Pub Date : 2013-12-01DOI: 10.1109/NCVPRIPG.2013.6776245
S. K. Chatterjee, I. Chakrabarti
The present paper proposes a fast algorithm and its VLSI architecture for fast quarter pixel (QP) accurate motion estimation (ME). The proposed algorithm is based on the distribution of the QP motion vectors (MVs) around the half pixel MV. The proposed algorithm efficiently explores the most likely QP locations and therefore skips the unlikely ones. The number of QP search locations for the proposed algorithm is reduced by 50% compared to the original full search method but results in only about 0.12 dB peak signal to noise ratio degradation. The VLSI architecture of the proposed algorithm theoretically can process thirty three 1280×720 HDTV frames per second. The power consumption of the proposed architecture is also reduced by 15? compared to a recently reported architecture.
{"title":"Algorithm and architecture for quarter pixel motion estimation for H.264/AVC","authors":"S. K. Chatterjee, I. Chakrabarti","doi":"10.1109/NCVPRIPG.2013.6776245","DOIUrl":"https://doi.org/10.1109/NCVPRIPG.2013.6776245","url":null,"abstract":"The present paper proposes a fast algorithm and its VLSI architecture for fast quarter pixel (QP) accurate motion estimation (ME). The proposed algorithm is based on the distribution of the QP motion vectors (MVs) around the half pixel MV. The proposed algorithm efficiently explores the most likely QP locations and therefore skips the unlikely ones. The number of QP search locations for the proposed algorithm is reduced by 50% compared to the original full search method but results in only about 0.12 dB peak signal to noise ratio degradation. The VLSI architecture of the proposed algorithm theoretically can process thirty three 1280×720 HDTV frames per second. The power consumption of the proposed architecture is also reduced by 15? compared to a recently reported architecture.","PeriodicalId":436402,"journal":{"name":"2013 Fourth National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG)","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123791583","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-01DOI: 10.1109/NCVPRIPG.2013.6776168
Sourabh Prajapati, P J Narayanan
We present an approach for fast registration of a Global Articulated 3D Model to RGBD data from Kinect. Our approach uses geometry based matching of rigid parts of the articulated objects in depth images. The registration is performed in a parametric space of transformations independently for each segment. The time for registering each frame with the global model is reduced greatly using this method. We experimented the algorithm with different articulated object datasets and obtained significantly low execution time as compared to ICP algorithm when applied on each rigid part of the articulated object.
{"title":"Fast registration of articulated objects from depth images","authors":"Sourabh Prajapati, P J Narayanan","doi":"10.1109/NCVPRIPG.2013.6776168","DOIUrl":"https://doi.org/10.1109/NCVPRIPG.2013.6776168","url":null,"abstract":"We present an approach for fast registration of a Global Articulated 3D Model to RGBD data from Kinect. Our approach uses geometry based matching of rigid parts of the articulated objects in depth images. The registration is performed in a parametric space of transformations independently for each segment. The time for registering each frame with the global model is reduced greatly using this method. We experimented the algorithm with different articulated object datasets and obtained significantly low execution time as compared to ICP algorithm when applied on each rigid part of the articulated object.","PeriodicalId":436402,"journal":{"name":"2013 Fourth National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125252325","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-01DOI: 10.1109/NCVPRIPG.2013.6776220
S. Winberg, S. Katz, A. Mishra
Computer-aided plant identification combines computer vision and pattern recognition. The Cape Floristic Kingdom is the most varied of plant kingdoms, comprising thousands of species of fynbos plants. While it is easier to classify fynbos when they are flowering, mostly flower for only a few weeks in a year. This paper concerns an image processing application for automatic identification of certain fynbos using leaf photographs. The architecture of this application is overviewed prior to focusing on the leaf recognition operations, and how these were experimentally tested using a series of experiments, culminating in a comprehensive test to measure identification accuracy, effectiveness of the online user interface, and the processing speed. Our conclusions reflect on the overall effectiveness of the application and our plans to take it further.
