Eckart Hundt

Method for processing ultrasonic echo signals of both directionally reflecting as well as nondirectionally scattering objects, particularly for ultrasonic image processing in the field of substance or tissue investigation

In an exemplary embodiment, the processing techniques of computerized x-ray tomography are applied to improve resolution transverse to the depth direction, for example with the use of compound ultrasonic scanning. In this way, an improved preliminary image for representing non-directionally scattering image points is attained. The ultrasonic echo signals are also processed to provide a preliminary image based predominantly or exclusively on directionally reflecting image points. The respective preliminary images for each image point or point region are then suitably combined to provide a stored ultrasonic image capable of display of the scanned region of improved accuracy, resolution, and freedom from noise.

A Face Verification Algorithm Integrating Geometrical and Template Features

One of the key problems in a face recognition system is verification. This paper presented a verification algorithm using SVM classifier with integrated geometrical features and template features based on our multi-view face recognition system. By using only six feature points located during the process of template matching, improved system performance is achieved.

Combining Two Imaging Modalities for Neuroradiological Diagnosis: 3D Representation of Cerebral Blood Vessels

Today the integration of information from different imaging modalities in medicine such as Computer Tomography or Magnetic Resonance Imaging (MRI) is left to the physician and gets little support from computers. In the case of neuroradiological diagnosis, information about cerebral blood vessels is available from 3D volume data from Magnetic Resonance Angiography (MRA) and from 2D images generated by Digital Subtraction Angiography (DSA). The DSA images have a higher resolution than MRA data, and therefore neuroradiologists are highly interested in a 3D reconstruction of cerebral blood vessels from different DSA projections. On the other hand, MRA contains important functional information, the velocity of blood flow. This paper describes work in progress to make available to the physician the full 3D information from both imaging modalities including an approach to 3D reconstruction from DSA im ages which makes use of the MRA data. The 3D DSA reconstruction also opens the way to an integration of information from DSA with completely different types of information, for example information on anatomical structure or soft tissue from MRI. An integral part of this work is a pilot system for clinical validation.

Topological approach to mathematical morphology

For low—bit—rate coding of video sequences, specifically in the context of the MPEG—4 proposal, morphological approaches proved to be highly attractive (see e.g. [16]). During a series of coding experiments performed at Siemens Research Laboratory in Munchen [9] the authors felt that there are some deficiencies of theory which need investigation. The aim of this paper is to sketch a theory which allows to understand the relationships between three classes of discrete concepts, namely discrete topology, discrete morphology and discrete metrics. Mathematical morphology is based on topologies for systems of subsets of a set [13]. The topology of the underlying set enters only indirectly. Therefore such concepts as connectedness of sets can cause difficulties if treated purely morphologically. These conceptual difficulties became especially apparent, when structures and algorithms were used practically which simultaneously involve both subset topologies and connectedness of subsets as is the case e.g. in watershed segmentation [16].

Wege zur Dokumentinterpretation: Schriftzeichenerkennung, Grafikerkennung, wissensbasierte Analyse

Die Dokumentinterpretation hat zum Ziel, den Informationsgehalt von Dokumenten zu erfassen und in elektronische Daten umzusetzen. Durch den zunehmenden Einsatz elektronischer Verarbeitungssysteme haben diese Verfahren ganz besonders an Bedeutung gewonnen. Entsprechend der Vielfalt der zu verarbeitenden Dokumente und den verschiedenen Anwendungen der Dokumentinterpretation werden unterschiedliche Verfahren der Mustererkennung eingesetzt. Es werden drei Wege der Dokumentinterpretation diskutiert, die auf spezifischen Verfahren beruhen und den Bedeutungsinhalt bis zu unterschiedlicher Tiefe erschliesen: Die Erkennung isolierter Schriftzeichen, die einen technisch hochwertigen Stand erreicht hat; die Erkennung grafischer Symbole und deren Beziehungen in Planen und Skizzen, die erst in prototypischen Systemen eingesetzt wird und die wissensbasierte Analyse, die mit Modellen aus dem Layout den logischen Inhalt erschliest und noch Gegenstand der Forschung ist.

