Talat Anand

Redevelopment of feature extraction algorithms (eg., SIFT) for omnidirectional sensors, by treating the sensor plane as a non-Euclidean Riemannian manifold, rather than the traditional Cartesian plane and through the calculation of the metric tensor for general sensor surfaces. This was in the context of driving assistance systems with four car-mounted omnidirectional cameras, ultimately to estimate structure from motion. Increased matching accuracy from ca. 7% to ca. 44%.

Interfacing a time-of-flight (TOF) camera over Ethernet with Simulink and MATLAB in Windows and with the Robot Operating System (ROS) in Unix. Creation of an API library for conversion of the serial data from the TOF cam into pointCloud2 and Octomap data structures in ROS and into timestamped arrays in Simulink. The data was ultimately used to perform simulateneous localization and mapping in combination with a laser scanner.

--Project I-- Wrote a software suite to train, use and maintain an artificial intelligence system for face recognition. The software was on display at CeBIT 2012 and ‘13. Development of a Bayesian framework based on fat-tail probability distributions for the interpretation of conflicting classification results from video frames. --Project II-- Object recognition. Probabilistic estimation of 3D depth and orientation of recognized objects using single camera 2D images, based on AI and gestalt theory.

Project: Mobile Satellite Communication in the Ka Band „MoSaKa“ Building a Skype-like video conferencing software working on a more robust version of TCP/IP for disaster scenarios.

Blind channel estimation. Error and performance analysis of tensor-based channel estimation algorithms.

Project: „CommonSim“ Simulation of the Ethernet protocol. Creation of a part of a network simulator.

Software development for the enterprise resource planning (ERP) suite, Microsoft Dynamics SL. Built a software to shift ERP systems running on proprietary softwares to MS Dynamics SL, using Microsoft Excel as a stop-over.

Focus on computer vision and artifical intelligence. Took off two years during studies to concentrate on research projects at Fraunhofer Institute für Digitale Medientechnology, Ilmenau and at Fraunhofer Institut für Optronik, Systemtechnik und Bildauswertung/ Angewandte Systemtechnik, Ilmenau

Focus on stochastic numerical analysis, signal processing, image processing.