Maytham Safar

Image retrieval by shape: a comparative study

Besides traditional applications (e.g. CAD/CAM and trademark registry), new multimedia applications, such as structured video, animation and the MPEG-7 standard, require the storage and management of well-defined objects. We focus on shape-based object retrieval and conduct a comparison study on four such techniques: Fourier descriptors (FD), the grid-based (GB) method, Delaunay triangulation (DT) and MBC-TPVAS (minimum bounding circles/touch-point vertex-angle sequence). Our results show that the similarity retrieval accuracy of our method (TVPAS) is as good as the other methods, while it has the lowest computational cost for generating the shape signatures of the objects. Moreover, it has low storage requirements and a comparable computation cost to compute the similarity between two shape signatures. In addition, TPVAS requires no normalization of the objects and is the only method that has direct support for RST (rotation/scaling/translation) query types. In this paper, we also introduce a new shape description taxonomy.

Voronoi-Based Continuous

Digital ecosystems are formed by “digital organisms” in complex, dynamic, and interrelated ecosystems and utilize multiple technologies to provide cost-efficient digital services and value-creating activities. A distributed wireless mobile network that serves as the underlying infrastructure to digital ecosystems provides important applications to the digital ecosystems, two of which are mobile navigation and continuous mobile information services. Most information and query services in a mobile environment are continuous mobile query processing or continuous k nearest neighbor (CKNN), which finds the locations where interest points or interest objects change while mobile users are moving. These locations are known as “split nodes.” All of the existing works on CKNN divide the query path into segments, which is a segment of road separated by two intersections, and then, the process to find split nodes is applied to each segment. Since there are many segments (due to many intersections, obviously), processing each segment is naturally inefficient. In this paper, we propose an alternative solution to overcome this problem. We use the Voronoi diagram for CKNN [called Voronoi CKNN (VCKNN)]. Our proposed approach does not need to divide the query path into segments, hence improving the overall query processing performance. Our experiment verified the applicability of the VCKNN approach to solve CKNN queries.

k

The use of Voronoi diagram has traditionally been applied to computational geometry and multimedia problems. In this paper, we will show how Voronoi diagram can be applied to spatial query processing, and in particular to Reverse Nearest Neighbor (RNN) queries. Spatial and geographical query processing, in general, and RNN in particular, are becoming more important, as online maps are now widely available. In this paper, using the concept of Voronoi diagram, we classify RNN into four types depending on whether the query point and the interest objects are the generator points of the Voronoi Polygon or not. Our approach is based on manipulating Network Voronoi Diagram properties and applying a progressive incremental network expansion for finding the polygon inner network distances required to solve RNN queries. Our experimentation results show that our approaches have good response times in answering RNN queries.

Nearest Neighbor Search in Mobile Navigation

A frequent type of query in a car navigation system is to find the k nearest neighbors (kNN) of a given query object (e.g., car) using the actual road network maps. With road networks (spatial networks), the distances between objects depend on their network connectivity and it is computationally expensive to compute the distances (e.g., shortest paths) between objects. In this paper, we propose a novel approach to efficiently and accurately evaluate kNN queries in a mobile information system that uses spatial network databases. The approach uses first order Voronoi diagram and Dijkstras algorithm. This approach is based on partitioning a large network to small Voronoi regions, and then pre-computing distances across the regions. By performing across the network computation for only the border points of the neighboring regions, we avoid global pre-computation between every object-pair. Our empirical experiments with real-world data sets show that our proposed solution outperforms approaches that are based on on-line distance computation by up to one order of magnitude. In addition, our approach has better response times than approaches that are based on pre-computation.

Voronoi-based reverse nearest neighbor query processing on spatial networks

With the wide availability of mobile devices (smart phones, iPhones, etc.), mobile location-based queries are increasingly in demand. One of the most frequent queries is range search which returns objects of interest within a pre-defined area. Most of the existing methods are based on the road network expansion method - expanding all nodes (intersections and objects) and computing the distance of each node to the query point. Since road networks are extremely complex, node expansion approaches are inefficient. In this paper, we propose a method, Voronoi Range Search (VRS) based on the Voronoi diagram, to process range search queries efficiently and accurately by partitioning the road networks to some special polygons. Then we further propose Voronoi Continuous Range (VCR) to satisfy the requirement for continuous range search queries (moving queries) based on VRS. Our empirical experiments show that VRS and VCR surpass all their rivals for both static and moving queries.

