Showing papers by "Santanu Chaudhury published in 2010"

Soft-computing based diagnostic tool for analyzing demyelination in magnetic resonance images

Abstract: This paper proposes a soft-computing based diagnostic tool for analyzing (white matter changes) demyelination due to radiation therapy given to brain tumor cases. The tool exploits the pattern of changes in gray level distribution using a temporal sequence of magnetic resonance (MR) images. Appearance of white matter changes due to demyelination varies from patient to patient. Further, there exists inherent impreciseness in the white matter change patterns. These characteristics make use of fuzzy features well suited for describing image based temporal patterns. Correlation between these temporal patterns and actual onset of demyelination can be captured by fuzzy rules because of the inherent uncertainty associated with changes in gray level pattern in the image and occurrence of the disease. The tool is based on hybrid approach of two popular approaches of genetic algorithm based machine learning (GBML) techniques namely Michigan and Pittsburgh approach. The genetic algorithm (GA) based machine learning tool generates an optimized rule set to indicate positive (P), negative (N) or doubtful (D) cases of demyelination.

Use of MKL as symbol classifier for Gujarati character recognition

Abstract: The present work is part of ongoing effort to improve the performance of Gujarati character recognition. In the recent advancement in kernel methods, the novel concept of multiple kernel learning(MKL) has given improved results for many problems. In this paper, we present novel application of MKL for Gujarati character recognition. We have applied three different feature representations for symbols obtained after zone wise segmentation of Gujarati text. The MKL based classification is proposed, where the MKL is used for learning optimal combination of different features for classification. In addition MKL based classification results for different features is also presented. The multiclass classification is performed in Decision DAG framework. The comparison results in 1-Vs-1 framework and using KNN classifier is also presented. The experiments have shown substantial improvement in earlier results.

Distributed calibration of pan-tilt camera network using multi-layered belief propagation

Abstract: In this paper, we present a technique for distributed self-calibration of pan-tilt camera network using multi-layered belief propagation. Our goal is to obtain globally consistent estimates of the camera parameters for each camera with respect to a global world coordinate system. The network configuration changes with time as the cameras can pan and tilt. We also give a distributed algorithm for automatically finding which cameras have overlapping views at a certain point in time. We argue that using belief propagation it is sufficient to have correspondences between three cameras at a time for calibrating a larger set of (static) cameras with overlapping views. Our method gives an accurate and globally consistent estimate of the camera parameters of each camera in the network.

Preservation of intangible heritage: a case-study of indian classical dance

Abstract: Cultural heritage is encoded in a variety of forms. The task of preserving heritage involves preserving the tangible and intangible resources that broadly define that heritage. A significant aspect of intangible heritage resources are performing arts which include classical dance and music. Digital heritage resources include heritage artefacts in digitized form as well as the background knowledge that puts them in perspective. We present an ontology based approach to capture and preserve the knowledge with digital heritage artefacts. Since the artefacts are generally preserved in multimedia format, we propose the use of Multimedia Web Ontology (MOWL) that supports probabilistic reasoning with media properties of domain concepts, to encode the domain knowledge. We propose an architectural framework that includes a method to construct the ontology with a labelled set of training data and use of the ontology to automatically annotate new instances of digital heritage artefacts. The annotations enable creation of a semantic navigation environment in a cultural heritage repository. We have realized a proof of concept in the domain of Indian Classical Dance and present some results.

Identity Determination with Offline Handwritten Input Using Multi Kernel Feature Combination

Abstract: The paper presents three novel features for handwritten data based identity recognition. A novel framework for combining the features for identification is presented. The framework combines the features in kernel space in MKL based framework. The application of features individually and in combination is presented for writer recognition and signature verification. The writer recognition results have been presented for Devanagari script input and signature verification results have been presented for open dataset [1]. The experiments have shown encouraging results.

Embedded Implementation of Change Detection Algorithm for Smart Camera Applications

Abstract: Smart cameras are important components in any Human Computer Interaction. In any remote surveillance scenario, smart cameras have to take intelligent decisions to select frames of interest to minimize communication and processing overhead. A clustering based change detection algorithm has been implemented in our smart camera system for filtering frames with significant changes. In this paper we propose a platform based framework for implementing clustering based change detection algorithm using HW-SW co-design based methodology. The complete system is implemented on Xilinx XUP Virtex-II Pro FPGA board. The overall algorithm is running on PowerPC405 and some of the blocks which are computationally intensive and more frequently called are implemented as custom IP using VHDL. Total gate count of the design is 2699K.

