Mandava Rajeswari

Review Article: Multi-objective nature-inspired clustering and classification techniques for image segmentation

This paper aims to provide a comprehensive review of nature-inspired techniques used in image segmentation problems. We focus particularly on multi-objective clustering and classification approaches. The approaches are classified based on the various aspects of optimization, various possible problem formulations, and types of datasets modeled. In the multi-objective clustering methods, the definition of the types of representation methods, encoding techniques, and number of clusters defined (fixed/variable) are presented. In the use of multi-objective nature-inspired techniques in classification, we describe issues related to diversity measures, accuracy measures, rule manipulation, and managing uncertainties. Through our analysis of the current state of research, we hope to address important challenges and provide specific directions for future modeling of similar problems with multi-objective optimization techniques.

Screening of Diabetic Retinopathy - Automatic Segmentation of Optic Disc in Colour Fundus Images

In this paper, a novel approach to automatically segment the optic disc contour using the center point of an optic disc candidate is proposed. The optic disc segmentation algorithm consists of 2 stages. The first stage involves the removal of blood vessels that obscure the optic disc. The blood vessel structures are detected using morphological operations. These detected structures are then removed by anisotropic diffusion smoothing. The second stage involves the detection of edge points belonging to the optic disc-contour. A number of one dimensional intensity profiles which pass through the center point of optic disc region are then obtained at multiple angles with fixed angular intervals. The modulus maxima of each intensity profile are identified as a contour point for the optic disc. Among these contour points, some of the outliers are removed by re-positioning to a new position which complies with optic discs shape using spline interpolation. Using these contour points, a coarse contour of the optic disc is constructed. Testing the approach on 23 colour fundus images demonstrates that the proposed algorithm is able to detect the optic disc contour to an accuracy of 92% to that drawn by a human expert

Wavelet energy-guided level set-based active contour: A segmentation method to segment highly similar regions

This paper introduces an approach to perform segmentation of regions in computed tomography (CT) images that exhibit intra-region intensity variations and at the same time have similar intensity distributions with surrounding/adjacent regions. In this work, we adapt a feature computed from wavelet transform called wavelet energy to represent the region information. The wavelet energy is embedded into a level set model to formulate the segmentation model called wavelet energy-guided level set-based active contour (WELSAC). The WELSAC model is evaluated using several synthetic and CT images focusing on tumour cases, which contain regions demonstrating the characteristics of intra-region intensity variations and having high similarity in intensity distributions with the adjacent regions. The obtained results show that the proposed WELSAC model is able to segment regions of interest in close correspondence with the manual delineation provided by the medical experts and to provide a solution for tumour detection.

Extrapolation detection and novelty-based node insertion for sequential growing multi-experts network

Artificial neural networks (ANNs) have been used to construct empirical nonlinear models of process data. Because networks are not based on physical theory and contain nonlinearities, their predictions are suspect when extrapolating beyond the range of original training data. Standard networks give no indication of possible errors due to extrapolation. This paper describes a sequential supervised learning scheme for the recently formalized Growing multi-experts network (GMN). It is shown that certainty factor can be generated by GMN that can be taken as extrapolation detector for GMN. On-line GMN identification algorithm is presented and its performance is evaluated. The capability of the GMN to extrapolate is also indicated. Four benchmark experiments are dealt with to demonstrate the effectiveness and utility of GMN as a universal function approximator.

Neural network-based robot visual positioning for intelligent assembly

A fundamental task in robotic assembly is the pick and place operation. Generally, this operation consists of three subtasks; guiding the robot to the target and positioning the manipulator in an appropriate pose, picking up the object and moving the object to a new location. In situations where the pose of the target may vary in the workspace, sensory feedback becomes indispensable to guide the robot to the object. Ideally, local image features must be clearly visible and un-occluded in multiple views of the object. In reality, this may not be always the case. Local image features are often are often rigidly constrained to a particular target and may require specialized feature localization algorithms. We present a visual positioning system that addresses feature extraction issues for a class of objects that have smooth or curved surfaces. In this work, the visual sensor consists of an arm mounted camera and a grid pattern projector that produces images with local surface description of the target. The projected pattern is always visible in the image and it is sensitive to variations in the object’s pose. A set of low-order geometric moments globally characterizes the observed pattern, eliminating the need for feature localization and overcoming the point correspondence problem. A neural network then learns the complex relationship between the robot’s pose displacements and the observed variations in the image features. After training, visual feedback guides the robot to the target from any arbitrary location in the workspace. Its applicability using a five degrees of freedom (DOF) industrial robot is demonstrated.

