Volume 5 Number 1 June 2015


Digital Watermarking of Images using Modified hybrid Approach in Dual Domain
M. Ranjitha, G. M. Nasira

Abstract: With the expansion of the World Wide Web an increased amount of digital information becomes available to a large number of people. Information hiding has become very challenging with the advancement of technology. In Telemedicine the integrity of received images are very critical. The spatial domain watermarking is simple and less complex in nature. Regardless of the subsequent processing, it can be applied to any image. On the other hand, it is very sensitive to even simple attacks such as jitter attack, stir mark attack etc. These types of watermarking are sensitive even to rotation, transformation or print and scan, as the location of the watermark in this situation may be lost. The frequency domain watermarking is more secure and robust compared to spatial domain watermarking. Frequency domain watermarking is more complex and relatively difficult to implement. An image once watermarked in frequency domain, subsequent processing is not possible. The size of the watermark image to be embedded in the host image in frequency domain is less when compared to spatial domain. We have proposed a new modified hybrid approach using variance, which uses dual domains-spatial and frequency. By this approach, the proposed algorithm enables greater control over the cryptic part of the watermark. It is embedded in the host image and the embedded watermark size is greatly enhanced. In our approach, we have taken care of information hiding as well as the clarity of the image.


An atlas based approach to segment the hippocampus from MRI of human head scans for the diagnosis of Alzheimer’s disease
K. Somasundaram, T. Genish, T. Kalaiselvi

Abstract: The Hippocampus in a human brain is the focus of neuro imaging research for the recent years due to its important role in memory processes and the significance in neurological and psychiatric disorders. Hence the segmentation of hippocampus from MRI is inevitable to identify the diagnosis and the disease progression. But, the extraction of hippocampus is a tedious task since it is smaller in size and has a vague boundary. To facilitate the segmentation, in this paper we propose a method to segment Hc from MRI of human head scans. This segmentation method constitutes two phases. In the first phase, the approximate location of Hc in the input image is identified by atlas based approach. From that location, an enclosed rectangle called region of interest (ROI) is derived. In the second phase the ROI is processed by applying conservative smoothing and top-hat filter to preserve the edges of hippocampus. The filtered image is then binarized using Riddler Calvard method to differentiate the hippocampus from other irrelevant structures. Finally, hippocampus alone is segmented by Connected Component Analysis (CCA).


Traffic Analysis on Highways based on Image Processing
R. Sofia Janet, J. Bagyamani

Abstract: A steady increase in population, and the exponential increase in the number of vehicles, leads to traffic jam often during peak hours. Traffic analysis becomes a challenging problem as well as the needed one to control the traffic in decent and safe manner. Normally the traffic signals are operated on predefined fixed program, based on the time of day. In case if there is no vehicle in the allotted road, the time will lapsed for the other vehicles who are all waiting on the other side, which leads congestion. To rectify this issue, this paper presents an approach for analysis and detecting vehicles in highways traffic images by means of image processing techniques such as background differencing, Otsu’s thresholding and morphological filters. To count the detected vehicle region properties are used. The result can then be used to control the traffic signals. The whole work has been developed using MATLAB environment.


Secure Irreversible Rapid Fourier Transform For Secure Communication In Video Steganography
R. Umadevi, Dr. G. M. Nasira

Abstract: Recently, several efficient data hiding algorithms has been developed successfully for video steganography. Data hiding is one promising way to accomplish better data communication by hiding information into a video medium carrier to form an unrecognizable code stream. Motion features-based approach is a popular type of steganographic algorithms related to video coding crafts. However, in most existing approaches, the choice of features on the perceived video quality mainly depends on blurring and blocking effects without considering the variance and intensity of temporal changes in irreversible video steganography. In this work, a novel method is introduced to reduce the complexity of data hiding on video steganography, Adaptive Irreversible Rapid Fourier Transform (AIRFT) technique is proposed. The polynomial hashing in AIRFT ensures lesser complexity of data hiding and achieves pseudo randomness of the output without any packet (i.e.,) information loss on the video frame. Based on the generated functions, an efficient Rapid Fourier Transform method for increasing the disguise level and generate a random like output by addressing the variance and intensity of temporal message changes is presented. Finally, the proposed video steganography method is evaluated via simulations. The simulation results evaluated with the aid of SD sequences by Video Quality Experts Group (VQEG) with parameter such as packet information loss on video frame, complexity on data hiding, Peak Signal to Noise Ratio. It shows that the method AIRFT enhance the security significantly compared with typical state-of-the-art methods.


New Seed Selection Technique for Protein Sequeunce Motif Identification
M. Chitralegha, Dr. K. Thangavel

Abstract: Bioinformatics is a field devoted to the interpretation and analysis of biological data using computational techniques. In recent years the study of bioinformatics has grown tremendously due to huge amount of biological information generated by the scientific community. Protein sequence motifs are short fragments of conserved amino acids often associated with specific function. Identifying such motifs is one of the challenging tasks in the area of bioinformatics. Data mining is one technique to explore sequence motif from protein sequences. In this proposed work, recurring sequence motifs are identified by adopting new seed initialization technique for K-Means clustering algorithm. This proposed work combine’s local density approximation utilizes sorted pair wise distance calculation for identifying potential seeds for K-Means clustering. This new initialization technique enhances K-Means learning characteristics towards better cluster separation to identify the significant motif patterns.


