. Laurini, Indexing ®eld values in ®eld oriented systems: interval quadtree, International Conference on Information Knowledge Management (CIKM), Kansas City, 1999.