Informatics Publishing Limited

S. Arumugan ¹, K. Senthamarai Kannan ², K. John Peter ³

Affiliations
1 Nandha Engineering College, Tamilnadu, IN
2 Department of Statstics, MS University, Tamilnadu, IN
3 Department of Information Technology, VINS Christian College of Engineering, Tamilnadu, IN

Abstract

In this paper one new compactive demosaicing Algorithm is used for color filter array (CFA). In all the existing algorithm they detect the edges in horizontal-, vertical- or Omni-direction. In the existing algorithm does not detect diagonal edges. We proposed a new approach of similarity-based demosaicing algorithm using unified high-frequency (UHF) map. Similarities between pixels are calculated on a local map called UHF map. A missing sample must be interpolated from the neighboring samples that are highly correlated with the missing sample. This highly correlated neighboring samples are detected by similarity calculation. It is, thus, expected that the proposed algorithm is able to deal with edges of any direction such as diagonal edges. The proposed algorithm improve PSNR and image quality. Again, the proposed algorithm requires fewer resources.

Keywords

Color Filter Array, Demosaicing, Edge-Directed Interpolation, Unified High-Frequency Map.

Full Text

     

V. Deneshkumar ¹, K. Senthamarai Kannan ², M. Manikandan ²

Affiliations
1 Department of Statistics, Manonmaniam Sundaranar University, Tirunelveli-627012, IN
2 Department of Statistics, Manonmaniam Sundaranar University, Tirunelveli, IN

Abstract

Outlier detection is important in many fields. In statistics, an outlier is a observation that is numerically far-away from the rest of the data. The handling of outlying observations in a data set is one of the most important tasks in data pre-processing. The large data base can be classified in an unsupervised manner using clustering and classification algorithms. Fuzzy C-means is a method of clustering which was developed by Dunn in (1973) and improved by Bezdek in (1981). This allocates one piece of data in two or more clusters and it is frequently used in pattern recognition. Herein a proposed method based on Fuzzy approach which combines outlier analysis and clustering technique is presented. Clustering validation technique adaptively evaluated the results of a clustering algorithm. A numerical example is provided for illustration using iris data set.

Keywords

Outlier Detection, Fuzzy Clustering, Silhouette Index, FCM Algorithm, and Random Number Simulation.

Full Text

     

K. John Peter ¹, G. Nagarajan ¹, S. Arumugan ², K. Senthamarai Kannan ³

Affiliations
1 VINS Christian College of Engineering, IN
2 Nandha Engineering College, Erode, IN
3 M.S University, IN

Abstract

Most cost-effective color cameras have single image sensors with a fixed repeating pattern of color filters with transmission spectra analogous to the human eye. To reconstruct a full color image a process often referred to as “demosaicking” is necessary. Almost every past method for demosaicking had the intention to create the most pleasant image for the human observer. The quality of demosaicked images is degraded due to the sensor noise introduced during the image acquisition process. According to the proposed Demosaicking algorithm we can extract edge information in terms of the direction of variation and the gradient from the mosaic image directly and accurately, and the extracted more accurate edge information will be utilized to assist the design of our proposed new demosaicking algorithm. The proposed algorithm is also capable of reducing the interpolation errors which are visually objectionable because they tend to correlate with object boundaries and edges. Proposed algorithm find out the missing color components in the mosaic images captured by the CFA and improve the visual quality of the resulting output.

Keywords

Demosaicking, Gradient, Mosaic Image, Correlation, Mosaic, CFA.

Full Text

     

M. Mohamed Sathik ¹, K. Senthamarai Kannan ², Y. Jacob Vetha Raj ²

Affiliations
1 Sadakathullah Appa College, Tirunelveli, IN
2 Manonmaniam Sundaranar University, Tirunelveli, IN

Abstract

The focus of the image is normally at the center of the photograph. The images with objects with different distance have the problem of lacking focus on some of the objects. In order to overcome this problem, a simple and effective method is proposed. The proposed method may be incorporated in a digital camera to capture images with different focuses of the same image. Then these images can be fused together to get a hybrid image with the details of focused area of all the images of a frame. Initially the RGB color images with different focuses of the same image are captured and converted to gray images. These images are subjected to a filtering operation with an overlapping window of order 3x3. The filtering operation finds the maximum difference of the center pixel with the neighboring eight pixels. A sliding window of order n x n is selected in all the images, the sum of the magnitude values of the difference are computed. The block in the image with higher block sum is selected and is fused with hybrid image. This hybrid image will have more details compared to the ordinary high quality image with single focus. The hybrid image generated helps in inspection to extract more details.

Keywords

Camera Automation, Edge Detection, Filtering, Image Fusion.

