M Tech Dissertations

Permanent URI for this collectionhttp://drsr.daiict.ac.in/handle/123456789/3

Browse

Subject: search.filters.subject.Human face recognition

Search Results

Now showing 1 - 3 of 3
  • Thumbnail Image
    ItemOpen Access
    Locality preserving projection: a study and applications
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2012) Shikkenawis, Gitam; Mitra, Suman K
    Locality Preserving Projection (LPP) is a recently proposed approach for dimensionality reduction that preserves the neighbourhood information and obtains a subspace that best detects the essential data manifold structure. Currently it is widely used for finding the intrinsic dimensionality of the data which is usually of high dimension. This characteristic of LPP has made it popular among other available dimensionality reduction approaches such as Principal Component Analysis (PCA). A study on LPP reveals that it tries to preserve the information about nearest neighbours of data points, thus may lead to misclassification in the overlapping regions of two or more classes while performing data analysis. It has also been observed that the dimension reducibility capacity of conventional LPP is much less than that of PCA. A new proposal called Extended LPP (ELPP) which amicably resolves two issues mentioned above is introduced. In particular, a new weighing scheme is designed that pays importance to the data points which are at a moderate distance, in addition to the nearest points. This helps to resolve the ambiguity occurring at the overlapping regions as well as increase the reducibility capacity. LPP is used for a variety of applications for reducing the dimensions one of which is Face Recognition. Face Recognition is one of the most widely used biometric technology for person identification. Face images are represented as highdimensional pixel arrays and due to high correlation between the neighbouring pixel values; they often belong to an intrinsically low dimensional manifold. The distribution of data in a high dimensional space is non-uniform and is generally concentrated around some kind of low dimensional structures. Hence, one of the ways of performing Face Recognition is by reducing the dimensionality of the data and finding the subspace of the manifold in which face images reside. Both LPP and ELPP are used for Face and Expression Recognition tasks. As the aim is to separate the clusters in the embedded space, class membership information may add more discriminating power. With this in mind, the proposal is further extended to the supervised version of LPP (SLPP) that uses the known class labels of data points to enhance the discriminating power along with inheriting the properties of ELPP
  • Thumbnail Image
    ItemOpen Access
    Person identification using face and speech
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2012) Parmar, Ajay; Joshi, Manjunath V.
    In this thesis, we present a multimodal biometric system using face and speech features. Multimodal biometrics system uses two or more intrinsic physical or behaviour traits to provide better recognition rate than unimodal biometric systems. Face recognition is built using principal component analysis (PCA) and the Gabor filters. In Face recognition, PCA is applied to Gabor filter bank response of the face images. Speaker recognition is built using amplitude modulation - frequency modulation (AM-FM) features. AM-FM features are weighted-instantaneous frequency of the analytical signal. Finally, weighted sum of score of face and speaker recognition system is used for person identification. Performance of our system is evaluated by using ORL database for face images and ELSDSR database for speech. Experimental results show better recognition rate for the multimodal sytem when compared to unimodal system
  • Thumbnail Image
    ItemOpen Access
    Eye localization in video: a hybrid approach
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2010) Kansara, Bena; Mitra, Suman K.
    Location of eyes is an important process for operations such as orientation correction, which are necessary pre-processes for face recognition. As eyes are one of the main features of the human face, the success of facial feature analysis and face recognition depends greatly on eye detection. It is advantageous to detect eyes before other facial features because the position of other facial features can be estimated using eye position and golden ratio. Since relative position of eyes and interocular distance are nearly constant for different individuals, eye localization is also useful in face normalization. Hence, eye localization is a very important component for any face recognition system. Various approaches to eye localization have been proposed and can be classified as feature based approaches, template based approaches and appearance based approaches. Feature based methods explore eye characteristics - such as edge and intensity of iris - to identify some distinctive features around the eyes. In template based methods, a generic model of eye shape is designed; this template is then matched to the face image pixel by pixel to find the eyes. Appearance based methods detect eyes based on their photometric appearance. Template based and appearance based methods can detect eyes accurately but they are not efficient when considering time factor while feature based methods are efficient but do not give accurate results. So, by combining feature based method with template based or appearance based method, we can get better results. In the proposed algorithm, we have combined feature based eye LEM approach proposed by Mihir Jain, Suman K. Mitra, Naresh D. Jotwani in 2008 and appearance based Bayesian classi_er approach proposed by Everingham M., Zisserman A. in 2006 to achieve eye localization. The work of localizing eyes in a video is motivated by some of the applications where eye localization can serve very useful purpose such as to find drowsiness of a person driving a car, eye based control of computer systems for people with motor difficulties. To carry out eye localization, after doing some preprocessing which include frame separation from video and to convert it into gray-scale images, the proposed algorithm is applied on each of these frames. For the experimentation, we have taken videos of few people in the normal blink condition as well as in the sleepy condition. All the videos have been taken in the lab environment. To check the accuracy of the proposed algorithm, we have performed various tests, namely, Wilcoxon signed rank test, Mann-Whitney U test, Kolmogorov-Smirnov test, Sensitivity and False alarm rate tests. And the results of these tests show that the proposed algorithm proves to be quite accurate in localizing the eyes in a video. All the experiments have been carried out in MATLAB.