Buildings are exposed to damage and deterioration during their life cycle. So, damage assessment plays an important role in Structural stability. Cracks in the structures are of common occurrence, hence early detection of cracks is necessary. Damages like cracks can be detected using Microwave Imaging of the columns. Damages like Horizontal and vertical cracks are determined by training the Bayesian classifier and the Artificial Neural Networks. Both these approaches are required as Structural health to be monitored for predicting damages in columns. Crack detection system is built in columns of civil structures based on Artificial Neural Network and Bayesian Classifiers, which are constructed upon probabilistic pattern recognition and data modelling. The frequency data was collected from 12 microwave sensors for 30 positions of column and is required to train and test the mathematical models. Since, mean and covariance of the statistical data are well known features used in feature extraction. Finally, performance analysis of the models has been provided in terms of Crack Error Rate (CER) justifies that dynamic modelling using ANN yields better results than Bayesian Classifiers and this can also be used in developing Automatic Crack detection systems of civil structures.

We build an emotion recognition system based on Artificial Neural Network (ANN) and compare the same with the one based upon the Hidden Markov Modeling (HMM) scheme. Both the systems were built upon probabilistic pattern recognition and acoustic phonetic modelling approaches. Since our native language is Kannada, a very rich South Indian language, we have used utterance in Kannada to train and test the schemes. Since Mel-Frequency Cepstral Coefficients (MFCC) are well known acoustic features of speech , we have used the Delta MFCC (DMFCC) and the Double Delta MFCC (DDMFCC) vectors in speech feature extraction. Finally, performance analysis of these models in terms of Emotion Error Rate (EER) justifies the fact that modeling using the ANN yields better results over other modeling schemes and can be used in developing Automatic Emotion Recognition systems.

We build an emotion detection system based on Bayesian multivariate modeling and compare the same with the one based upon Hidden Markov Modeling (HMM) scheme. Both the systems were built upon probabilistic pattern recognition and acoustic phonetic recognition. Since our native language is Kannada, one of very rich South Indian language, we have used 4 Emotions uttered in Kannada to train and test the schemes. Since Mel-Frequency Cepstral Coefficients (MFCC) are well known acoustic features of speech [1][2][4], we have used the same in speech feature extraction. Finally performance analysis of these models in terms of Emotion Error Rate (EER) justifies the fact that Dynamic modeling using HMM yields better results over other modeling schemes and can be used in developing Automatic Speech Recognition systems.

This work aims to make use of compressed sensing to exploit the redundant nature hidden in the image and reduce the computational complexity involved in DRR generation. As a result, radiation risk to the patient can be reduced whilst maintaining an acceptable level of accuracy thus resulting in speed-up in DRR generation using the multi-resolution approach compared to the conventional ray casting approach. Also in this research, different gradient based similarity metrics were compared on the basis of accuracy to achieve robustness against image content mismatch.

Recent advancements in spine surgery use image guided intervention systems to enhance surgical accuracy for incorporating pedicle screw constructs. Two-dimensional (2D) to three-dimensional (3D) image registration helps in treatment planning and verification by mapping the Digitally Reconstructed Radiograph (DRR) images, rendered from 3D CT volumetric data to 2D C-arm images. Conventional approach of ray casting which is followed for DRR generation helps to reduce the risk of wrong site surgery and provide superior accuracy. However, it consumes significant amount of time due to complex computational intensity. An effective method is proposed to reduce the time taken for DRR generation by exploiting the redundant information hidden in the CT images with the help of sparse sampling and multi-resolution using Discrete Wavelet Transform. Beginning with 2D to 3D image registration methods, this study emphasizes on acceleration of DRR by spatial and temporal resolution of sagittal slices of the CT data, while maintaining the quality consistent with that needed for image registration. This method was found to generate DRR at a much faster rate of 84%.

We build and compare phoneme recognition systems based on Bayesian Multivariate Modeling scheme and Hidden Markov Modeling (HMM) scheme. Both models were built by using Stochastic pattern recognition and Acoustic phonetic schemes to recognise phonemes. Since our native language is Kannada, a rich South Indian Language, we have used 15 Kannada phonemes to train and test these models. Since Mel - Frequency Cepstral Coefficients (MFCC) are well known Acoustic features of speech, we have used the same in speech feature extraction. Finally performance analysis of both models in terms of Phoneme Error Rate (PER) justifies the fact that Dynamic modeling yields better results over Static modeling and can be used in developing Automatic Speech Recognition systems.

We build and compare phoneme recognition systems based on Gaussian Mixture Modeling (GMM) which is a static modeling scheme and Hidden Markov Modeling (HMM) which is a Dynamic modeling scheme. Both models were built by using Stochastic pattern recognition and Acoustic phonetic schemes to recognise phonemes. Since our native language is Kannada, a rich South Indian Language, we have used 15 Kannada phonemes to train and test these models. Since Mel - Frequency Cepstral Coefficients (MFCC) are well known Acoustic features of speech, we have used the same in speech feature extraction. Finally performance analysis of both models in terms of Phoneme Error Rate (PER) justifies the fact that Dynamic modeling yields better results over Static modeling and can be used in developing Automatic Speech Recognition systems.

We build an automatic phoneme recognition system based on Bayesian Multivariate Modeling which is a static scheme. Phoneme models were built by using stochastic pattern recognition and acoustic phonetic schemes to recognise phonemes. Since our native language is Kannada, a rich South Indian Language, we have used 15 Kannada phonemes to train and test these models. As Mel – Frequency Cepstral Coefficients (MFCC) are well known acoustic features of speech, we have used the same in speech feature extraction. Finally performance analysis of models in terms of Phoneme Error Rate (PER) justifies the fact that though static modeling yields good results, improvization is necessary in order to use it in developing Automatic Speech Recognition systems. KeywordsBayesian Classification, Kannada, MFCC, Pattern Recognition; PER, Phoneme Modeling