Full Text

     

M. Mohamed Sathik ¹, K. Senthamarai Kannan ², Y. Jacob Vetha Raj ²

Affiliations
1 Department of Computer Science, Sadakathullah Appa College, Tirunelveli, IN
2 Department of Statistics, Manonmaniam Sundaranar University, Tirunelveli, IN

Abstract

Color Filter Arrays (CFAs) are used in Single sensor cameras to capture the images. The sensors in a CFA are arranged based on Bayer pattern to capture any one of the Red, Green or Blue components of a pixel. The image captured by this type of CFAs is known as mosaic image. The compression ratios achieved while compressing these mosaic images are normally low for all the classical compression methods. Different strategy has to be adopted for compressing mosaic images due to the roughness caused by interleaving of components. This paper proposes a hybrid method by which the mosaic images are compressed by preserving the main subject after segmenting using histogram. The background of the image is stored with some level of loss. Compression ratio is increased by using the proposed hybrid method than the existing methods.

Keywords

Image Compression, Mosaic Image, Segmentation.

Full Text

     

V. Vallinayagam ¹, N. Senthil Vel Murugan ², K. Senthamarai Kannan ³

Affiliations
1 Department of Mathematics, St. Joseph's College of Engineering, Chennai, IN
2 Department of Mathematics, Rohini College of Engineering and Technology, Kanyakumari Dist-629401, IN
3 Department of Statistics, Manonmaniam Sundaranar University, Tirunelveli-627012, IN

Abstract

Medical dataset is a vital ingredient used in predicting patient's health condition. In other to have the best Prediction, there calls for a technique with high degree of accuracy. With the computerization in hospitals, a huge amount of data is collected. Although human decision-making is often optimal, it is poor when there are huge amounts of data to be classified. Medical data mining has great potential for exploring hidden patterns in the data sets of medical domain. These patterns can be used for clinical diagnosis. In this paper, we modeled data from diabetes patients and used it to predict the diabetes probability of any patient. In this paper a method for constructing fuzzy membership functions from data collected from Hospitals and diagnoses in the medical application area of diabetics is being presented. Fuzzy membership functions are generated to represent linguistic medical concepts for the data to symbol conversion unit of the medical knowledge based system. This data consists of six variables collected from private hospital in kanyakumari district. The variables chosen are low, medium and high. Based on the variables the risk of diabetics can be diagnosed. The aim of this paper is analyze the Diabetics data and Mamdani's fuzzy inference system is used. The reasonable results verify the validity of our method. This method tries to use the data mining technique effectively than other models available. Moreover the variables taken are linguistic variables which focuses on accuracy of the results.

Keywords

Data Mining, Diabetics, Membership Function and Fuzzy Inference System.

Full Text

     

V. Deneshkumar ¹, K. Senthamarai Kannan ²

Affiliations
1 Department of Statistics, Manonmaniam Sundaranar University, Tirunelveli-627012, IN
2 Department of Statistics, Manonmaniam Sundaranar University, Tirunelveli, IN

Abstract

For many data mining applications, finding the outliers is more interesting than finding the common patterns of the data. Outliers are frequently adapted in time series data mining analysis. The main objective of this paper, outliers on forecasting in agricultural production is analyzed. Outliers in time series data was carried out by Fox (1972). Outlier detection has been used for detect and, where appropriate, remove inconsistent observations from data. The original outlier detection methods were arbitrary but new, Principled and systematic techniques are used, drawn from the full scope of computer science and statistics. In agricultural production outliers are initially detected and then forecast using ARIMA model. Predictions made after detecting outliers are compared with numerically and graphically the predictions made before detecting outliers.

Keywords

Data Mining, Outliers, Forecasting, Mean Square Error and ARIMA Model.

Full Text

     

N. Senthil Vel Murugan ¹, V. Vallinayagam ², K. Senthamarai Kannan ³

Affiliations
1 Department of Mathematics, Rohini College of Engineering and Technology, Kanyakumari Dist, 629401, IN
2 Department of Mathematics, St. Joseph's College of Engineering, Chennai-600119, IN
3 Department of Statistics, Manonmaniam Sundaranar University, Tirunelveli-627012, IN

Abstract

Data mining allows users to discover novelty in huge amounts of data. The recent studies have used individual structures for study while this study focuses on sequential pattern mining. This study attempts to study sequential patterns extracted from gene data. The data for the present study were collected from the Gen Bank. The data taken for study is DNA sequence of samples affected by Liver cancer. It can be inferred from the analysis that increases or decrease in protein level, hormone level contributes to Lever cancer. The aim of this paper is analyze the above liver cancer DNA sequence data and reduce the variable size by Principal Component Analysis and Singular value decomposition technique and which proteins will affect quickly as possible using Similarity techniques. The reasonable results verify the validity of our method.

