Mukesh C. Motwani

3D face pose discrimination using wavelets

This paper describes a robust method for discriminating 3D face pose (face orientation) from a video sequence featuring views of a human head under variable lighting and facial expression conditions. The wavelet transform is used to decompose the image into multiresolution face images containing both spatial and spatial-frequency information. Principal component analysis (PCA) is used to project a low-resolution sub-band face pose onto a pose eigen-space where the first 3 eigen coefficients are found most sensitive to pose and follow a trajectory as the pose changes. Any unknown pose of an query image can then be estimated by finding the Euclidean distance of the first 3 eigen coefficients of the query image from the estimated trajectory. The wavelet transform reduces the computational load on the PCA and makes the algorithm robust against illumination changes and facial expression. An accuracy of 84% was obtained for test images unseen during training under different environment conditions, facial expressions, and even different human subjects.

Adaptive Fuzzy Watermarking for 3D Models

Watermarking algorithms have a basic requirement that the watermark amplitude should be as high as possible for robustness and at the same time the watermark should not introduce any perceptible artifacts. Thus, the design of watermarking algorithms involves a tradeoff between imperceptibility and robustness. This paper proposes a novel state of the art algorithm, which is based on wavelet and fuzzy logic, to determine an optimal value for the watermark amplitude to be inserted in a 3D model. The system being adaptive to the local geometry of the mesh inserts an 8 bit grey scale image as watermark as compared to inserting a binary image in existing algorithms. Simulation results prove it to be robust against smoothing, cropping, affine operations and noise attacks.

Towards a scalable and interoperable global environmental sensor network using Service Oriented Architecture

Sensor networks are critical infrastructures for monitoring environmental variables, allowing evaluation of long-term trends and changes in the interaction of atmospheric, ecologic, and hydrologic processes. However, due to lack of coordination between large-scale environmental observation systems that have been set up around the globe, the wealth of information collected by sensor networks is not exploited to its full potential. Since there is no central organization tying these various systems together, data acquisition and dissemination methods are inconsistent and public accessibility to these observation systems is restricted to the methods chosen at the individual project level. The ability to easily discover, access, and interact with observational instruments and use real-time sensory data spread across different environmental observatories is limited. Interoperability of sensor network assets is essential to producing improved projections, models, analyses, and assessments at a global scale. To this end this paper presents the architectural design for a scalable, interoperable and real-time environmental sensor network. The proposed architectural design also provides support for dynamic reconfiguration of sensors in response to changing environmental conditions, and facilitates secure sensor access and delivery of observations to a distributed community of users using standardized mechanisms based on Open Geospatial Consortium standards. The outlined architecture uses an Event-Driven Service Oriented Architecture, with Enterprise Service Bus as its backbone messaging transport layer.

Watermark Embedder Optimization for 3D Mesh Objects Using Classification Based Approach

This paper presents a novel 3D mesh watermarking scheme that utilizes a support vector machine(SVM) based classifier for watermark insertion. Artificial intelligence(AI)based approaches have been employed by watermarking algorithms for various host mediums such as images, audio, and video. However, AI based techniques are yet to be explored by researchers in the 3D domain for watermark insertion and extraction processes. Contributing towards this end, the proposed approach employs a binary SVM to classify vertices as appropriate or inappropriate candidates for watermark insertion. The SVM is trained with feature vectors derived from the curvature estimates of a 1-ring neighborhood of vertices taken from normalized 3D meshes. A geometry-based non-blind approach is used by the watermarking algorithm. The robustness of proposed technique is evaluated experimentally by simulating attacks such as mesh smoothing, cropping and noise addition.

Wavelet based fuzzy perceptual mask for images

One of the characteristics of the Human Visual System (HVS) is to model the sensitivity of the human eye at each coordinate location in the image. This paper explores the use of fuzzy logic for building a non-linear HVS model for perceptual masking in wavelet domain. The fuzzy input variables corresponding to brightness, edge sensitivity, and texture are computed for each wavelet coefficient at different scales in an image. The output of the fuzzy system is a single value which gives a perceptual value for each corresponding wavelet coefficient. This paper proposes a novel method of constructing perceptual mask in wavelet domain using fuzzy logic.

