Krishna Chandramouli

Image Classification using Chaotic Particle Swarm Optimization

Particle swarm optimization is one of several meta-heuristic algorithms inspired by biological systems. The chaotic modeling of particle swarm optimization is presented in this paper with application to image classification. The performance of this modified particle swarm optimization algorithm is compared with standard particle swarm optimization. Numerical results of this comparative study are performed on binary classes of images from the Corel dataset.

Combining image captions and visual analysis for image concept classification

We present a framework for efficiently exploiting free-text annotations as a complementary resource to image classification. A novel approach called Semantic Concept Mapping (SCM) is used to classify entities occurring in the text to a custom-defined set of concepts. SCM performs unsupervised classification by exploiting the relations between common entities codified in the Wordnet thesaurus. SCM exploits Targeted Hypernym Discovery (THD) to map unknown entities extracted from the text to concepts in Wordnet. We show how the result of SCM/THD can be fused with the outcome of Knowledge Assisted Image Analysis (KAA), a classification algorithm that extracts and labels multiple segments from an image. In the experimental evaluation, THD achieved an accuracy of 75%, and SCM an accuracy of 52%. In one of the first experiments with fusing the results of a free-text and image-content classifier, SCM/THD + KAA achieved a relative improvement of 49% and 31% over the text-only and image-content-only baselines.

Integrating Image Segmentation and Classification for Fuzzy Knowledge-Based Multimedia Indexing

In this paper we propose a methodology for semantic indexing of images, based on techniques of image segmentation, classification and fuzzy reasoning. The proposed knowledge-assisted analysis architecture integrates algorithms applied on three overlapping levels of semantic information: i) no semantics, i.e. segmentation based on low-level features such as color and shape, ii) mid-level semantics, such as concurrent image segmentation and object detection, region-based classification and, iii) rich semantics, i.e. fuzzy reasoning for extraction of implicit knowledge. In that way, we extract semantic description of raw multimedia content and use it for indexing and retrieval purposes, backed up by a fuzzy knowledge repository. We conducted several experiments to evaluate each technique, as well as the whole methodology in overall and, results show the potential of our approach.

CAE-L: An Ontology Modelling Cultural Behaviour in Adaptive Education

The presentation of learning materials in Adaptive Education Hypermedia is influenced by several factors such as learning style, background knowledge and cultural background, to name a few. In this paper, we introduce the notion of the CAE-L Ontology for modelling stereotype cultural artefacts in adaptive education. The Ontology design is based on the user study gathered from the respondents to the CAE questionnaire which determines the cultural artefacts that influence a learner¿s behaviour within an educational environment. We present a brief overview of the implementation and discuss the stereotype presentation styles from three different countries, namely China, Ireland and UK.

Semantic structuring and retrieval of event chapters in social photo collections

The phenomenal growth of multimedia content on the web over the last couple of decades has paved the way for content management systems integrating intelligent information retrieval and indexing techniques. Also, in order to improve the performance of retrieval techniques while searching and navigating the database, many relevance feedback algorithms are implemented, in which the subjective semantics of individual users are included in the image search. Following the recent developments in social networking, there is an emerging interest to share experiences online with friends using multimedia data. As the experiences to be shared among social peers vary from a simple social gathering to a tourism visit with a group of peers, there is a critical need for intelligent content management tools driven by a social perspective. Addressing the challenges related to socially-driven content management, the objective of this paper is twofold. First, we investigate techniques to intelligently structure multimedia content to enable efficient browsing of photo albums. The proposed structuring schemes exploit EXIF metadata, visual content and social peer relationships. Second, we propose a retrieval model based on social context to identify users with similar interests. The retrieval model aims to allow increased interaction among social peers. The proposed techniques have been evaluated against tourism pictures captured across Europe.

Particle Swarm Optimisation and Self Organising Maps Based Image Classifier

Neural network based image classification has been dominated by self organising maps. Following the recent developments in biologically inspired optimisation techniques, the application of particle swarm optimization algorithm for updating the weights of self organising maps has been studied in this paper, along with different network configurations of self organising maps. Experimental results are presented for 6 different classes from the Corel database based on MPEG -7 visual descriptor.

A Comparative Study of Classification Techniques for Knowledge-Assisted Image Analysis

In this paper, four individual approaches to region classification for knowledge-assisted semantic image analysis are presented and comparatively evaluated. All of the examined approaches realize knowledge-assisted analysis via implicit knowledge acquisition, i.e. are based on machine learning techniques such as support vector machines (SVMs), self organizing maps (SOMs), genetic algorithm (GA)and particle swarm optimization (PSO). Under all examined approaches, each image is initially segmented and suitable low-level descriptors are extracted for every resulting segment. Then, each of the aforementioned classifiers is applied to associate every region with a predefined high-level semantic concept. An appropriate evaluation framework has been employed for the comparative evaluation of the above algorithms under varying experimental conditions.

Image classification using biologically inspired systems

In this paper the problem of the image classification based on biologically inspired optimization systems is addressed. Recent developments in applied and heuristic optimization have been strongly influenced and inspired by natural and biological system. The findings of recent studies are showing strong evidence to the fact that some aspects of the collaborative behavior of social animals such as ants and birds can be applied to solve specific problems in science and engineering. Two algorithms based on this paradigm Ant Colony Optimization and Particle Swarm Optimization are investigated in this paper. The comparative evaluation of the recently developed techniques by the authors for optimizing the COP-K-means and the Self Organizing Feature Maps for the application of Binary Image Classification is presented. The precision and retrieval results are used as the metrics of comparison for both classifiers.

Towards semantic tagging in collaborative environments

Tags pose an efficient and effective way of organization of resources, but they are not always available. A technique called SCM/THD investigated in this paper extracts entities from free-text annotations, and using the Lin similarity measure over the WordNet thesaurus classifies them into a controlled vocabulary of tags. Hypernyms extracted from Wikipedia are used to map uncommon entities to Wordnet synsets. In collaborative environments, users can assign multiple annotations to the same object hence increasing the amount of information available. Assuming that the semantics of the annotations overlap, this redundancy can be exploited to generate higher quality tags. A preliminary experiment presented in the paper evaluates the consistency and quality of tags generated from multiple annotations of the same image. The results obtained on an experimental dataset comprising of 62 annotations from four annotators show that the accuracy of a simple majority vote surpasses the average accuracy obtained through assessing the annotations individually by 18%. A moderate-strength correlation has been found between the quality of generated tags and the consistency of annotations.

Geo-tagging online videos using semantic expansion and visual analysis

The association of geographical tags to multimedia resources enables browsing and searching online multimedia repositories using geographical criteria, but millions of already online but non geo-tagged videos and images remain invisible to the eyes of this type of systems. This situation calls for the development of automatic geo-tagging techniques capable of estimating the location where a video or image was taken. This paper presents a bimodal geo-tagging system for online videos based on extracting and expanding the geographical information contained in the textual metadata and on visual similarity criteria. The performance of the proposed system is evaluated on the MediaEval 2011 Placing task data set, and compared against the participants in that workshop.