1. Faculty Publications
Permanent URI for this communityhttps://idr.nitk.ac.in/handle/1/5
Browse
31 results
Search Results
Item New sparse matrix storage format to improve the performance of total SPMV time(2012) Neelima, B.; Prakash, S.R.; Ram Mohana Reddy, GuddetiGraphics Processing Units (GPUs) are massive data parallel processors. High performance comes only at the cost of identifying data parallelism in the applications while using data parallel processors like GPU. This is an easy effort for applications that have regular memory access and high computation intensity. GPUs are equally attractive for sparse matrix vector multiplications (SPMV for short) that have irregular memory access. SPMV is an important computation in most of the scientific and engineering applications and scaling the performance, bandwidth utilization and compute intensity (ratio of computation to the data access) of SPMV computation is a priority in both academia and industry. There are various data structures and access patterns proposed for sparse matrix representation on GPUs and optimizations and improvements on these data structures is a continuous effort. This paper proposes a new format for the sparse matrix representation that reduces the data organization time and the memory transfer time from CPU to GPU for the memory bound SPMV computation. The BLSI (Bit Level Single Indexing) sparse matrix representation is up to 204% faster than COO (Co-ordinate), 104% faster than CSR (Compressed Sparse Row) and 217% faster than HYB (Hybrid) formats in memory transfer time from CPU to GPU. The proposed sparse matrix format is implemented in CUDA-C on CUDA (Compute Unified Device Architecture) supported NVIDIA graphics cards. 2012 SCPE.Item An intelligent content-based image retrieval system for clinical decision support in brain tumor diagnosis(2013) Arakeri, M.P.; Ram Mohana Reddy, GuddetiAccurate diagnosis is crucial for successful treatment of the brain tumor. Accordingly in this paper, we propose an intelligent content-based image retrieval (CBIR) system which retrieves similar pathology bearing magnetic resonance (MR) images of the brain from a medical database to assist the radiologist in the diagnosis of the brain tumor. A single feature vector will not perform well for finding similar images in the medical domain as images within the same disease class differ by severity, density and other such factors. To handle this problem, the proposed CBIR system uses a two-step approach to retrieve similar MR images. The first step classifies the query image as benign or malignant using the features that discriminate the classes. The second step then retrieves the most similar images within the predicted class using the features that distinguish the subclasses. In order to provide faster image retrieval, we propose an indexing method called clustering with principal component analysis (PCA) and KD-tree which groups subclass features into clusters using modified K-means clustering and separately reduces the dimensionality of each cluster using PCA. The reduced feature set is then indexed using a KD-tree. The proposed CBIR system is also made robust against misalignment that occurs during MR image acquisition. Experiments were carried out on a database consisting of 820 MR images of the brain tumor. The experimental results demonstrate the effectiveness of the proposed system and show the viability of clinical application. 2013, Springer-Verlag London.Item A novel spatio-temporal registration framework for video copy localization based on multimodal features(2013) Roopalakshmi, R.; Ram Mohana Reddy, GuddetiFighting movie piracy requires copy detection followed by the accurate frame alignments of master and copy videos, in order to estimate distortion model and capture location in a theater. Existing research on pirate video registration utilizes only visual features for aligning pirate and master videos, while no effort is made to employ acoustic features. Further, most studies in illegal video registration concentrate on the alignment of watermarked videos, while few attempts are made to address the alignment of non-watermarked sequences. We attempt to solve these issues, by proposing a novel spatio-temporal registration framework that utilizes content-based multimodal features for frame alignments. The proposed scheme includes three stages: first, a video sequence is compactly represented using Speeded Up Robust Features (SURF) and audio spectral signatures; second, sliding window based dynamic time warping (DTW) is employed to compute temporal frame alignments; third, robust SURF descriptors are utilized to generate accurate geometric frame alignments. The results of experiments on three different datasets demonstrate the robustness and efficiency of the proposed method against various video transformations. 2012 Elsevier B.V.Item Recent trends in content-based video copy detection(2010) Roopalakshmi, R.; Ram Mohana Reddy, GuddetiVideo copy detection is an interesting & challenging problem, as it is becoming increasingly important with the growing rate of illegal digital media and huge piracy issues. Therefore, in the present era of Internet &Multimedia technologies,the existence of ubiquitous digital videos, has led to the requirement of robust video copy detection techniques, for content management and copyright protection. In this paper, an overview of content based video copy detection (CBVCD) is presented along with the different state-of-the-art techniques used for video feature/ signature description. This article also explores the future key directions and highlights the research challenges that need to be addressed in the CBVCD paradigm. � 2010 IEEE.Item Recent advances and future potential of computer aided diagnosis of liver cancer on computed tomography images(2011) Megha, P.A.; Ram Mohana Reddy, GuddetiLiver cancer has been known as one of the deadliest diseases. It has become a major health issue in the world over the past 30 years and its occurrence has increased in the recent years. Early detection is necessary to diagnose and cure liver cancer. Advances in medical imaging and image processing techniques have greatly enhanced interpretation of medical images. Computer aided diagnosis (CAD) systems based on these techniques play a vital role in the early detection of liver cancer and hence reduce death rate. The concept of computer aided diagnosis is to provide a computer output as a second opinion in analysis of liver cancer. It assists radiologist's image interpretation by improving accuracy and consistency of radiological diagnosis and also by reducing image analysis time. The main objective of this paper is to provide an overview of recent advances in the development of CAD systems for analysis of liver cancer. Medical imaging