{"title":"Fynbos leaf online plant recognition application","authors":"S. Winberg, S. Katz, A. Mishra","doi":"10.1109/NCVPRIPG.2013.6776220","DOIUrl":"https://doi.org/10.1109/NCVPRIPG.2013.6776220","url":null,"abstract":"Computer-aided plant identification combines computer vision and pattern recognition. The Cape Floristic Kingdom is the most varied of plant kingdoms, comprising thousands of species of fynbos plants. While it is easier to classify fynbos when they are flowering, mostly flower for only a few weeks in a year. This paper concerns an image processing application for automatic identification of certain fynbos using leaf photographs. The architecture of this application is overviewed prior to focusing on the leaf recognition operations, and how these were experimentally tested using a series of experiments, culminating in a comprehensive test to measure identification accuracy, effectiveness of the online user interface, and the processing speed. Our conclusions reflect on the overall effectiveness of the application and our plans to take it further.","PeriodicalId":436402,"journal":{"name":"2013 Fourth National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128267122","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-01DOI: 10.1109/NCVPRIPG.2013.6776270
Sudeshna Roy, Sukhendu Das
Bottom-up saliency detection algorithms identify distinct regions in an image, with rare occurrence of local feature distributions. Notable among those works published recently, use local and global contrast, spectral analysis of the entire image or graph based feature mapping. Whereas, we propose a novel unsupervised method using color compactness and statistical modeling of the background cues, to segment the salient foreground region and thus the salient object. At the first stage of processing, the image is segmented into clusters using color feature. First component proposed for our saliency measure combines disparity in color and spatial distance between patches. In addition to rarity of feature, we propose another component for saliency computation that estimates the divergence of the color of a patch from those in the set of patches at the boundary of the image, representing the background. Combination of these two complementary components provides a much improved saliency map for salient object detection.We verify the performance of our proposed method of saliency detection on two popular benchmark datasets, with one or more salient regions and diverse saliency characteristics. Experimental results show that our method out-performs many existing state-of-the-art methods.
{"title":"Spatial variance of color and boundary statistics for salient object detection","authors":"Sudeshna Roy, Sukhendu Das","doi":"10.1109/NCVPRIPG.2013.6776270","DOIUrl":"https://doi.org/10.1109/NCVPRIPG.2013.6776270","url":null,"abstract":"Bottom-up saliency detection algorithms identify distinct regions in an image, with rare occurrence of local feature distributions. Notable among those works published recently, use local and global contrast, spectral analysis of the entire image or graph based feature mapping. Whereas, we propose a novel unsupervised method using color compactness and statistical modeling of the background cues, to segment the salient foreground region and thus the salient object. At the first stage of processing, the image is segmented into clusters using color feature. First component proposed for our saliency measure combines disparity in color and spatial distance between patches. In addition to rarity of feature, we propose another component for saliency computation that estimates the divergence of the color of a patch from those in the set of patches at the boundary of the image, representing the background. Combination of these two complementary components provides a much improved saliency map for salient object detection.We verify the performance of our proposed method of saliency detection on two popular benchmark datasets, with one or more salient regions and diverse saliency characteristics. Experimental results show that our method out-performs many existing state-of-the-art methods.","PeriodicalId":436402,"journal":{"name":"2013 Fourth National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129697933","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-01DOI: 10.1109/NCVPRIPG.2013.6776167
Sailik Sengupta, P. Chaudhuri
In this paper we present a system to create 3D garments from 2D patterns. Once placed over the 3D character, our system can quickly stitch the patterns into the 3D garment. The stitched cloth is then simulated to obtain the drape of the garment over the character. Our system can accurately and efficiently resolve cloth-body and cloth-cloth collisions.
