Mansur R. Kabuka

Ontology matching with semantic verification

ASMOV (Automated Semantic Matching of Ontologies with Verification) is a novel algorithm that uses lexical and structural characteristics of two ontologies to iteratively calculate a similarity measure between them, derives an alignment, and then verifies it to ensure that it does not contain semantic inconsistencies. In this paper, we describe the ASMOV algorithm, and then present experimental results that measure its accuracy using the OAEI 2008 tests, and that evaluate its use with two different thesauri: WordNet, and the Unified Medical Language System (UMLS). These results show the increased accuracy obtained by combining lexical, structural and extensional matchers with semantic verification, and demonstrate the advantage of using a domain-specific thesaurus for the alignment of specialized ontologies.

Position verification of a mobile robot using standard pattern

As mobile robots are taking on more and more of the tasks that were normally delegated to humans, they need to acquire higher degrees of autonomous operation, which calls for accurate and efficient position determination and/or verification. The critical geometric dimensions of a standard pattern are used here to locate the relative position of the mobile robot with respect to the pattern; by doing so, the method does not depend on values of any intrinsic camera parameters, except the focal length. In addition, this method has the advantages of simplicity and flexibility. This standard pattern is also provided with a unique identification code, using bar codes, that enables the system to find the absolute location of the pattern. These bar codes also assist in the scanning algorithms to locate the pattern in the environment. A thorough error analysis and experimental results obtained through software simulation are presented, as well as the current direction of our work.

Edge detection in medical images using a genetic algorithm

An algorithm is developed that detects well-localized, unfragmented, thin edges in medical images based on optimization of edge configurations using a genetic algorithm (GA). Several enhancements were added to improve the performance of the algorithm over a traditional GA. The edge map is split into connected subregions to reduce the solution space and simplify the problem. The edge-map is then optimized in parallel using incorporated genetic operators that perform transforms on edge structures. Adaptation is used to control operator probabilities based on their participation. The GA was compared to the simulated annealing (SA) approach using ideal and actual medical images from different modalities including magnetic resonance imaging (MRI), computed tomography (CT), and ultrasound. Quantitative comparisons were provided based on the Pratt figure of merit and on the cost-function minimization. The detected edges were thin, continuous, and well localized. Most of the basic edge features were detected, Results for different medical image modalities are promising and encourage further investigation to improve the accuracy and experiment with different cost functions and genetic operators.

A robust framework for content-based retrieval by spatial similarity in image databases

A framework for retrieving images by spatial similarity (FRISS) in ima ge databases is presented. In this framework, a robust retrieval by spatial similarity (RSS) algorithm is defined as one that incorporates both directional and topological spatial constraints, retrieves similar images, and recognized images even after they undergo translation, scaling, rotation (both perfect and multiple), or any arbitrary combination of transformatioins. The FRISS framework is discussed and used as a base for comparing various existing RSS algorithms. Analysis shows that none of them satisfies all the FRISS specifications. An algorithm, <italic>SIM<subscrpt>dtc</subscrpt></italic>, is then presented. <italic>SIM<subscrpt>dtc</subscrpt></italic> introduces the concept of a <italic>rotation correction angle</italic>(RCA) to align objects in one image spatially closer to matching objects in another image for more accurate similarity assessment. Similarity between two images is a function of the number of common objects between them and the closeness of directional and topological spatial relationships between object pairs in both images. The <italic>SIM<subscrpt>dtc</subscrpt></italic> retrieval is invariant under translation, scaling, and perfect rotation, and the algorithm is able to rank multiple rotation variants. The algorithm was tested using synthetic images and the TESSA image database. Analysis shows the robustness of the <italic>SIM<subscrpt>dtc</subscrpt></italic> algorithm over current algorithms.

A novel feature recognition neural network and its application to character recognition

Presents a feature recognition network for pattern recognition that learns the patterns by remembering their different segments. The base algorithm for this network is a Boolean net algorithm that the authors developed during past research. Simulation results show that the network can recognize patterns after significant noise, deformation, translation and even scaling. The network is compared to existing popular networks used for the same purpose, especially the Neocognitron. The network is also analyzed as regards to interconnection complexity and information storage/retrieval. >

Automatic moving object extraction for content-based applications

Rapid developments in the Internet and multimedia applications allow us to access large amounts of image and video data. While significant progress has been made in digital data compression, content-based functionalities are still quite limited. Many existing techniques in content-based retrieval are based on global visual features extracted from the entire image. In order to provide more efficient content-based functionalities for video applications, it is necessary to extract meaningful video objects from scenes to enable object-based representation of video content. Object-based representation is also introduced by MPEG-4 to enable content-based functionality and high coding efficiency. In this paper, we propose a new algorithm that automatically extracts meaningful video objects from video sequences. The algorithm begins with the robust motion segmentation on the first two successive frames. To detect moving objects, segmented regions are grouped together according to their spatial similarity. A binary object model for each moving object is automatically derived and tracked in subsequent frames using the generalized Hausdorff distance. The object model is updated for each frame to accommodate for complex motions and shape changes of the object. Experimental results using different types of video sequences are presented to demonstrate the efficiency and accuracy of our proposed algorithm.