Road traffic model using distributed camera network

Abstract: Traffic monitoring/prediction using a distributed camera network is presented in this paper. The activities on each road link are monitored and features are derived to identify the pattern. Then it is learnt, classified, predicted and communicated to neighboring road links. We used GMM-EM based classification and HMM based prediction. Optimum path is determined by assigning proportional weights to the predicted states of the connected road links. The proposed method is neither based on tracking nor on vehicle detection. Apart from this the method is flexible, adaptive, robust and computationally light. Unlike the existing methods it does not assume or draws analogies of traffic moving as particles, neither does it impose restriction on road conditions or road tributaries and distributaries. The model is validated using traffic simulator and tested on real road network.

Document Image Retrieval Using Feature Combination in Kernel Space

Abstract: The paper presents application of multiple features for word based document image indexing and retrieval. A novel framework to perform Multiple Kernel Learning for indexing using the Kernel based Distance Based Hashing is proposed. The Genetic Algorithm based framework is used for optimization. Two different features representing the structural organization of word shape are defined. The optimal combination of both the features for indexing is learned by performing MKL. The retrieval results for document collection belonging to Devanagari script are presented.

Restricted affine motion compensation in video coding using particle filtering

Abstract: We propose a novel particle filter-based motion compensation strategy for video coding. We use a higher order linear model in place of the traditional translational model used in standards such as H.264. The measurement/observation process in the particle filter is a computationally efficient mechanism as opposed to traditional search methods. We use a multi-resolution framework for efficient parameter estimation. Results of our experimentation show reduced residual energy and better PSNR as compared to traditional video coding methods, especially in regions of complex motion such as zooming and rotation.

Distributed framework for composite event recognition in a calibrated pan-tilt camera network

Abstract: In this paper, we propose a real-time distributed framework for composite event recognition in a calibrated pan-tilt camera network. A composite event comprises of events that occur simultaneously or sequentially at different locations across time. Distributed composite event recognition requires distributed multi-camera multi-object tracking and distributed multi-camera event recognition. We apply belief propagation to reach a consensus on the global identities of the objects in the pan-tilt camera network and to arrive at a consensus on the event recognized by multiple cameras simultaneously observing it. We propose a hidden Markov model based approach for composite event recognition. We also propose a novel probabilistic Latent Semantic Analysis based algorithm for pair-wise interaction recognition and present an application of our distributed composite event recognition framework, where the events are interactions between pairs of objects.

Multiple kernel learning for image indexing

Abstract: The paper presents a novel framework for learning the hash functions for indexing through Multiple Kernel Learning. The Distance Based Hashing function is applied which does the object projection to hash space by preserving inter object distances. In recent works, the kernel matrix has been proved to be more accurate representation of similarity in various recognition problems. Our framework learns the optimal kernel for hashing by parametrized linear combination of base kernels. A novel application of Genetic Algorithm for the optimization of kernel combination parameters is presented. We also define new texture based feature representation for images. Our proposed framework can also be applied for optimal combination of multiple sources for indexing. The evaluation of the proposed framework is presented for CIFAR-10 dataset by applying individual and combination of different features. Additionally, the primary experimental results with MNIST dataset is also presented.

Texture mapping based video compression scheme

Abstract: We present a video coding technique based on texture synthesis via stitching small patches of previous images as done in texture transfer. Each macroblock is identified as a texture or non texture block by means of an edge based criterion. The texture blocks are reconstructed using the luminance values as the control block information to guide the texture transfer algorithm. By modifying the texture transfer algorithm we ensure both spatial as well as temporal consistency. The non texture blocks and the luminance values for texture blocks can be coded by any encoder and thus our method can act as a pre-processor to any video codec e.g. H.264 and add to the compression obtained.

RBF based spatio-temporal representation technique for video compression

Abstract: Parametric coding is a technique in which data is processed to extract meaningful information and then representing it compactly using appropriate parameters. Parametric Coding exploits redundancy in information to provide a very compact representation and thus achieves very high compression ratios. However, this is achieved at the cost of higher computation complexity. This disadvantage is now being offset by the availability of high speed processors, thus making it possible to exploit the high compression ratios of the parametric video coding techniques. In this paper a novel idea for efficient parametric representation of video is proposed. We perform Oct-Tree Decomposition on a video stack, followed by parameter extraction using Radial Basis Function Networks (RBFN) to achieve exceptionally high compression ratios, even higher than the state of art H.264 codec. The proposed technique exploits spatial-temporal redundancy and therefore inherently achieves multiframe prediction.

Method, apparatus and system for orienting a disoriented image

Abstract: A method, apparatus and system for orienting a disoriented image, and a method, apparatus and system for training a plurality of Gaussian mixture models (GMMs) to orient the disoriented image are provided. The method of training the plurality of GMMs includes: obtaining a plurality of color and texture features from the disoriented image; selecting a plurality of discriminative features from the color and texture features; calculating probabilities of each of the GMMs orienting the disoriented image, where each of the GMMs represents one of a plurality of rotation classes, and each of the rotation classes represents a rotation angle that is a multiple of a right angle. Furthermore, the system includes an electronic device that includes an embedded platform including a processor which processes the disoriented image.