Kuntal Sengupta - Academia.edu (original) (raw)

Uploads

Papers by Kuntal Sengupta

Proceedings of International Symposium on Computer Vision - ISCV

We present two new methods for the semiautonomous registration of spatial data from disparate sou... more We present two new methods for the semiautonomous registration of spatial data from disparate sources, in particular image and elevation data. One method stems from the synthetic image idea first proposed by Horn and Bachman, while the second is a line alignment technique. The methods estimate the affine transform between digital terrain elevation data (DTED) and aerial photographs from the National High Altitude Photography (NHAP) program. We also propose a novel measure of transform consistency and demonstrate the quality of the results. Both systems are accessed via graphical interfaces which are not described here.

Proceedings 1998 IEEE and ATR Workshop on Computer Vision for Virtual Reality Based Human Communications

At ATR Media Integration & Communications Research Laboratories, our research group is aiming... more At ATR Media Integration & Communications Research Laboratories, our research group is aiming at realizing virtual communication environments in which the users at remote sites can feel that they are virtually co-located. This paper presents the ongoing research in two subareas chosen to address this problem: (a) Generating human images in virtual scenes. (b) An image based approach for generation of (background) scenes. In the virtual metamorphosis system, 3D models of characters are created in advance. For metamorphosis, the facial expressions and body posture of a person are detected in real-time from the face images and monocular thermal images of the person, respectively. The detected expressions and body movements are reproduced by deforming the 3D models. Demonstrations using 3D models of a Kabuki actor and a dinosaur show good performance. For novel view generation, we use as affine coordinate based reprojection scheme, which is also extended to merging real and synthetic objects.

Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2004. CVPR 2004.

Proceedings of 13th International Conference on Pattern Recognition, 1996

We present an eigenvalue or spectral representation for CAD models to be used in conjunction with... more We present an eigenvalue or spectral representation for CAD models to be used in conjunction with the more traditional attributed graph based representation of these models. The eigenvalues provide a gross description of the structure of the objects, and help to divide a large modelbase into structurally homogeneous partitions. Models in each partition are next hierarchically organized according to the algorithm presented in Sengupta and Boyer (1995). In recognition, gross features computed from a hypothesized object in a range image are used to prune the modelbase by selecting a few "favorable" partitions in which the correct object model is likely to lie. The partitioning experiments presented here are for real range images using a modelbase of 125 CAD objects with planar, cylindrical, and spherical surfaces.

Addresses two problems related to organizing CAD models represented as random parametric structur... more Addresses two problems related to organizing CAD models represented as random parametric structural descriptions (RPSDs). First, the authors present a method to estimate the probability information for RPSDs from CAD models. The randomness arising from viewpoint variation is captured in the random variables corresponding to the attributes of the primitives and the relationship tuples. Next, and more significantly, they present a method to subdivide a large, heterogeneous set of models into smaller, structurally homogeneous subsets based on the distribution of the eigenvalues of property matrices derived from 2/spl frac12/ descriptions of the CAD models. Once this is done, each sublibrary of RPSDs can be hierarchically organized.<<ETX>>

... as input comes from the cheap and widely available Creative Webcam 3 desktop ... MANIPULATION... more ... as input comes from the cheap and widely available Creative Webcam 3 desktop ... MANIPULATION OF REMOTE 3D AVATAR THROUGH FACIAL FEATURE DETECTION AND REAL TIME TRACKING ... stable in the sense that it may change shape during speaking and moving of ...

... modality. The purpose of this research is to show that it is possible to recognize color thou... more ... modality. The purpose of this research is to show that it is possible to recognize color thoughts solely from EEG signals and use this, together with gesture recognition, to interface, communicate, and modify the virtual world. The ...

We discuss a novel nonlinear generalized type of classical stochastic hidden Markov model (HMM) b... more We discuss a novel nonlinear generalized type of classical stochastic hidden Markov model (HMM) based on fuzzy integrals and fuzzy measures, with its first application to speech recognition. The new model relaxes the additivity constraint of probability measures by using fuzzy measures whose key property is monotonicity with respect to set inclusion. This is far more relaxed and flexible than the additivity constraint of probability measures. Furthermore, the generalized fuzzy HMM does not require the statistical independence assumption of classical stochastic HMMs. Another advantage is that the generalized model reduces to classical HMM by utilization of Choquet integrals and probability measures, thus allowing full compatibility with traditional HMM systems. This generalized form of the classical HMM can be used to design different classifiers, and these classifiers can then be combined to obtain a better recognition performance. Hence the most important point to note is that the generalized fuzzy HMM allows the creation of multiple classifiers to increase system recognition accuracy. In this research, the new model is applied to acoustic speech recognition for the first time, and its performance is compared with the classical HMM. The experimental results showed that the new model improved the recognition accuracy, and furthermore increased the training speed of the HMMs. It was found that creating a combined fuzzy and classical speech recognition system significantly increased recognition accuracy.

2022 IEEE International Joint Conference on Biometrics (IJCB)

In this article we have proposed an abstract representation for a sequence using a constant sized... more In this article we have proposed an abstract representation for a sequence using a constant sized 3D matrix. Subsequently the representation may be utilized for many analytical purposes. We have attempted to use it for comparing sequences, and analyzed the method's asymptotic complexity. Providing a metric for sequence comparison is an underlying operation to many bioinformatics applications. In order to show the effectiveness of the proposed sequence comparison technique we have generated some phylogeny over two sets of bio-sequences and compared them with the ones available in literature. The results prove that our technique is comparable to the standard ones. The technique, called the correlogram-based method, is borrowed from the image analysis area. We have also done some experiments with synthetically generated sequences in order to compare correlogram-based method with the well-known dynamic programming method. Finally, we have discussed some other possibilities on how our method can be used or extended.

In a phishing attack, a perpetrator attempts to obtain the online credentials of a user by impers... more In a phishing attack, a perpetrator attempts to obtain the online credentials of a user by impersonating a trusted entity such as a bank, email service provider, etc. Sophisticated phishers attempt to deceive spam filters by structuring the visual look-and-feel of their fake emails to be nearly but not precisely identical to emails sent by a trusted entity, such that spam filters allow the fake email to reach a user’s inbox. This disclosure describes use of hand-crafted visual features of emails or landing pages, and classification based on earth-mover’s distance, to assess the visual similarity of genuine and phished emails. The techniques detect visual near-duplicates of a trusted entity’s email and thereby achieve resilience against phishing attacks

Pattern Recognition Letters, 2002

Proceedings 15th International Conference on Pattern Recognition. ICPR-2000