{"title":"Virtual garment simulation","authors":"Sailik Sengupta, P. Chaudhuri","doi":"10.1109/NCVPRIPG.2013.6776167","DOIUrl":"https://doi.org/10.1109/NCVPRIPG.2013.6776167","url":null,"abstract":"In this paper we present a system to create 3D garments from 2D patterns. Once placed over the 3D character, our system can quickly stitch the patterns into the 3D garment. The stitched cloth is then simulated to obtain the drape of the garment over the character. Our system can accurately and efficiently resolve cloth-body and cloth-cloth collisions.","PeriodicalId":436402,"journal":{"name":"2013 Fourth National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125705542","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-01DOI: 10.1109/NCVPRIPG.2013.6776213
Suranjana Samanta, T. Selvan, Sukhendu Das
In this paper, we propose a method to improve the results of clustering in a target domain, using significant information from an auxiliary (source) domain dataset. The applicability of this method concerns the field of transfer learning (or domain adaptation), where the performance of a task (say, classification using clustering) in one domain is improved using knowledge obtained from a similar domain. We propose two unsupervised methods of cross-domain clustering and show results on two different categories of benchmark datasets, both having difference in density distributions over the pair of domains. In the first method, we propose an iterative framework, where the clustering in the target domain is influenced by the clusters formed in the source domain and vice-versa. Similarity/dissimilarity measures have been appropriately formulated using Euclidean distance and Bregman Divergence, for cross-domain clustering. In the second method, we perform clustering in the target domain by estimating local density computed using a non-parametric (NP) density estimator (due to less number of samples). Prior to clustering, the NP-density scattering in the target domain is modified using information of cluster density distribution in source domain. Results shown on real-world datasets suggest that the proposed methods of cross-domain clustering are comparable to the recent start-of-the-art work.
{"title":"Cross-domain clustering performed by transfer of knowledge across domains","authors":"Suranjana Samanta, T. Selvan, Sukhendu Das","doi":"10.1109/NCVPRIPG.2013.6776213","DOIUrl":"https://doi.org/10.1109/NCVPRIPG.2013.6776213","url":null,"abstract":"In this paper, we propose a method to improve the results of clustering in a target domain, using significant information from an auxiliary (source) domain dataset. The applicability of this method concerns the field of transfer learning (or domain adaptation), where the performance of a task (say, classification using clustering) in one domain is improved using knowledge obtained from a similar domain. We propose two unsupervised methods of cross-domain clustering and show results on two different categories of benchmark datasets, both having difference in density distributions over the pair of domains. In the first method, we propose an iterative framework, where the clustering in the target domain is influenced by the clusters formed in the source domain and vice-versa. Similarity/dissimilarity measures have been appropriately formulated using Euclidean distance and Bregman Divergence, for cross-domain clustering. In the second method, we perform clustering in the target domain by estimating local density computed using a non-parametric (NP) density estimator (due to less number of samples). Prior to clustering, the NP-density scattering in the target domain is modified using information of cluster density distribution in source domain. Results shown on real-world datasets suggest that the proposed methods of cross-domain clustering are comparable to the recent start-of-the-art work.","PeriodicalId":436402,"journal":{"name":"2013 Fourth National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG)","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130012940","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-01DOI: 10.1109/NCVPRIPG.2013.6776250
V. B. Surya Prasath, J. Moreno
Anisotropic diffusion based schemes are widely used in image smoothing and noise removal. Typically, the partial differential equation (PDE) used is based on computing image gradients or isotropically smoothed version of the gradient image. To improve the denoising capability of such nonlinear anisotropic diffusion schemes, we introduce a multi-direction based discretization along with a selection strategy for choosing the best direction of possible edge pixels. This strategy avoids the directionality based bias which can over-smooth features that are not aligned with the coordinate axis. The proposed hybrid discretization scheme helps in preserving multi-scale features present in the images via selective smoothing of the PDE. Experimental results indicate such an adaptive modification provides improved restoration results on noisy images.
