Asian Science Citation Index is committed to provide an authoritative, trusted and significant information by the coverage of the most important and influential journals to meet the needs of the global scientific community.  
ASCI Database
308-Lasani Town,
Sargodha Road,
Faisalabad, Pakistan
Fax: +92-41-8815544
Contact Via Web
Suggest a Journal
 
Articles by A. Kannan
Total Records ( 12 ) for A. Kannan
  P. Anandhakumar , M. Boopathi Raja and A. Kannan
  A new approach is proposed for retrieving video sequences from large video databases based on code generated from tree structures of video frames. Video retrieval in a distributed environment faces the challenges of communicating the query to sites and fast retrieval of sequences. Effective representation of query frame is more important in retrieving the video sequence from video databases. For better representation, query frame is hexagonally sampled, which has the advantages of 12-fold symmetry, high packing density, nearly circular in shape and well-behaved connectivity, all of which lead to faster image processing. A Ternary tree is constructed from the hexagonally sampled values. In this study a string of 24 numbers are generated called identification code for each and every frame’s Ternary tree of the video database. The code generated is of size 96 bytes and the code is used to retrieve the video sequence by comparing the code of query frame and database frame in a distributed environment efficiently. Storage requirement for the code is very less compared to the volume of database and the overall accuracy is nearly 100%.
  S. Behin Sam , Amna Omar , A. Kannan and P. Vivekanandan
  Generally data communication is done by using the routing algorithms that uses metrics such as path length, communication cost etc., to determine the optimal path to a destination. This paper approaches the problem of routing in communication network from a new angle by using data mining technologies to determine the optimal path in a network. The new routing model combines several scalable data mining methods, such as decision tree, association rules and neural networks. The model predicts the quality of service in the communication paths, which can be used by routing algorithms to determine the optimal path to a destination.
  G. Suresh Kumar , R. Baskaran and A. Kannan
  Automatic image annotation is a process of assigning semantic keywords to images and these annotations are used to retrieve the unlabeled images from large image collections by using semantic query texts. We are proposing the AIAS (Automatic Image Annotation System), which provides a effective mechanism for Annotating images using an active learning framework. The visual features like color and shape gives a great evidence for representing image blobs and its usage for image annotation has been explored in this study. The extracted image feature vectors (color, shape) and training keywords are used by machine learning techniques to automatically apply annotations to new images. During training phase the SVM (Support Vector machine) generation process learns the correlations between image features and training keywords. The trained model provides the mapping between training image data set and semantic keywords and then the trained decision model can be used for the automatic image annotation process.
  P. Yogesh , S. Bose and A. Kannan
  Rate control is an important issue in video streaming application for both wired and wireless networks. A widely accepted rate control method in wired networks is TCP Friendly Rate Control (TFRC). TFRC assumes that packet loss in wired networks is primarily due to congestion and as such is not be applicable to wireless networks in which the bulk of packet loss is at the physical layer. Hence the multiple TFRC Connections (MULTFRC) are being used as an existing approach, which is an end-to-end based solution to this problem. By opening appropriate number of TFRC connections, MULTFRC not only avoids modifications to the network infrastructure but also results in full utilization of the wireless bandwidth. However MULTFRC suffers form underutilization of resources and loss of granularity in number of connections. To address these drawbacks, we propose a modified TFRC rate control scheme that combines multiple connections to one connection and improves the performance of streaming over wireless networks. We name our rate control scheme as CTFRC (Consolidated TFRC) with loss handling mechanism. We carried out NS-2 simulations to compare the performance of MULTFRC and our own rate control scheme.
  H. Khanna Nehemiah , A. Kannan and D. Senthil Kumar
  The retrieval of stored medical images matching an input medical image is an imperative form of content-based retrieval. For efficient similarity image retrieval and integration, the medical images should be processed systematically to extract a representing feature space vector for each member image. This study explains a system, which takes a fractured image as a query image and retrieves the similar images from the image database using distance metrics and also provides the radiologists with details about the type of fracture and the treatment recommended. The key objective of present research is to retrieve similar X-ray images of fractured reports using K-Nearest Neighbor. Images are matched using color in gray level and texture attributes. Similarity between images is established based on the respective numeric values (Signature). Features are extracted from X-ray images. Indexing is also performed on extracted features using a k-d tree data structure for images and is stored in a backend database for effective retrieval.
  Angelina Geetha , R. Srinivasan and A. Kannan
  In this study, we propose a method to improve the precision of top N retrieved documents retrieved from the web by re-ordering the retrieved documents from a search engine. The user query is accepted and the search process is initiated by employing an external search engine. On the retrieved search results, content analysis is carried out and various measures of relevance are calculated. Based on the overall relevance measure, the search results are reranked. The search context plays a vital role in framing of the query and search process. Hence we propose an algorithm to perform the context analysis on the reranked results. The benefit of this is two fold. First, the user is given a preview about on what context the keywords are used in a document thus reducing the irrelevant document browsing time. Second, by viewing the context, the user can fine tune the search query to get a closer search result. From the experimental results we have found that the reranking based on our relevance measure shows improvement in the search result obtained from search results.
