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 B. Sujatha
Total Records ( 7 ) for B. Sujatha
  B. Sujatha and S. Chenthur Pandian
  Mining predominant patterns in a varying time series databases is a significant data mining problem with numerous applications. The existing closed sequential patterns permit us to improve efficiency without bringing down the accuracy. The narrative technique developed a earlier research follows a multiplex tree pruning technique which combines both the prefix and suffix tree patterns in an activity normalized time periodicity data sequences. The combinatorial point of prefix and suffix trees is on the threshold of predominant data pattern occurrence rate which efficiently identify the regularity of all observed patterns but still obtains the interlaced unwanted data. To separate the interlaced unwanted data from the predominant pattern mining, researchers are going to implement a new technique termed Optimized Discrete Interested Pattern technique (ODIP). This technique identifies the optimal value using the repetition occurrence in the pattern. An analytical and empirical result offers an efficient and effective predominant pattern mining framework for highly dynamic online time series data. Performance of the optimized discrete interested pattern technique is measured in terms of interlaced data removal efficiency, time taken for online pattern mining based on the frequency. Experiments are conducted with online time series data obtained from research repositories of both synthetic and real data sets.
  U.S.N. Raju , B. Eswara Reddy , V. Vijaya Kumar and B. Sujatha
  A novel method for dominant skeleton extraction of textures using different wavelet transforms, is proposed in this study. The skeleton varies depending on the shape of structuring element. If the structuring element is homothetic to the object, the object is covered with only one magnification of the structuring element. By this, the skeleton is reduced to one point. The present study considers the skeleton from a binary texture. The proposed method derives from the above that a total number of pixels within the skeleton is the minimum when structuring element is homothetic to the primitive. This provides the scope that the texture is composed of one primitive, which minimizes the total number of pixels. For evaluating such skeleton primitive the present study utilized a 3x3 structuring element, as the skeleton primitives. All possible skeleton primitives combinations of 3x3 mask are evaluated on all textures. The skeleton primitive that is making the least number of skeleton points is considered as dominant skeleton primitive. Based on the extraction of skeleton primitives a classification is made on textures using Haar, Daubechies, Coiflet and Symlet wavelets. Experimental results indicate a good classification and also a comparison is made among these four wavelet results. Present method is experimented on Brodatz textures using these four wavelets.
  B. Sujatha and T. Santhanam
  Lipreading a perception of speech for listeners with hearing impairment is purely based on observing the lip movements under noisy conditions where visual speech information plays an important role. Lipreading, a visual modality which involves watching the movement of lips constitutes 1/3 of the conveyed message. This study investigates the use of two feature extraction methodologies for recognizing isolated words. The first type is based on a geometric approach which extracts the features like inner height, inner width, outer height and outer width of the lips while the second type is based on a set of Block-Based Gabor-Wavelet co-efficient extracted from each frame. Then these two features are given as input into the Ergodic Hidden Markov Model for recognizing the words.
  B. Sujatha and S. Chenthur Pandian
  A time series is collection of well-defined data sets obtained through repeated measurements of time. Extraction of periodic pattern in a time series database is significant one in data mining problem that predicts and forecasts the future behavior of the data at regular time interval. Periodic pattern mining involves several applications such as prediction, forecasting, detection of unusual activities. The difficulty is not trivial because the data to be examined are regularly noisy and diverse periodicity types (that is symbol, sequence and segment) are to be examined. The whole time series or in a subsection of it to effectively handle various types of noise (to a definite degree) and at the same time to detect different types of periodic patterns. The existing suffix tree based periodic pattern mining algorithm can detect symbol, sequence and segment periodicity in time series data with noise filters for diverse noise kinds. But the running time desired to identify the patterns without redundancy is high. So, to overcome this issue, in this study, Predominant Pattern Distribution Model is introduced with which redundant and unwanted noisy patterns are identified and discarded from the time series data. Predominant patterns are extracted with automatic or user defined threshold of pattern of interest, generated from the dynamic online time series data. Experiments conducted on both synthetic and real data sets of research repositories including protein sequences. Performance of proposed framework is measured and evaluated in terms of periodic pattern mining accuracy, noise distribution rate and predominant pattern occurrence.
  B. Sujatha and S. Chenthur Pandian
  Discovering specified patterns in a time series database has expected much consideration and is nowadays a comparatively mature field. Existing Periodic Pattern Mining algorithms concentrates on mining which involves subsequences. However, huge portion of requests for example, genetic DNA and protein pattern mining requires estimated patterns that are adjacent in nature. The existing algorithms applied to discover such estimated pattern mining comprises of complicated problems such as deprived scalability and complexity while applying towards certain other applications. To overcome these limitations, a novel technique is presented that evolves a set of periodic pattern if the regularity of the occurrence changes from that estimated pattern. The technique is based on the combination of both suffix and prefix tree patterns, to develop a multiplex tree pruning, for an activity normalized time periodicity data sequences. The integrative sequence of prefix and suffix trees is based on the threshold factor of predominant data pattern occurrence rate. The conceptual model of multiplex tree pruning technique presented in this study, in combination with the prefix and suffix tree model for pruning items identifies the regularity of all observed patterns in an efficient manner. The detailed experimental study shows strong gains in periodic pattern mining, ensure fast storage of all the time series for a specified item. Empirical studies with varied time series data obtained from bank and car data set using UCI repositories is measured and evaluated in terms of time efficiency of pruning patterns of interest, sensitivity and accuracy.
  B. Sujatha and S. Chenthur Pandian
  Extracting predominant pattern in a time series database is a major data mining problem with several applications. The existing closed sequential patterns permit us to improve efficiency without bringing down the accuracy. The narrative technique developed a previous research follows a multiplex tree pruning technique which combines both the prefix and suffix tree patterns in an activity normalized time periodicity data sequences. The combinatorial point of prefix and suffix trees is on the threshold of predominant data pattern occurrence rate which efficiently identify the regularity of all observed patterns but still obtains the interlaced unwanted data. To separate the interlaced unwanted data from the predominant pattern mining, researchers are going to implement a new technique termed Optimized Discrete Interested Pattern technique (ODIP). This technique identifies the optimal value using the repetition occurrence in the pattern. An analytical and empirical result offers an efficient and effective predominant pattern mining framework for highly dynamic online time series data. Performance of the optimized discrete interested pattern technique is measured in terms of interlaced data removal efficiency, time taken for online pattern mining based on the frequency. Experiments are conducted with online time series data obtained from research repositories of both synthetic and real data sets.
  B. Sujatha and T. Santhanam
  Problem statement: Deaf and dumb needs assistance from a technical box that takes movements of lips to identify the words. This technical article provided appropriate model implementation of flexible lip model for better visual lip reading system. Approach: From the frame sequence of words, Active Shape Model (ASM) based lip model provided local tracking and extraction of geometric lip-feature. Two geometric criteria define required geometric features and its variations in the sequence. Results: The feature established machine classification using Analytic Hierarchy Process (AHP), a relative weight finder. AHP presents weight vector to fuzzy classifier to decide the video frame sequence belonging to a respective word. Conclusion: The suggested model tested on a total of 5 different sample databases results in 83.2% accuracy over the other combinational algorithms.
 
 
 
Copyright   |   Desclaimer   |    Privacy Policy   |   Browsers   |   Accessibility