{"title":"Feature preserving anisotropic diffusion for image restoration","authors":"V. B. Surya Prasath, J. Moreno","doi":"10.1109/NCVPRIPG.2013.6776250","DOIUrl":"https://doi.org/10.1109/NCVPRIPG.2013.6776250","url":null,"abstract":"Anisotropic diffusion based schemes are widely used in image smoothing and noise removal. Typically, the partial differential equation (PDE) used is based on computing image gradients or isotropically smoothed version of the gradient image. To improve the denoising capability of such nonlinear anisotropic diffusion schemes, we introduce a multi-direction based discretization along with a selection strategy for choosing the best direction of possible edge pixels. This strategy avoids the directionality based bias which can over-smooth features that are not aligned with the coordinate axis. The proposed hybrid discretization scheme helps in preserving multi-scale features present in the images via selective smoothing of the PDE. Experimental results indicate such an adaptive modification provides improved restoration results on noisy images.","PeriodicalId":436402,"journal":{"name":"2013 Fourth National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG)","volume":"235 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114417841","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-01DOI: 10.1109/NCVPRIPG.2013.6776231
Arvind R. Yadav, M. Dewal, R. S. Anand, Sangeeta Gupta
In this paper, an approach for the classification of different hardwood species of open access database, using texture feature extraction and supervised machine learning technique has been implemented. Edges of complex cellular structure of microscopic images of hardwood are enhanced with the application of Gabor filter, and Gray Level Co-occurrence Matrix (GLCM) as an effective texture feature extraction technique is being revalidated. About, 44 features have been extracted from GLCM; these features have been further normalized in the range [0.1, 1]. Multilayer Perceptron Backpropagation Artificial Neural Network have been used for classification. Experiments conducted on 25 wood species have resulted in recognition accuracy of about 88.60% and 92.60% using Levenberg-Marquardt backpropagation training function with two different datasets for training, validation and testing ratio (70%, 15%, 15% and 80%, 10%, 10%) respectively. Proposed methodology can be extended with optimized machine learning techniques for online identification of wood.
{"title":"Classification of hardwood species using ANN classifier","authors":"Arvind R. Yadav, M. Dewal, R. S. Anand, Sangeeta Gupta","doi":"10.1109/NCVPRIPG.2013.6776231","DOIUrl":"https://doi.org/10.1109/NCVPRIPG.2013.6776231","url":null,"abstract":"In this paper, an approach for the classification of different hardwood species of open access database, using texture feature extraction and supervised machine learning technique has been implemented. Edges of complex cellular structure of microscopic images of hardwood are enhanced with the application of Gabor filter, and Gray Level Co-occurrence Matrix (GLCM) as an effective texture feature extraction technique is being revalidated. About, 44 features have been extracted from GLCM; these features have been further normalized in the range [0.1, 1]. Multilayer Perceptron Backpropagation Artificial Neural Network have been used for classification. Experiments conducted on 25 wood species have resulted in recognition accuracy of about 88.60% and 92.60% using Levenberg-Marquardt backpropagation training function with two different datasets for training, validation and testing ratio (70%, 15%, 15% and 80%, 10%, 10%) respectively. Proposed methodology can be extended with optimized machine learning techniques for online identification of wood.","PeriodicalId":436402,"journal":{"name":"2013 Fourth National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132750015","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-01DOI: 10.1109/NCVPRIPG.2013.6776200
Jingyong Su, Anuj Srivastava, F. Souza, Sudeep Sarkar
An important problem in speech, and generally activity, recognition is to develop analyses that are invariant to the execution rates. We introduce a theoretical framework that provides a parametrization-invariant metric for comparing parametrized paths on Riemannian manifolds. Treating instances of activities as parametrized paths on a Riemannian manifold of covariance matrices, we apply this framework to the problem of visual speech recognition from image sequences. We represent each sequence as a path on the space of covariance matrices, each covariance matrix capturing spatial variability of visual features in a frame, and perform simultaneous pairwise temporal alignment and comparison of paths. This removes the temporal variability and helps provide a robust metric for visual speech classification. We evaluated this idea on the OuluVS database and the rank-1 nearest neighbor classification rate improves from 32% to 57% due to temporal alignment.
