Ashok Samal

Automatic recognition and analysis of human faces and facial expressions: a survey

Abstract Humans detect and identify faces in a scene with little or no effort. However, building an automated system that accomplishes this task is very difficult. There are several related subproblems: detection of a pattern as a face, identification of the face, analysis of facial expressions, and classification based on physical features of the face. A system that performs these operations will find many applications, e.g. criminal identification, authentication in secure systems, etc. Most of the work to date has been in identification. This paper surveys the past work in solving these problems. The capability of the human visual system with respect to these problems is also discussed. It is meant to serve as a guide for an automated system. Some new approaches to these problems are also briefly discussed.

A feature-based approach to conflation of geospatial sources

A Geographic Information System (GIS) populated with disparate data sources has multiple and different representations of the same real-world object. Often, the type of information in these sources is different, and combining them to generate one composite representation has many benefits. The first step in this conflation process is to identify the features in different sources that represent the same real-world entity. The matching process is not simple, since the identified features from different sources do not always match in their location, extent, and description. We present a new approach to matching GIS features from disparate sources. A graph theoretic approach is used to model the geographic context and to determine the matching features from multiple sources. Experiments on implementation of this approach demonstrate its viability.

A dual encryption protocol for scalable secure multicasting

We propose a dual encryption protocol for scalable secure multicasting. Multicasting is a scalable solution for group communication. It however poses several unique security problems. We use hierarchical subgrouping to achieve scalability. Third-party hosts or members of the multicast group are designated as subgroup managers. They are responsible for secret key distribution and group membership management at the subgroup level. Unlike existing secure multicast protocols, our protocol need not trust the subgroup managers with the distribution of data encryption keys. The dual encryption protocol proposed in this paper distributes encrypted data encryption keys via subgroup managers. We also present a classification of the existing secure multicast protocols, compare their relative merits and show the advantages of our protocol.

Computation of a face attractiveness index based on neoclassical canons, symmetry, and golden ratios

Analysis of attractiveness of faces has long been a topic of research. Literature has identified many different factors that can be related to attractiveness. In this research we analyze the role of symmetry, neoclassical canons, and golden ratio in the determination of attractiveness of a face. We focus on the geometry of a face and use actual faces for our analysis. We find there are some differences in the criteria used by males and females to determine attractiveness. The model we have developed to predict the attractiveness of a face using its geometry is accurate with low residual errors.

Semi-Automated Road Detection From High Resolution Satellite Images by Directional Morphological Enhancement and Segmentation Techniques

Extraction of map objects such roads, rivers and buildings from high resolution satellite imagery is an important task in many civilian and military applications. We present a semi-automatic approach for road detection that achieves high accuracy and efficiency. This method exploits the properties of road segments to develop customized operators to accurately derive the road segments. The customized operators include directional morphological enhancement, directional segmentation and thinning. We have systematically evaluated the algorithm on a variety of images from IKONOS, QuickBird, CARTOSAT-2A satellites and carefully compared it with the techniques presented in literature. The results demonstrate that the algorithm proposed is both accurate and efficient.

A fuzzy clustering approach to delineate agroecozones

Agroecozones are geographic areas that share similar biophysical characteristics for crop production, such as soil, landscape, and climate, which define the potentials for agricultural productivity. Delineation and characterization of agroecozones would greatly enhance agricultural decision-making and management, as well as the extrapolation of experiment station research and field trials to forms and landscapes of similar agronomic behavior. Currently, agroecozones are often represented with static and rigid boundaries derived by using data obtained by averaging observations over a period of time. The boundaries of these regions, however, are fuzzy and reflect change over time and space. Furthermore, the measurements used as the basis of the delineation are themselves uncertain. Agroecozones may be obtained by treating the input data as well as the resultant regions as fuzzy. Clustering is one of the most common approaches to derive agroecozones delineation. In this paper, we explore the suitability of some fuzzy clustering approaches for this problem. Experimental results show that fuzzy algorithms generate more accurate delineations as measured by their closeness to the Major Land Resources Areas (MLRA) map. However, they both require greater computational resources. Some additional advantages of using a fuzzy approach are also illustrated. This approach should be viewed as an additional tool available for modeling and analysis of important processes in spatial environmental decision support systems.

Generalized Hough transform for natural shapes

Abstract The Hough transform and their extensions can not adequately handle shapes characterized by the fact that different instances of the same shape are similar , but not identical . We present an extension to recognize natural shapes. It is based on the principle that natural shapes can be represented by two extremes: the inner extreme and the outer extreme.

Analysis of sexual dimorphism in human face

Human beings can easily distinguish between a male and a female face without much difficulty. The science of recognizing and differentiating different faces by humans is not completely understood and is still under research. Sexual dimorphism is common in humans and indeed in other species of animals as well. Significant differences between males and females exist in many aspects like size, color, body shapes, and weight. In this research, we characterize and analyze the sexual dimorphism in human face as a function of age and of face features. Features are grouped into six categories: head, eyes, orbits, nose, lips, and mouth, and ears. We demonstrate that the face of adult males is significantly different from adult females. We also identify the features that significantly contribute to the dimorphism of the face. This provides a basis for gender-based classification of faces.

Texture as the basis for individual tree identification

Recognizing plants from imagery is a complex task due to their irregular nature. In this research, three tree species, Japanese yew (Taxus cuspidata Sieb. & Zucc.), Hicks yew (Taxus x media), and eastern white pine (Pinus strobus L.), were identified using their textural properties. First, the plants were separated from their backgrounds in digital images based on a combination of textural features. Textural feature values for energy, local homogeneity, and inertia were derived from the co-occurrence matrix and differed significantly between the trees and their backgrounds. Subsequently, these features were used to construct the feature space where the nearest-neighbor method was applied to discriminate trees from their backgrounds. The recognition rates for Japanese yew, Hicks yew, and eastern white pine were 87%, 93%, and 93%, respectively. The study demonstrates that the texture features selected and the methods employed satisfactorily separated the trees from their relatively complex backgrounds and effectively differentiated between the three species. This research can lead to potentially useful applications in forestry and related disciplines.

Sentence recognition from articulatory movements for silent speech interfaces

Recent research has demonstrated the potential of using an articulation-based silent speech interface for command-and-control systems. Such an interface converts articulation to words that can then drive a text-to-speech synthesizer. In this paper, we have proposed a novel near-time algorithm to recognize whole-sentences from continuous tongue and lip movements. Our goal is to assist persons who are aphonic or have a severe motor speech impairment to produce functional speech using their tongue and lips. Our algorithm was tested using a functional sentence data set collected from ten speakers (3012 utterances). The average accuracy was 94.89% with an average latency of 3.11 seconds for each sentence prediction. The results indicate the effectiveness of our approach and its potential for building a real-time articulation-based silent speech interface for clinical applications.