Department of Computer Science & Engineering

&ldquo;IIT Jodhpur is committed to impart quality education to facilitate the creation of qualified, competent and responsible human resources to meet the emerging technological challenges of the world. It is working towards creating an equitable learning platform that is positive, accessible and effective in the face of the modern societal and scientific advancements.

Flexibility in curriculum, in all its academic programs, encompassing elements of teaching, research, creativity, innovation and entrepreneurship is the most unique feature of IIT Jodhpur. The Department of Science and Technology (DST), under the National Mission on Cyber Physical Systems, has identified IIT Jodhpur as a Technology &amp; Innovation Hub in the area of Augmented Reality and Virtual Reality. The institute has also been nominated as the coordinator for the Jodhpur City and Knowledge Cluster by the Office of Principal Scientific Adviser, Government of India.

The sprawling 852-acre IIT Jodhpur campus is a majestic tribute to the deep-rooted diverse cultural heritage of the region. The campus is an architectural masterpiece, which is slated to become an international exemplar of sustainability with net-zero energy, and advanced water and waste management strategies.&rdquo;

IIT Jodhpur

The quality of input data has an important role in the performance of a biometric system. Images such as fingerprint and face captured under non-ideal conditions may require additional preprocessing. This chapter presents intelligent SVM techniques for quality assessment and enhancement. The proposed quality assessment algorithm associates the quantitative quality score of the image that has a specific type of irregularity such as noise, blur, and illumination. This enables the application of the most appropriate quality enhancement algorithm on the non-ideal image. We further propose a SVM quality enhancement algorithm which simultaneously applies selected enhancement algorithms to the original image and selects the best quality regions from the global enhanced image. These selected regions are used to generate single high quality image. The performance of the proposed algorithms is validated by considering face biometrics as the case study. Results show that the proposed algorithms improve the verification accuracy of face recognition by around 10-17%. © 2008 Springer-Verlag Berlin Heidelberg.

World University root node

Studies in Computational Intelligence

SVM based adaptive biometric image enhancement using quality assessment

This paper presents a two level hierarchical fusion of face images captured under visible and infrared light spectrum to improve the performance of face recognition. At image level fusion, two face images from different spectrums are fused using DWT based fusion algorithm. At feature level fusion, the amplitude and phase features are extracted from the fused image using 2D log polar Gabor wavelet. An adaptive SVM learning algorithm intelligently selects either the amplitude or phase features to generate a fused feature set for improved face recognition. The recognition performance is observed under the worst case scenario of using single training images. Experimental results on Equinox face database show that the combination of visible light and short-wave IR spectrum face images yielded the best recognition performance with an equal error rate of 2.86%. The proposed image-feature fusion algorithm also performed better than existing fusion algorithms. © 2006 Elsevier B.V. All rights reserved.

Information Fusion

Hierarchical fusion of multi-spectral face images for improved recognition performance

This paper presents a novel formulation of multiclass support vector machine by integrating the concepts of soft labels and granular computing. The proposed multiclass mv-granular soft support vector machine uses soft labels to address the issues due to noisy and incorrectly labeled data, and granular computing to make it adaptable to data distributions both globally and locally. The proposed multiclass classifier is used for dynamic selection in a multispectral face recognition application. Specifically, for the given probe face images, mv-GSSVM is used to optimally choose one of the four options: visible spectrum face recognition, short-wave infrared face recognition, multispectral face image fusion, and multispectral match score fusion. Experimental results on a multispectral face database show that the proposed algorithm improves the verification accuracy and also decreases the computational time. © 2008 IEEE.

Proceedings - International Conference on Pattern Recognition

Multiclass mv-granular soft support vector machine: A case study in dynamic classifier selection for multispectral face recognition

In this work, we provide new and improved attacks against 22, 23 and 24-step SHA-2 family using a local collision given by Sanadhya and Sarkar (SS) at ACISP '08. The success probability of our 22-step attack is 1 for both SHA-256 and SHA-512. The computational efforts for the 23-step and 24-step SHA-256 attacks are respectively 211.5 and 228.5 calls to the corresponding step reduced SHA-256. The corresponding values for the 23 and 24-step SHA-512 attack are respectively 216.5 and 232.5 calls. Using a look-up table having 232 (resp. 264) entries the computational effort for finding 24-step SHA-256 (resp. SHA-512) collisions can be reduced to 215.5 (resp. 222.5) calls. We exhibit colliding message pairs for 22, 23 and 24-step SHA-256 and SHA-512. This is the first time that a colliding message pair for 24-step SHA-512 is provided. The previous work on 23 and 24-step SHA-2 attacks is due to Indesteege et al. and utilizes the local collision presented by Nikolić and Biryukov (NB) at FSE '08. The reported computational efforts are 218 and 228.5 for 23 and 24-step SHA-256 respectively and 2 43.9 and 253 for 23 and 24-step SHA-512. The previous 23 and 24-step attacks first constructed a pseudo-collision and later converted it into a collision for the reduced round SHA-2 family. We show that this two step procedure is unnecessary. Although these attacks improve upon the existing reduced round SHA-2 attacks, they do not threaten the security of the full SHA-2 family. © 2008 Springer Berlin Heidelberg.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

New collision attacks against up to 24-step SHA-2

Although a fuzzy rule based system offers interpretability, its application in gene expression data analysis becomes difficult due to the very high dimensional nature of the data. Here we propose an interesting scheme of combining fuzzy modeling with neural networks for designing fuzzy rule based classifiers for gene expression data analysis. A neural system is used for selecting a set of informative genes. Considering only these selected set of genes, we cluster the expression data with a fuzzy clustering algorithm. Each cluster is then converted into a fuzzy if-then rule, which models an area in the input space. These rules are tuned using a gradient descent technique to improve the classification performance. The rule base is tested on a leukemia data set containing two classes and it is found to produce good results. We propose some simple criteria to simplify membership functions and the rules. Our rule extraction scheme can be automated. Unlike other classifiers, it produces human interpretable rules which are not expected to give poor generalization because fuzzy rules do not respond to areas not represented by the training data. The last two properties are very important for problems like diagnosis of cancer. © 2008 - IOS Press.