system based on computer tomography will be focused as it is particularly suitable for detecting liver tumors. The paper begins with introduction to liver tumors and medical imaging techniques. Then the key CAD techniques developed recently for liver tumor detection, classification, case-based reasoning based on image retrieval and 3D reconstruction are presented. This article also explores the future key directions and highlights the research challenges that need to be addressed in the development of CAD system which can help the radiologist in improving the diagnostic accuracy. � Springer-Verlag 2011.Item Projection and interaction with ad-hoc interfaces on non-planar surfaces(2013) Dere, K.S.; Ram Mohana Reddy, GuddetiProjector-based display systems have been used in area of computer interaction as Ad-hoc interface in recent time. The mobile hand-held projectors are becoming more popular. Many human centric user interfaces with the human wearable computer are being developed. Most of such system uses daily objects for projection and the interaction. But most of ignores the fact that these object surfaces are not planar. Hence such interfaces suffers from the distortion due to non-planar projection surface. Besides this projection quality also suffers from the radiometric distortion as well. Further more the interaction proposed with such interfaces bound to the planar surface only. Hence this paper is targeted to address the geometric distortion free projection of and interaction with such interfaces on non planar surfaces. Kinect is used as depth sensor for 3D scenario acquisition. We use imageper warping to mesh from Kinect. We use colored fingertip gloves for interaction. Here our system aims any day to day object surface for distortion free projection such as human body, curved wall, room corners, curtain's and many more objects. � 2013 IEEE.Item Primary education for the specially-abled(2013) Chandra, S.; Sagar, M.; Rout, P.; Khattri, P.; Domanal, S.; Ram Mohana Reddy, GuddetiTechnology can support the children who require special needs, who if given the proper training and opportunity, can compete on a basis of equality with their peers. This should be the basic philosophy of a programmer who designs programs, standards for programs, or evaluation of programs. Proper education for these children will lead to enhancing their capability to lead a dignified life and also help them to earn a square meal. Technology is needed to teach them and hence the necessity can clearly be seen for further research and development in this field. In addition to software being used as teaching tools at schools and at-home, the learning process should be more interesting so that child should feel engaged. A daily routine, not only a syllabus/homework, using technologies such as text to speech conversion and image morphology is needed to both help them understand concepts of classroom syllabus and motivate them to learn more at home as specially-abled children need to be given enough motivation, as well as time, to succeed. The system developed proves to be useful to specially-abled children to memorize rhymes, recognize common sounds (like that of animals) as well as develop haptic abilities using a game like interface. � 2013 IEEE.Item Parallelized K-Means clustering algorithm for self aware mobile Ad-hoc networks(2011) Thomas, L.; Manjappa, K.; Annappa, B.; Ram Mohana Reddy, GuddetiProviding Quality of Service (QoS) in Mobile Ad-hoc Network (MANET) in terms of bandwidth, delay, jitter, throughput etc., is critical and challenging issue because of node mobility and the shared medium. The work in this paper predicts the best effective cluster while taking QoS parameters into account. The proposed work uses K-Means clustering algorithm for automatically discovering clusters from large data repositories. Further, iterative K-Means clustering algorithm is parallelized using Map-Reduce technique in order to improve the computational efficiency and thereby predicting the best effective cluster. Hence, parallel K-Means algorithm is explored for finding the best effective cluster containing the hops which lies in the best cluster with the best throughput in self aware MANET. Copyright � 2011 ACM.Item Parallel implementation of 3D modelling of indoor environment using Microsoft Kinect sensor(2013) Manojkumar, P.; Ram Mohana Reddy, Guddeti3D visual modelling of indoor space will not only provide the detailed knowledge about the environment but also rich contextual information of the existing objects. In this paper, we propose a parallel implementation of 3D modelling of indoor environment using Microsoft Kinect depth camera. 3D maps are generated by Simultaneous Localization And Mapping [SLAM] technique. These 3-D maps will be more useful in Context aware and in Robotics applications. Iterative Closest Point [ICP] with initial guess by RANSAC is used for pair alignment. Many tasks of pair registration are parallelised using OpenMP and the performance evaluation of both sequential and parallel implementations are compared. Simulation results demonstrate that OpenMP based parallel implementation has achieved a speedup factor of 3.7. � 2013 IEEE.Item Optimized Termite: A bio-inspired routing algorithm for MANET's(2012) Hoolimath, P.K.G.; Kiran, M.; Ram Mohana Reddy, GuddetiA Mobile Adhoc Network (MANET) is a collection of mobile nodes connected by the Wireless medium and each mobile node is aware of only its neighbours. Due to mobility of these mobile nodes the topology changes dynamically. Such a dynamic network topology makes the task of routing a challenging one. Recently, a new class of routing algorithms based on Swarm Intelligence has emerged. These algorithms are inspired by nature's self-organizing systems like ant-colonies, bird-flocks, honey-bees, school of fish, spiders and fireflies. The characteristics of such algorithms are their capability of self-organization, adaptation to the changing conditions, self healing and local decision making. In this work, a routing protocol inspired by the termite activity in nature, called Optimized-Termite (Opt-Termite), is proposed. Opt-Termite uses concept of stigmergy for self-organization, thereby reducing the control packet overhead. Opt-Termite mainly concentrates on load balancing for optimization. With Opt-Termite, a route with less loaded mobile nodes in terms of traffic will be chosen to reach destination. The routing information at each node gets influenced by the movement of packets and the routing table will be updated accordingly. It also allows the use of multiple paths and each packet is routed randomly and independently. Opt-Termite is implemented in ns-2 and its performance is compared with traditional routing protocol AODV. Opt-Termite's performance has been promising. � 2012 IEEE.