Keywords

Data Mining, Liver Cancer, Principal Component Analysis, Singular Value Decomposition, DNA.

Full Text

     

C. Vijayalakshmi ¹, K. Senthamarai Kannan ²

Affiliations
1 Department of Mathematics, Sathyabama University, Chennai-119, IN
2 Department of Statistics, Manonmaniam Sundaranar University, Tirunelveli, IN

Abstract

This paper mainly deals with the recent advances in DNA Micro array technologies and the abundance of Genomic information which plays a vital role for analyzing the molecular mechanisms. Statistical and Machine Learning Algorithms are used to analyze the biological implications in order to discover complex gene expression patterns. A Hidden Markov model (HMM) is designed and the maximum penalized likelihood is used to estimate the parameters in this model. This method is applied to lung cancer micro array experiment. Several regions identified through the HMM are consistent with known recurrent regions of amplifications or deletions in cancer. The association of these abnormal expression regions with the measures of disease status, such as tumor stage, differentiation, and survival are being analyzed. Numerical calculations and graphical representations reveals that genes in these regions may play a major role in the process of carcinogenesis of the lungs.

Keywords

DNA, Molecular Information, HMM, Machine Learning Algorithm, Lung Cancer, Markov.

Full Text

     

V. Balamurugan ¹, K. Senthamarai Kannan ², S. Selvakumar ²

Affiliations
1 Department of Computer Science and Engineering, Sri Vidya College of Engineering and Technology, Tamil Nadu, IN
2 Department of Statistics, Manonmaniam Sundaranar University, Tamil Nadu, IN

Abstract

Searching for similar sequence in large database is an important task in temporal data mining. Similarity search is concerned with efficiently locating subsequences or whole sequences in large archives of sequences. It is useful in typical data mining applications and it can be easily extended to image retrieval. In this work, time series similarity analysis that involves dimensionality reduction and clustering is adapted on digital images to find similarity between them. The dimensionality reduced time series is represented as clusters by the use of K-Means clustering and the similarity distance between two images is found by finding the distance between the signatures of their clusters. To quantify the extent of similarity between two sequences, Earth Mover’s Distance (EMD) is used. From the experiments on different sets of images, it is found that this technique is well suited for measuring the subjective similarity between two images.

Keywords

Similarity Search, Vector Quantization, Similarity Measures, Clustering, EMD.

Full Text

     

Jacob Vetha Raj ¹, M. Mohamed Sathik ², K. Senthamarai Kannan ³

Affiliations
1 Department of Computer Science, Nesamony Memorial Christian College, IN
2 Sadakathullah Appa College, IN
3 Department of Statistics, Manonmaniam Sundaranar University, IN

Abstract

A hybrid image compression method is proposed by which the Nuclei-Cytoplasm of the image is completely restorable and the background part of the image is restorable with insignificant loss. In Hybrid Compression of Cervical Images by Segmenting Nuclei-Cytoplasm, the image is subjected to binary segmentation to detect Background and Nuclei-Cytoplasm. The image is compressed by standard lossy compression method. The difference between the lossy image and the original image is computed as residue. The residue at the Nuclei-Cytoplasm area is compressed by standard lossless compression method by which the Nuclei-Cytoplasm area is completely restorable. This method gives a low bit rate than the lossless compression methods.

Keywords

Edge Detection, Segmentation, Image Compression.

Full Text

     

S. Abirami ¹, K. Chitra ², K. Senthamarai Kannan ³

Affiliations
1 Dept.of CA&IT, Thiagarajar College (Autonomous), Madurai-9,Tamil Nadu, IN
2 Dept.of Computer Science, Govt.Arts College, Melur, Madurai, Tamilnadu, IN
3 Dept. of Statistics, Manonmaniam Sundaranar University, Tirunelveli, Tamilnadu, IN

Abstract

Miraculous growth in mobile technology increases the mobile usage day by day. Mobile data traffic and the limitation of power backup are still the major problem in disseminating consistent data on mobile database environment. In this paper, the author propose a consistent replicated mobile data broadcast algorithm(CRMB) to improve the performance of broadcast approach by providing high data availability, consistency and currency with the minimum data access delay and uplink communication. When an update transaction is executed on replicated mobile database environment, it preserves the consistency among the fixed host server, replica servers and the read-write mobile clients. The implementation results show that the performance of proposed approach is better in most cases. Key features of the proposed approach are:

• It disseminates consistent data quickly to all mobile clients in the same order in which they were updated through minimum bandwidth communication 
• It allows the mobile clients to get broadcasted data items from any nearest replica server without contacting the server.

Keywords

Mobile Computing, Data Replication, Consistency, Data Dissemination, Temporal Database.

Full Text