K nearest neighbor search in navigation systems

Due to the universality and importance of range search queries processing in mobile and spatial databases as well as in geographic information system (GIS), numerous approaches on range search algorithms have been proposed in recent years. But ordinary range search queries focus only on a specific type of point objects. For queries which require to retrieve objects of interest locating in a particular region, ordinary range search could not get the expected results. In addition, most existing range search methods need to perform a searching on each road segments within the pre-defined range, which decreases the performance of range search. In this paper, we design a weighted network Voronoi diagram and propose a high-performance multilevel range search query processing that retrieves a set of objects locating in some specified region within the searching range. The experimental results show that our proposed algorithm runs very efficiently and outperforms its main competitor.

Voronoi-based range and continuous range query processing in mobile databases

Geographical information systems (GIS) and applications assist us in commuting, traveling and locating our points of interests. The efficient implementation and support of spatial queries in those systems is of particular interest and importance. The use of a Voronoi diagram has traditionally been applied to computational geometry. In this paper, we will show how a Voronoi diagram can be applied to support spatial queries in GIS systems, and in particular to reverse nearest neighbor (RNN) queries. An RNN query retrieves the set of interest objects having the query object as the nearest neighbor among other objects. Two cases of RNN queries are: monochromatic (MRNN) and bichromatic (BRNN). In the MRNN, the interest objects and the query object are of the same type, whereas in the BRNN they are of two different types. Due to the shortcomings of solutions for BRNN in the literature, we develop a new approach and algorithm, named the ‘2Vor BRNN algorithm’, for processing this query type in the context of the spatial network database (SNDB). Our novel approach extends the previous work and uses the ‘order-2 network Voronoi diagram’ to provide a more efficient solution for the BRNN. In addition, we experimentally confirm that the proposed algorithm outperforms the previous one in terms of memory used and response time.

Voronoi-based multi-level range search in mobile navigation

One of the problems that arise in geographical information systems is finding objects that are influenced by other objects. While most research focuses on kNN (k Nearest Neighbor) and RNN (Reverse Nearest Neighbor) queries, an important type of proximity queries called Reverse Farthest Neighbor (RFN) has not received much attention. Since our previous work shows that kNN and RNN queries in spatial network databases can be efficiently solved using Network Voronoi Diagram (NVD), in this paper, we aim to introduce a new approach to process reverse proximity queries including RFN and RkNN/RkFN queries. Our approach is based on NVD and pre-computation of network distances, and is applicable for spatial road network maps. Being the most fundamental Voronoi-based approach for RFN and RkNN/RkFN queries, our solutions show that they can be efficiently used for networks that have a low and medium level of density.

Spatial Network RNN Queries in GIS

In traditional reverse nearest-neighbor (RNN) search, the goal is to find the set of interest objects taking the query object as the nearest neighbor given that the interest objects and the query object have the same type. However, when the interest objects and the query object are of different types, this problem is called bichromatic reverse nearest neighbor (BRNN) search. The BRNN query has an increasing number of mobile applications and requires efficient algorithms for processing. In this paper, we present a novel approach for the BRNN search in the context of spatial network databases (SNDB). The main idea behind our approaches is to use Euclidean-based range search to prune the search space and use Euclidean distance in addition to network distance to minimize processing time. Finally, the experimental results confirm that our proposed algorithm have good performance on different network densities.

Reverse k Nearest Neighbor and Reverse Farthest Neighbor Search on Spatial Networks

Besides traditional applications (e.g., CAD/CAM and Trademark registry), new multimedia applications such as structured video, animation, and MPEG-7 standard require the storage and management of well-defined objects. These object databases are then queried and searched for different purposes. A sample query might be “find all the scenes that contain a certain object.” Shape of an object is an important feature for image and multimedia similarity retrievals. Therefore, in this study we focus on shape-based object retrieval and conduct a comparison study on four of such techniques (i.e., Fourier descriptors, grid based, Delaunay triangulation, and our proposed MBC-based methods (e.g., MBC-TPVAS)). We measure the effectiveness of the similarity retrieval of the four different shape representation methods in terms of recall and precision. Our results show that the similarity retrieval accuracy of our method (MBC-TPVAS) is as good as that of the other methods, while it observes the lowest computation cost to generate the shape signatures of the objects. Moreover, it has low storage requirement, and a comparable computation cost to compute the similarity between two shape signatures. In addition, MBC-TPVAS requires no normalization of the objects, and is the only method that has direct support for S-RST query types. In this paper, we also propose a new shape description taxonomy.