Robust watermarking of 3D skinning mesh animations

This paper presents a novel robust watermarking algorithm for 3D skinning mesh animations by embedding the watermark in skin weights in addition to key frames. This method can be used for copyright protection, tamper proofing or content annotation purposes. The proposed watermark is immune to noise attacks on key frames and skin weights, key frame dropping and frame modification and is perceptible invisible as well. Experimental results verified that the proposed algorithm has good robustness against attacks and maintains invisibility by preprocessing the animation data sets by key frame decimation.

Towards Benchmarking of Video Motion Tracking Algorithms

The environment in which video motion needs to be tracked, places several constraints on the design of the tracking system. Current datasets which are used to evaluate and compare video motion tracking algorithms use a cumulative performance measure without thoroughly analyzing the effect of these different constraints imposed by the environment. There is need to build a heuristic framework which analyses these constraints as parameters of the framework and their effect on selection or design of tracking algorithm. The emphasis in this paper is to identify these parameters which will lay a foundation for defining subjective measures for the comparison of performance evaluation of tracking algorithms.

Tamper Proofing 3D Models

This paper describes a novel algorithm designed for tamper proofing of 3D models. Fragile watermarking is a known technique for detecting location of tamper in the artwork. However, to detect even minute changes, the watermark needs to be distributed throughout the 3D model. This poses as a challenge since watermarking all vertices can cause perceptible distortion. The proposed algorithm solves this problem by inserting a watermark in all the vertices of a 3D model. The watermark is randomly added to each and every vertex of the 3D model by modifying the coordinate location of the vertices. Genetic Algorithms have been used to find the near optimal coordinate location of the watermarked vertex. The fitness function chosen is the Signal to Noise Ratio of the 1-ring neighbourhood of the vertex. This ensures that there exists no distortion in the watermarked model. The proposed approach is computationally inexpensive and experimental results indicate that the algorithm can efficiently detect location of any kind of unauthorized data modification.

An Intelligent Learning Approach for Information Hiding in 3D Multimedia

This paper presents a new watermarking algorithm for 3D triangular mesh models that is based on surface curvature estimation and supervised learning. A feedforward backpropagation neural network is adopted for selecting vertices for watermark insertion. A variety of 3D models with varying degrees of surface curvature are used to train and simulate the neural network. An array of neural networks is used for vertices with different valences to achieve higher watermark embedding capacity. A gray scale bitmap image is used as the watermark. The watermark extraction process is informed and needs the original watermark and 3D model. Experimental results evaluate the embedding capacity, imperceptibility and robustness of the proposed algorithm and simulate various attacks including noise addition, smoothing and cropping.

Using Radial Basis Function Networks for Watermark Determination in 3D Models

In this paper, we investigate the ability of a radial basis function network to determine the values of the watermark to be inserted in a 3D triangulated mesh model. The challenge in a watermarking algorithm is to achieve high watermark embedding capacity without causing perceptual distortion to the model. The proposed technique overcomes this challenge. The principal, mean and Gaussian curvature values computed at each vertex of the 3D model collectively represent the local geometry of the vertex and its neighborhood. These curvature values are used as input feature vectors to train the neural network. The amount of watermark to be inserted at a vertex is non-linearly proportional to these feature vectors. The neural network is trained with watermark values to learn this non-linear relationship. A wide range of 3D models are used to train the neural network such that the large variations in the geometry of the vertices is incorporated by the training phase. The algorithm adopts a non-blind extraction process to retrieve the watermark. Experimental results prove that the watermarking algorithm achieves imperceptibility, high capacity and robustness to noise and cropping attacks up to 20% level. I. INTRODUCTION