  S. Sujatha , P. Vivekanandan , P. Yogesh and A. Kannan
  Mobile Ad hoc network (MANET) is gaining more importance and popularity due to the proliferation of miniature yet powerful mobile computing devices. MANET is more vulnerable due to its networks characteristics, such as dynamic topology, distributed cooperation and open medium. Ad hoc On demand Distance Vector (AODV) protocol is the most popular reactive routing protocol designed for the MANET. AODV is vulnerable to both external and internal security attacks. In this study, we analyze the vulnerabilities of AODV protocol, specifically the internal attacks and we propose solutions to monitor the attack by using an Intrusion Detection System (IDS). Our solution is based on a Fuzzy Based Response Model (FBRM).
  R. Baskaran and A. Kannan
  Tracking faces and identifying expression from a video files is a complex mining application due to the fact that even a delicate change in the emotion of the face could convey a great variation as an expression. This paper presents a robust novel method to track both male and female faces and recognize the facial expression in video files. Face localization in a digital image, which is being grabbed from the video source is performed using the hue-saturation and YC C based algorithms. Morphological algorithm over this region b r signals the eyes and mouth position in that region. Lips, eyebrows and eyelids are then identified. Distance between eyebrow and eyelid and the slope of the curvature of the lips are derived which facilitates in deciding the facial expression of the given digital image.
  S. Bose , P. Yogesh and A. Kannan
  This study proposes a distributed intrusion detection system for adhoc wireless networks using self organizing maps and mobile agents. In this research, we efficiently use log file data obtained from the local host for training the neural network, to analyze the adhoc wireless network for detecting intrusions. Security agents are used to monitor multiple clients of the wireless network to determine the correlation among the observed anomalous patterns and to report such abnormal behavior to the administrator and the user in order to take possible actions. From the system developed in this research, we obtained high intrusion-detection rates (99.2%) and low false-alarm rates. The main contribution of this paper is the provision of an agent based framework that is capable of detecting intruders and to forecast the anomalies using the neural classifier, self organizing maps.
  H. Khanna Nehemiah and A. Kannan
  In this study we propose an Intelligent Lung Cancer Diagnosis System (ILCDS) that has been developed to detect all possible lung nodules from chest radiographs. Our system uses image processing techniques and feed forward neural networks for detection and validation of nodules. Nodules are relatively low-contrast white circular objects within the lung fields. As nodules are the most common sign of lung cancer, nodule detection in chest radiographs is a major diagnostic problem. Even experienced radiologists have trouble while distinguishing the normal pattern of blood vessels and nodules that indicate the Lung cancer. Our work is centered around two major sub systems namely Nodule Detection Subsystem (NDS) and Nodule Validation Subsystem (NVS). The Nodule Detection Subsystem is constructed using wavelet based image-processing techniques such as Besov ball projections, Laplacian of Gaussian filter and Gabor wavelet networks which are used to remove the noise from the image, find the edges of the image and detect the nodule, size and its location. The NDS detects all the possible nodules and gives the nodule-detected image. The processed image shows all nodules in the chest radiograph. Since all nodules are not cancerous, the nodules detected by the NDS are validated by the NVS. The NVS is constructed using Feed forward neural network classifiers, which classifies the nodules into non-cancerous and cancerous nodules.
  K. Kulothungan , S. Ganapathy , S. Indira Gandhi , P. Yogesh and A. Kannan
  In wireless sensor networks, the nodes are deployed randomly over a large geographical region for monitor and collect the data. These nodes are powered by battery and are impossible to get recharged after deployment. Thus, minimizing the energy consumption of the sensor nodes is a challenging issue in wireless sensor networks for guaranteeing the network’s lifetime. Clustering is one of way to maintain an effective topology in wireless sensor networks in order to reduce the overall network’s energy consumption. In wireless sensor network providing an energy efficient in sensor networks is more difficult due to lack of infrastructure in the predicting the location of each node in the network. In this study, researchers propose a new clustering scheme that optimizes energy using a clustering algorithm to enhance the performance of the network. In addition, fault tolerant routing is proposed in this research by considering cluster sub heads. Therefore, we have modified the AODV with a clustering and security features to reduce the number of nodes involved in routing and to avoid denial of service attacks. We compare the performance of this proposed protocol with AODV with respect to distance and energy metrics. This modified AODV with clustering improves the QoS in terms of packet loss, packet delivery ratio and end-to-end delay.
  S. Behin Sam , S. Venkateswaran , A. Kannan and P. Vivekanandan
  Generally in ad hoc it is necessary for one mobile host to enlist the aid of others in forwarding a packet to its destination. In a fast changing ad hoc network, the routing tables will be out of data on a regular basis. Current routing algorithms are not adequate to tackle this problem. This study focuses on the use of topological information for routing in ad hoc network. The approach is to use topological information to train the Artificial Neural Network (ANN) for identifying the various subnets. Once the subnets are identified another ANN is used to elect the backbone node. This collection of backbone nodes forms the backbone network. A combination of AODV and DSDV protocols were used on the backbone network and on the local subnet for performance analysis. The AODV-DSDV protocols achieved a better delivery fraction and end-to-end delay than by using AODV alone. Implementation was carried out by using NS2.
 
 
 
Copyright   |   Desclaimer   |    Privacy Policy   |   Browsers   |   Accessibility