Randorientierte Methoden zur Verdünnung - Ein Zugang zur Fuzzy Topologie

Es wird eine neue Charakterisierung der einfachen Punkte einer digitalen Menge vorgestellt. Einfache Punkte sind solche, deren Elimination die topologischen Eigenschaften der digitalen Menge nicht andert. Die hier gegebene Charakterisierung stutzt sich auf das Konzept der perfekten Randpunkte. Damit ergibt sich eine neue Moglichkeit, topologische Methoden, wie etwa die Verdunnung von Binarbildern, ohne direkten Ruckgriff auf topologische Begriffe zu begrunden. Eine Anwendungsmoglichkeit ist die unscharfe (Fuzzy) Ubertragung der Verdunnung auf Grauwertbilder.

Automatic Face Modeling and Synthesis Based on Image Pairs

Unlike traditional 3D model based or image based animation methods, in this paper a novel approach is presented to generate both facial actions and head rotations for photo-realistic facial animation based on one frontal and one half-profile facial image taken with an uncalibrated camera. We represent faces with 2D wire-frame models and use MPEG4 FAPs to encode basic facial actions. Hierarchical Direct Appearance Model is employed for facial feature localization. 3D deformable model is applied for pose estimation. By affine projection 3D deformable model and facial actions are mapped to 2D facial models and actions at various head poses. Coarse 2D models are refined with extracted facial features by RBF interpolation. Pose-variable facial animation is generated by synthesizing facial actions on 2D models and morphing facial textures between frontal and halfprofile views. Experimental results demonstrate the effectiveness of our approach.

A versatile DCT coding system for HDTV with interlaced and progressive scanning

Abstract HDTV implies various new technologies with strong implications for information and imaging technology, and telecommunications as well. Various standards, image formats and scanning modes are under discussion for various applications. Until now of HDTV signals via satellite and B-ISDN a video data rate of 125 Mbit/s was considered to be necessary for distribution. But the fast development of efficient coding methods indicates that also for HDTV the trend is moving toward lower bit-rates. That could open new areas of application in the range of 50 Mbit/s. We present a DCT coding concept that is versatile with respect to scanning mode - interlaced and progressive - and data rate. It shows very good image quality also at the low data rate of 50 Mbit/s. It contains efficient mechanisms for local adaptivity which are applied for switching between different coding modes and for the control of quality parameters. For this purpose local image analysis is largely performed. Criteria and activity measures are computed in the spatial domain in a regular ways as sums of absolute spatial differences. To some extent analysis is also performed in the DCT domain for coefficient thresholding and an adaptive assignment of Huffman coding classes. An advanced computationally efficient motion estimation method is described. It improves the coding performance and open the possibility for coding with much lower bit-rates. We have performed a comparison experiment at 125 Mbit/s between interlaced and progressive scanned scenes. The most noticeable gain with the progressive scene is the absence of aliasing defects. This positive effect influences the quality more than a slightly increased coding noise and advocates the usage of the progressive scan format for high quality studio application. In a second experiment an interlaced scene is coded at 50 Mbit/s with the usage of the advanced motion estimated prediction scheme. The result is very promising with respect to a possible application for distribution.

Mustererkennung zur benutzerfreundlichen Informationseingabe am Arbeitsplatz

Die elektronische Datenverarbeitung hat fur die Arbeitsplatze in Buro, Konstruktion und Verwaltung vollig neue Moglichkeiten der Informationsverarbeitung geschaffen. Dadurch entsteht das Bedurfnis nach einer benutzerfreundlichen Schnittstelle, die es erlaubt, vorliegende Information schnell, einfach und sicher einzugeben. Ein Weg dazu sind Mustererkennungsverfahren, welche Papiervorlagen in elektronische Daten umsetzen, die mit elektronischen Systemen verarbeitet werden. Je nach Aufgabe, Anwendung und vorgegebenen Vorlagen werden verschiedene Verfahren und Systeme eingesetzt: Beleg- und Formularleser, Zeichnungserfassung fur CAD, Interpretation von Handskizzen und wissensbasierte Dokumentanalyse, die den Inhalt von Dokumenten interpretiert. Diese Systeme werden beschrieben, die wichtigsten Verfahrensansatze diskutiert und der Stand der Technik an ausgewahlten Beispielen erortert. Auserdem wird auf die Dokumentarchitektur ODA eingegangen, die zur Beschreibung von Dokumenten in modernen Burosystemen und zum Datenaustausch angewandtwird.