Image Processing for Broadcast and Video Production

Overview.- The MONA LISA Project, General Presentation.- Session 1: Hardware.- A Parallel Accelerator for Using Synthetic Images in TV and Video Production.- Real-Time Mixing of Live Action and Synthetic Backgrounds Based on Depth Values.- Visual Tools for Parallel Server Handling.- P3I, a Multi-Paradigm Real-Time Video Engine.- Session 2: Analysis 1 - Tracking.- Global Motion Estimation for Registering Real and Synthetic Images.- Real-Time Camera Tracking Server on the ELSET Accelerator.- Tracking and Recognition of Face Sequences.- Session 3: Analysis 2 - Stereoscopic Imaging.- Depth Estimation from Stereoscopic Image Pairs Assuming Piecewise Continuous Surfaces.- 3D Scene Modeling from Stereoscopic Image Sequences.- The Usage of Turntable Sequences for Disparity/Depth Estimation.- Session 4: Analysis 3 - 3D Automatic Modelling.- Joint Estimation of Depth Maps and Camera Motion in the Construction of 3D Models from a Mobile Camera.- Scene Reconstruction Based on Penta-Ocular Image Sequences for 3D-Imaging.- C3D(TM): A Novel Vision-Based 3D Data Acquisition System.- Automatic Modelling of 3D Natural Objects from Multiple Views.- Session 5: Analysis 3 (Continued).- Modelling Buildings from Single Images.- Knowledge Based Modelling of Natural Scenes.- Synthesis 1.- Real-Time Collision Checking for 3D Object Positioning in Sparse Environments.- Session 6: Synthesis 2.- Depth Sensitive Image Compositing in a Multimedia System.- Inverse Kinetics for Center of Mass Position Control and Posture Optimization.- Real-Time Walkthrough with Specular Effects.- Real-Time Depth of Field Algorithm.- Session 7: Synthesis 3.- Combining Method of Generation Realistic Images of Clouds.- Methods for Volume Metamorphosis.- Author Index.