A hybrid intelligent active force controller for articulated robot arms using dynamic structure network

The key feature of this paper is the application of a robotic control concept – Active Force Control (AFC). In this type of control, the unknown friction effect of the robotic arm may be compensated by the AFC method. AFC involves the direct measurement of the acceleration and force quantities and therefore, the process of estimating the system ‘disturbance’ due to friction becomes instantaneous and purely algebraic. However, the AFC strategy is very practical provided a good estimation of the inertia matrix of articulated robot arm is acquired. A dynamic structure neural network – Growing Multi-experts Network (GMN) is developed to estimate the robot inertia matrix. The growing and pruning mechanism of GMN ensures the optimum size of the network that results in an excellent generalization capability of the network. Active Force Control (AFC) in conjunction with GMN successfully reduces the velocity and position tracking errors in spite of robot joint friction. The embedded GMN is capable of coupling the inertia matrix estimation on-line that clearly enhances the performance of AFC controller. The robustness and effectiveness of the new hybrid neural network-based AFC scheme are demonstrated clearly with regard to two link articulated robot and a simulated two-degree of freedom Puma 560 robot.

MOBILE ROBOT PATH PLANNING USING HYBRID GENETIC ALGORITHM AND TRAVERSABILITY VECTORS METHOD

Abstract The shortest/optimal path generation is essential for the efficient operation of a mobile robot. Recent advances in robotics and machine intelligence have led to the application of modern optimization method such as the genetic algorithm (GA), to solve the path-planning problem. However, the genetic algorithm path planning approach in the previous works requires a preprocessing step that captures the connectivity of the free-space in a concise representation. In this paper, GA path-planning approach is enhanced with feasible path detection mechanism based on traversability vectors method. This novel idea eliminates the need of free-space connectivity representation. The feasible path detection is performed concurrently while the GA performs the search for the shortest path. The performance of the proposed GA approach is tested on three different environments consisting of polygonal obstacles with increasing complexity. In all experiments, the GA has successfully detected the near-optimal feasible...

Novel direct and self-regulating approaches to determine optimum growing multi-experts network structure

This work presents two novel approaches to determine optimum growing multi-experts network (GMN) structure. The first method called direct method deals with expertise domain and levels in connection with local experts. The growing neural gas (GNG) algorithm is used to cluster the local experts. The concept of error distribution is used to apportion error among the local experts. After reaching the specified size of the network, redundant experts removal algorithm is invoked to prune the size of the network based on the ranking of the experts. However, GMN is not ergonomic due to too many network control parameters. Therefore, a self-regulating GMN (SGMN) algorithm is proposed. SGMN adopts self-adaptive learning rates for gradient-descent learning rules. In addition, SGMN adopts a more rigorous clustering method called fully self-organized simplified adaptive resonance theory in a modified form. Experimental results show SGMN obtains comparative or even better performance than GMN in four benchmark examples, with reduced sensitivity to learning parameters setting. Moreover, both GMN and SGMN outperform the other neural networks and statistical models. The efficacy of SGMN is further justified in three industrial applications and a control problem. It provides consistent results besides holding out a profound potential and promise for building a novel type of nonlinear model consisting of several local linear models.

Using Correlation Based Subspace Clustering for Multi-label Text Data Classification

With the boom of web and social networking, the amount of generated text data has increased enormously. Much of this data can be considered and modeled as a stream and the volume of such data necessitates the application of automated text classification strategies. Although streaming data classification is not new, considering text data streams for classification purposes has been extensively researched only recently. Before applying any classification method in text data streams, it is imperative that we apply them for existing well-known non-stream text data sets and evaluate their performance. One of the many characteristics of text data that has been pursued for research is its multi-labelity. A single text document may cover multiple class-labels at the same time and hence gives rise to the concept of multi-labelity. From classification perspective, an immediate drawback of such a characteristic is that traditional binary or multi-class classification techniques perform poorly on multi-label text data. In this paper, we extend our previously formulated SISC (Semi-supervised Impurity based Subspace Clustering) [1] approach and its multi-label variation SISC-ML [2]. We call this new algorithm H-SISC (Hierarchical SISC). H-SISC captures the underlying correlation that exists between each pair of class labels in a multi-label environment. Developing a robust multi-label classifier will allow us to apply such a model in classifying streaming text data more effectively. We have experimented with well known text data sets and empirical evaluation on these real world multi-label NASA ASRS (Aviation Safety Reporting System), Reuters and 20 Newsgroups data sets reveals that our proposed approach outperforms other state-of-the-art text classification as well as subspace clustering algorithms.

A Web-Based Framework for Distributed Medical Image Processing Using Image Markup Language (IML)

Image processing plays an important role in computer science, making complex image manipulation more feasible. It can be used either in a general manner or in specific domains such as for medical purposes. With the availability of the internet, image processing applications can be distributed to be available for different people, regardless of their geographical location. Although there are a handful of professional image processing solutions available, there is no well-designed framework that can be simply distributed and customized to be used in specific areas. Other than image processing tools, image storage and retrieval is another issue that needs to be addressed. This research proposes a general platform that can be used for implementing a distributed image processing framework. It provides solutions for storing and retrieving images, particularly large-size ones as well as an approach for executing image processing functions that are physically located anywhere within the distributed system.