{"title":"Rate-invariant comparisons of covariance paths for visual speech recognition","authors":"Jingyong Su, Anuj Srivastava, F. Souza, Sudeep Sarkar","doi":"10.1109/NCVPRIPG.2013.6776200","DOIUrl":"https://doi.org/10.1109/NCVPRIPG.2013.6776200","url":null,"abstract":"An important problem in speech, and generally activity, recognition is to develop analyses that are invariant to the execution rates. We introduce a theoretical framework that provides a parametrization-invariant metric for comparing parametrized paths on Riemannian manifolds. Treating instances of activities as parametrized paths on a Riemannian manifold of covariance matrices, we apply this framework to the problem of visual speech recognition from image sequences. We represent each sequence as a path on the space of covariance matrices, each covariance matrix capturing spatial variability of visual features in a frame, and perform simultaneous pairwise temporal alignment and comparison of paths. This removes the temporal variability and helps provide a robust metric for visual speech classification. We evaluated this idea on the OuluVS database and the rank-1 nearest neighbor classification rate improves from 32% to 57% due to temporal alignment.","PeriodicalId":436402,"journal":{"name":"2013 Fourth National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG)","volume":"93 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134290329","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-01DOI: 10.1109/NCVPRIPG.2013.6776249
Laishram Rahul, Salam Nandakishor, L. J. Singh, S. K. Dutta
This paper aims to discuss the implementation of phoneme based Manipuri Keyword Spotting System (MKWSS). Manipuri is a scheduled Indian language of Tibeto-Burman origin. Around 5 hours of read speech are collected from 4 male and 6 female speakers for development of database of MKWSS. The symbols of International Phonetic Alphabet (IPA)(revised in 2005) are used during the transcription of the data. A five state left to right Hidden Markov Model (HMM) with 32 mixture continuous density diagonal covariance Gaussian Mixture Model (GMM) per state is used to build a model for each phonetic unit. We have used HMM tool kit (HTK), version 3.4 for modeling the system. The system can recognize 29 phonemes and a non-speech event (silence) and will detect the present keywords formed by these phonemes. Continuous Speech data have been collected from 5 males and 8 females for analysing the performance of the system. The performance of the system depends on the ability of detection of the keywords. An overall performance of 65.24% is obtained from the phoneme based MKWSS.
{"title":"Design of Manipuri Keywords Spotting System using HMM","authors":"Laishram Rahul, Salam Nandakishor, L. J. Singh, S. K. Dutta","doi":"10.1109/NCVPRIPG.2013.6776249","DOIUrl":"https://doi.org/10.1109/NCVPRIPG.2013.6776249","url":null,"abstract":"This paper aims to discuss the implementation of phoneme based Manipuri Keyword Spotting System (MKWSS). Manipuri is a scheduled Indian language of Tibeto-Burman origin. Around 5 hours of read speech are collected from 4 male and 6 female speakers for development of database of MKWSS. The symbols of International Phonetic Alphabet (IPA)(revised in 2005) are used during the transcription of the data. A five state left to right Hidden Markov Model (HMM) with 32 mixture continuous density diagonal covariance Gaussian Mixture Model (GMM) per state is used to build a model for each phonetic unit. We have used HMM tool kit (HTK), version 3.4 for modeling the system. The system can recognize 29 phonemes and a non-speech event (silence) and will detect the present keywords formed by these phonemes. Continuous Speech data have been collected from 5 males and 8 females for analysing the performance of the system. The performance of the system depends on the ability of detection of the keywords. An overall performance of 65.24% is obtained from the phoneme based MKWSS.","PeriodicalId":436402,"journal":{"name":"2013 Fourth National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133495676","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：