Brain Extraction Algorithm for T1-W and T2-W MRI of Human Head Scans
K Somasundaram, P A Kalaividya, T Kalaiselvi

Abstract: In this paper, we extend a brain segmentation algorithm developed for T1-W and T2-W Magnetic Resonance Images (MRI). The proposed scheme consists of image denoising, intensity thresholding and largest connected component analysis. Usually an image diffusion is done to blur the image without losing edge properties and an intensity threshold is found for the diffused image. Using the diffused image and threshold T, a binary image is obtained for extraction of brain. In the proposed method, after computing the threshold T using diffused image, we use the undiffused image. We used T1-W and T2-W images collected from Internet Brain Service Repository (IBSR) and The whole Brain Atlas (WBA). Experimental results show that the proposed scheme works well on T2-W images and gave satisfactory results on T1-W images. The performance of the method is evaluated using the Jaccard and Dice similarity coefficients.


Neural Network Based Signature Authentication System with Regional Properties, Fractal Dimensions and QR code
D. Ashok Kumar, S. Dhandapani

Abstract: Every transactions authorized by handwritten signatures are accepted worldwide. Utmost care is to be taken for the verification of genuineness of the signature. A novel method for offline signature verification in bank cheque is proposed. The system uses connected Components Labeling, Fractal Dimensions, Quick Response (QR) code and Neural Networks. The signature is scanned and preprocessed. Using connected components labeling, the regional property features are extracted and normalised. Extracted feature values and fractal dimensions are compared with the sample signature’s feature values for its genuineness. A Neural Network is used to classify the signature into genuine or forged. Some complex signatures may require human intervention. An optimum signature verification model consumes less time and memory space in the database server. Conventionally the features extracted are stored in the database. Instead, the proposed model prints features in the QR code format on the cheque. Whenever the cheque comes for transaction, the QR code and the signature is scanned and verified. The proposed verification system shows very good results with good sensitivity and specificity with the CEDAR signature database. The system attains an accuracy of maximum 95% with very low false acceptance rate and false rejection rate. It is observed that, using fractal dimensions for verification purpose, improves the accuracy rate. Also the proposed model reduces the time, memory and cost for the signature verification process and may aid the banking community.


Automatic Brain Extraction from MRI of Human Head Scans using Fuzzy Logic and Bridge Building Algorithm
K. Somasundaram, K. Ezhilarasan, T. Kalaiselvi

Abstract: Brain extraction or skull stripping is a primary process in the brain image analysis. In this paper, we have proposed an automatic brain extraction method for Magnetic Resonance Images (MRI) using fuzzy logic, bridge building algorithm and morphological operations. We applied our proposed method on 5 volumes of MRI head scans taken from Internet Brain Segmentation Repository (IBSR) and extracted the brain portion. The performance of the proposed method is evaluated by computing similarity indices Jaccard (J) and Dice (D).


Abnormal Slices Identification Technique using GLCM Features and Least Square Line Fitting Technique for MRI T2- FLAIR Brain Scans
T. Kalaiselvi, S. Karthigai selvi

Abstract: The proposed work is used to extract abnormal slices from a Magnetic resonance image (MRI) volume taken from an abnormal patient. This process is an essential part in brain image processing pipeline and is an initial work in any brain segmentation process. It drives the automatic diagnostic process to reach the target images directly and quickly. The proposed work is developed by using Gray level co occurrence matrix (GLCM) features and least square line fitting techniques. The neighborhood features related to hyper intense regions are targeted for constructing the GLCM. Then the mean value of each slice GLCM is processed to separate the abnormal range. Finally, the least square line fitting technique is used to fix the lower and the upper limits of abnormal range. 20 high grade tumor volumes and 10 low grade tumor volumes of T2-FLAIR sequences are taken from BRATS database. Among them some volumes are suspected by artifacts and some are in good quality for further processing. The 3D volumes are converted into 2D slices for our experiment. For 12 volumes, the proposed technique yields 97% accuracy. The experiments were also carried out on the other artifact affected volumes and the noisy volumes. It results that the proposed method detects the abnormal slices (image) with more accuracy in good quality volume and abnormal slices along with the noisy and artifact affected slices.


Performance Analysis of Genetic and Bees Colony Optimization Techniques for Finest Routing in Ad-Hoc Network
T. Sudhakar, Dr. H. Hannah Inbarani

Abstract: An ad-hoc network is an infrastructure less network. Ad-hoc network is a seldom topology rottenly change their positions. The main destination of an ad-hoc network is to detect the shortest path between sources to destinations. Here using some evolutionary techniques to get an optimal path among topology. The proposed modified genetic algorithm is employed for the premature convergence of genes (PCG) with the help of a novel mutation operator and modified topology crossover (MTC), and also a simple bee’s colony optimization algorithm also implemented and compared with MGA and AODV. Both the algorithms are applied with an Ad-hoc On-Demand Distance Vector (AODV) routing protocol. In a previous work genetic algorithm compared with the DSR routing protocol.Here QoS applied for evaluating the performance of routing protocols.The simulation results are managed with the help of network simulator 2 (ns2) tools. The proposed modified genetic algorithm shows the best results compared with other methods.