Efficient content-based indexing of large image databases

Large image databases have emerged in various applications in recent years. A prime requisite of these databases is the means by which their contents can be indexed and retrieved. A multilevel signature file called the Two Signature Multi-level Signature File (2SMLSF) is introduced as an efficient access structure for large image databases. The 2SMLSF encodes image information into binary signatures and creates a tree structures can be efficiently searched to satisfy a users query. Two types of signatures are generated. Type I signatures are used at all tree levels except the leaf level and are based only on the domain objects included in the image. Type II signatures, on the other hand, are stored at the leaf level and are based on the included domain objects and their spatial relationships. The 2SMLSF was compared analytically to existing signature file techniques. The 2SMLSF significantly reduces the storage requirements; the index structure can answer more queries; and the 2SMLSF performance significantly improves over current techniques. Both storage reduction and performance improvement increase with the number of objects per image and the number of images in the database. For an example large image database, a storage reduction of 78% may be archieved while the performance improvement may reach 98%.

Hidden Markov models-based 3D MRI brain segmentation

Abstract This paper introduces a 3D MRI segmentation algorithm based on Hidden Markov Models (HMMs). The mathematical models for the HMM that forms the basis of the segmentation algorithm for both the continuous and discrete cases are developed and contrasted with Hidden Markov Random Field in terms of complexity and extensibility to larger fields. The presented algorithm clearly demonstrates the capacity of HMM to tackle multi-dimensional classification problems. The HMM-based segmentation algorithm was evaluated through application to simulated brain images from the McConnell Brain Imaging Centre, Montreal Neurological Institute, McGill University as well as real brain images from the Internet Brain Segmentation Repository (IBSR), Harvard University. The HMM model exhibited high accuracy in segmenting the simulated brain data and an even higher accuracy when compared to other techniques applied to the IBSR 3D MRI data sets. The achieved accuracy of the segmentation results is attributed to the HMM foundation and the utilization of the 3D model of the data. The IBSR 3D MRI data sets encompass various levels of difficulty and artifacts that were chosen to pose a wide range of challenges, which required handling of sudden intensity variations and the need for global intensity level correction and 3D anisotropic filtering. During segmentation, each class of MR tissue was assigned to a separate HMM and all of the models were trained using the discriminative MCE training algorithm. The results were numerically assessed and compared to those reported using other techniques applied to the same data sets, including manual segmentations establishing the ground truth for real MR brain data. The results obtained using the HMM-based algorithm were the closest to the manual segmentation ground truth in terms of an objective measure of overlap compared to other methods.

Design of supervised classifiers using Boolean neural networks

In this paper we present two supervised pattern classifiers designed using Boolean neural networks. They are: 1) nearest-to-an-exemplar classifier; and 2) Boolean k-nearest neighbor classifier. The emphasis during the design of these classifiers was on simplicity, robustness, and the ease of hardware implementation. The classifiers use the idea of radius of attraction to achieve their goal. Mathematical analysis of the algorithms presented in the paper is done to prove their feasibility. Both classifiers are tested with well-known binary and continuous feature valued data sets yielding results comparable with those obtained by similar existing classifiers.

Labeling of MR brain images using Boolean neural network

Presents a knowledge-based approach for labeling two-dimensional (2-D) magnetic resonance (MR) brain images using the Boolean neural network (BNN), which has binary inputs and outputs, integer weights, fast learning and classification, and guaranteed convergence. The approach consists of two components: a BNN clustering algorithm and a constraint satisfying Boolean neural network (CSBNN) labeling procedure. The BNN clustering algorithm is developed to initially segment an image into a number of regions. Then the segmented regions are labeled with the CSBNN, which is a modified version of BNN. The CSBNN uses a knowledge base that contains information on image-feature space and tissue models as constraints. The method is tested using sets of MR brain images. The regions of the different brain tissues are satisfactorily segmented and labeled. A comparison with the Hopfield neural network and the traditional simulated annealing method for image labeling is provided. The comparison results show that the CSBNN approach offers a fast, feasible, and reliable alternative to the existing techniques for medical image labeling.