K. Naveen Kumar

Bio: K. Naveen Kumar is an academic researcher from GITAM University Hyderabad Campus. The author has contributed to research in topics: Mixture model & Image segmentation. The author has an hindex of 1, co-authored 2 publications receiving 6 citations.

Studies on Texture Segmentation Using D- Dimensional Generalized Gaussian Distribution integrated with Hierarchical Clustering

Abstract: Texture deals with the visual properties of an image. Texture analysis plays a dominant role for image segmentation. In texture segmentation, model based methods are superior to model free methods with respect to segmentation methods. This paper addresses the application of multivariate generalized Gaussian mixture probability model for segmenting the texture of an image integrating with hierarchical clustering. Here the feature vector associated with the texture is derived through DCT coefficients of the image blocks. The model parameters are estimated using EM algorithm. The initialization of model parameters is done through hierarchical clustering algorithm and moment method of estimation. The texture segmentation algorithm is developed using component maximum likelihood under Bayesian frame. The performance of the proposed algorithm is carried through experimentation on five image textures selected randomly from the Brodatz texture database. The texture segmentation performance measures such as GCE, PRI and VOI have revealed that this method outperform over the existing methods of texture segmentation using Gaussian mixture model. This is also supported by computing confusion matrix, accuracy, specificity, sensitivity and F-measure.

Texture Segmentation based on Multivariate GeneralizedGaussian Mixture Model

Abstract: Texture Analysis is one of the prime considerations for image analysis and processing. Texture segmentation gained lot of importance due to its ready applicability in automation of scene identification and computer vision. Several texture segmentation methods have been developed and analysed with the assumption that the feature vector associated with the texture of the image region is modelled as Gaussian mixture model. Due to the limitations of the Gaussian model being meso kurtic, it may not characterise the texture of all image regions accurately. Hence in this paper, a texture segmentation algorithm is developed and analysed with the assumption that the feature vector of the texture associated with the whole image is characterised by multivariate generalized Gaussian mixture model. The generalized Gaussian mixture model includes several lepto kurtic, platy kurtic and meso kurtic distributions as particular cases. The model parameters are estimated through EM algorithm. The segmentation algorithm is developed using maximum likelihood under Bayesian framework. The performance of the proposed algorithm is evaluated through segmentation quality metrics and conducting experimentation with a set of 8 sample images taken from Brodatz texture database. A comparative study of the proposed algorithm with that of Gaussian mixture model revealed that the proposed algorithm outstandthe existing algorithms.

Texture-based image segmentation using neutrosophic clustering

Abstract: This study presents an effective segmentation method which is based on neutrosophic clustering with the integration of texture features for images. The proposed method transforms the image into the neutrosophic domain and then extracts the texture features using analogies of human preattentive texture discrimination mechanisms. Finally, the neutrosophic clustering is employed to segment the images. This method can handle the indeterminacy of pixels to have strong clusters and to perform segmentation effectively with the noisy images. Experiments are performed with various types of natural and medical images to exhibit the performance of proposed segmentation method. The evaluation of proposed method has been done with other segmentation methods to measure its performance which shows its robustness for noisy and textured images.

Unsupervised Human Action Categorization Using a Riemannian Averaged Fixed-Point Learning of Multivariate GGMM

Abstract: We present a novel learning algorithm for Human action recognition and categorization. Our purpose here is to develop a Riemannian Averaged Fixed-Point estimation algorithm (RA-FP) for learning the multivariate generalized Gaussian mixture model’s parameters (MGGMM). Experiments in a large datasets of human action images have shown the merits of our approach.

A Fixed-Point Estimation Algorithm for Learning the Multivariate GGMM: Application to Human Action Recognition

Abstract: Multivariate generalized Gaussian distribution has been an attractive solution to many signal and image processing applications. Therefore, efficient estimation of its parameters is of significant interest for a number of research problems. The main contribution of this paper is to develop a fixed-point estimation algorithm for learning the multivariate generalized Gaussian mixture model's parameters (MGGMM). A challenging application that concerns Human action recognition is deployed to validate our statistical framework and to show its merits.

A texture-based 3D region growing approach for segmentation of ICA through the skull base in CTA

Abstract: The internal carotid artery (ICA) segmentation is a complicated task at skull base in computed tomography angiography (CTA) images. The ICA enters into from skull cavity and shows close proximity to bone and surrounding soft tissues. For this reason, there exists a robust intensity overlap between vessels, bone and other surrounding tissues. Thus, these similar objects are not separated properly in images only according to the intensity level. In this paper, a texture-based 3D region growing approach is proposed and applied to the ICA through the skull base. The main contribution of this study is that the method does not ask for an extra computed tomography scan for bone masking. Moreover, the method dynamically sets the segmentation parameters according to texture knowledge. The proposed method was evaluated by the experiments on 15 actual clinical data. The performance evaluations were performed by comparing the automatic outputs with manual segmentations which are done by two radiologist observers. As a result, dice similarity rate of 89% was achieved together with 99% accuracy and 0.32 mm mean surface distance (Msd) for ICA segmentation through the skull base. The results show that the average overlap for the observers are similar. The proposed texture-based approach decreases significantly explosions, over-segmentations and increases rate of area overlap, sensitivity, precision at skull base. Therefore, the method is clinically useful and has potential to segment carotid arteries at skull base efficiently.