Yiannis Xirouhakis

Affine-invariant curve normalization for object shape representation, classification, and retrieval

Abstract. A novel method for two-dimensional curve normalization with respect to affine transformations is presented in this paper, which allows an affine-invariant curve representation to be obtained without any actual loss of information on the original curve. It can be applied as a preprocessing step to any shape representation, classification, recognition, or retrieval technique, since it effectively decouples the problem of affine-invariant description from feature extraction and pattern matching. Curves estimated from object contours are first modeled by cubic B-splines and then normalized in several steps in order to eliminate translation, scaling, skew, starting point, rotation, and reflection transformations, based on a combination of curve features including moments and Fourier descriptors.

Least squares estimation of 3D shape and motion of rigid objects from their orthographic projections

The extraction of motion and shape information of three-dimensional objects from their two-dimensional projections is a task that emerges in various applications such as computer vision, biomedical engineering, and video coding and mining especially after the recent guidelines of the Motion Pictures Expert Group regarding MPEG-4 and MPEG-7 standards. Present work establishes a novel approach for extracting the motion and shape parameters of a rigid three-dimensional object on the basis of its orthographic projections and the associated motion field. Experimental results have been included to verify the theoretical analysis.

Affine-invariant curve normalization for shape-based retrieval

A novel method for two-dimensional curve normalization with respect to affine transformations is presented in this paper, allowing an affine-invariant curve representation to be obtained without any actual loss of information on the original curve. It can be applied as a pre-processing step to any shape representation, classification, recognition or retrieval technique, since it effectively decouples the problem of affine-invariant description from feature extraction and pattern matching. Curves estimated from object contours are first modeled by cubic B-splines and then normalized in several steps in order to eliminate translation, scaling, skew, starting point, rotation and reflection transformations, based on a combination of curve features including moments and Fourier descriptors.

Efficient optical camera tracking in virtual sets

Optical tracking systems have become particularly popular in virtual studios applications tending to substitute electromechanical ones. However, optical systems are reported to be inferior in terms of accuracy in camera motion estimation. Moreover, marker-based approaches often cause problems in image/video compositing and impose undesirable constraints on camera movement, present work introduces a novel methodology for the construction of a two-tone blue screen, which allows the localization of camera in three-dimensional (3-D) space on the basis of the captured sequence. At the same time, a novel algorithm is presented for the extraction of cameras 3-D motion parameters based on 3-D-to-two-dimensional (2-D) line correspondences. Simulated experiments have been included to illustrate the performance of the proposed system.

Main Mobile Object Detection and Localization in Video Sequences

Main mobile object detection and localization is a task of major importance in the fields of video understanding, object-based coding and numerous related applications, such as content-based retrieval, remote surveillance and object recognition.The present work revisits the algorithm proposed in [13] for mobile object localization in both indoor and outdoor sequences when either a static or a mobile camera is utilized. The proposed approach greatly improves the trade-off between accuracy and time-performance leading to satisfactory results with a considerably low amount of computations. Moreover, based on the point gatherings extracted in [13], the bounding polygon and the direction of movement are estimated for each mobile object; thus yielding an adequate representation in the MPEG-7 sense.Experimental results over a number of distinct natural sequences have been included to illustrate the performance of the proposed approach.

User choices for efficient 3D motion and shape extraction from orthographic projections

The extraction of structure-from-motion emerges in several research fields such as computer vision, video coding, biomedical engineering and human-computer interaction. The present work focuses on the algorithmic approach of structure-from-motion extraction under orthography providing, at the same time, guidelines in matters of implementation. Relative principles, constraints and stability are discussed. The improvement of the algorithms performance w.r.t. the proposed user-choices is illustrated by means of experimental results.

Efficient browsing in multimedia databases using intelligent agents and content-based retrieval schemes

An intelligent multimedia system is developed for efficient retrieval and processing of information stored in multimedia databases. Intelligent agents are employed in order to limit the search into minor subsets of the database; a training strategy is introduced for quick browsing in the database contents. Content-based query and retrieval schemes are also employed to increase the effectiveness of the browsing procedure.

Optical camera tracking in virtual studios: Degenerate cases

Over the past few years, virtual studios applications have significantly attracted the attention of the entertainment industry. Optical tracking systems for virtual sets production have become particularly popular tending to substitute electro-mechanical ones. In this work, an existing optical tracking system is revisited, in order to tackle with inherent degenerate cases; namely, reduction of the perspective projection model to the orthographic one and blurring of the blue screen. In this context, we propose a simple algorithm for 3D motion estimation under orthography using 3D-to-2D line correspondences. In addition, the watershed algorithm is employed for successful feature extraction in the presence of defocus or motion blur.

Estimation of 3D Motion and Structure of Human Faces

The extraction of motion and shape information of three dimensional objects from video sequences emerges in various applications especially within the framework of the MPEG-4 and MPEG-7 standards. Particular attention has been given to this problem within the scope of model-based coding and knowledge-based ZD modeling. In this chapter, a novel algorithm is proposed for the 3D reconstruction of a human face from 2D projections. The obtained results can contribute to several fields with an emphasis on 3D modeling and characterization of human faces.

Efficient use of content-based retrieval in quick browsing: a realization

Due to recent growth of interest in multimedia applications, an increasing demand has emerged for efficient storage, management and browsing in multimedia databases. This fact has made the content-based retrieval (CBR) concept very popular during the past decade. However, it has been practically shown that CBR tools respond successfully to the inexperienced user queries very rarely. It is for this reason that in commercial systems, queries are supported by the use of annotations or manual indexing. In this work, content-based query, retrieval and indexing capabilities have been combined with an intelligent agent framework over a simple database architecture. The proposed system is based on the ideas presented in (Xirouhakis et al., 1998) which are implemented and further extended in this work.