Journal of Intelligent and Fuzzy Systems

Deriving meaningful rules from gene expression data for classification

Recognition of object categories from their images is extremely challenging due to the large intra-class variations, and variations in pose, illumination and scale, in addition to lack of depth information of the object. Recovering the depth information from multiple images or from image cues such as variations in illumination or focus, is both computationally intensive and error prone. In contrast, the appearance based approaches are more robust and computationally efficient. However, they lack the potential accuracy of 3D feature based approaches due to the lack of shape information. We propose the use of structured lighting patterns projected on the object, which gets deformed according to the shape of the object for recognition. Since our goal is object classification and not shape recovery, we characterize the deformations using simple texture measures, thus avoiding depth recovery step. Moreover, the shape information present in the deformations is implicitly used for classification. We show that the information thus derived can significantly improve the accuracy of object category recognition from arbitrary-pose images. © 2008 IEEE.

Proceedings - 6th Indian Conference on Computer Vision, Graphics and Image Processing, ICVGIP 2008

Object category recognition with projected texture

Algorithms for classification of 3D objects either recover the depth information lost during imaging using multiple images, structured lighting, image cues, etc. or work directly the images for classification. While the latter class of algorithms are more efficient and robust in comparison, they are less accurate due to the lack of depth information. We propose the use of structured lighting patterns projected on the object, which gets deformed according to the shape of the object. Since our goal is object classification and not shape recovery, we characterize the deformations using simple texture measures, thus avoiding the error prone and computationally expensive step of depth recovery. Moreover, since the deformations encode depth variations of the object, the 3D shape information is implicitly used for classification. We show that the information thus derived can significantly improve the accuracy of object classification algorithms, and derive the theoretical limits on height variations that can be captured by a particular projector-camera setup. A 3D texture classification algorithm derived from the proposed approach achieves a ten-fold reduction in error rate on a dataset of 30 classes, when compared to state-of-the-art image based approaches. We also demonstrate the effectiveness of the approach for a hand geometry based authentication system, which achieves a four-fold reduction in the equal error rate on a dataset containing 149 users. © 2008 Springer Berlin Heidelberg.

Projected texture for object classification

We have attempted the problem of novel view synthesis of scenes containing man-made objects from images taken by arbitrary, uncalibrated cameras. Under the assumption of availability of the correspondence of three vanishing points, in general position, we propose two techniques. The first is a transfer-based scheme which synthesizes new views with only a translation of the virtual camera and computes z-buffer values for handling occlusions in synthesized views. The second is a reconstruction-based scheme which synthesizes arbitrary new views in which the camera can undergo rotation as well as translation. We present experimental results to establish the validity of both formulations. © 2006.

Pattern Recognition Letters

On the view synthesis of man-made scenes using uncalibrated cameras

This paper presents a novel multi-level wavelet based fusion algorithm that combines information from fingerprint, face, iris, and signature images of an individual into a single composite image. The proposed approach reduces the memory size, increases the recognition accuracy using multi-modal biometric features, and withstands common attacks such as smoothing, cropping, JPEG 2000, and filtering due to tampering. The fusion algorithm is validated using the verification algorithms we developed, existing algorithms, and commercial algorithm. In addition to our multi-modal database, experiments are also performed on other well known databases such as FERET face database and CASIA iris database. The effectiveness of the fusion algorithm is experimentally validated by computing the matching scores and the equal error rates before fusion, after reconstruction of biometric images, and when the composite fused image is subjected to both frequency and geometric attacks. The results show that the fusion process reduced the memory required for storing the multi-modal images by 75%. The integrity of biometric features and the recognition performance of the resulting composite fused image is not affected significantly. The complexity of the fusion and the reconstruction algorithms is O(n log n) and is suitable for many real-time applications. We also propose a multi-modal biometric algorithm that further reduces the equal error rate compared to individual biometric images. © 2005 Elsevier B.V. All rights reserved.

Robust memory-efficient data level information fusion of multi-modal biometric images

Hydrophobicity is probably the most important property of amino acid that is being often exploited o predict protein folds. There are many amino acid hydrophobicity scales available in the literature. Here we propose two computational approaches based on self-organizing map (SOM) and fuzzy clustering to find some consensus scales. Although SOM and fuzzy clustering produce centroids, we propose new schemes to compute more effective representative scales that exploit the properties of SOM and fuzzy memberships. To demonstrate the utility of the new scales, we apply them to predict the protein folds of a benchmark data set using neural networks. Our experiments show that it is possible to generate useful scales with better utility compared to some existing scales. There can be other applications of the proposed scales. © Dynamic Publishers, Inc.