[Paroxysmal tachycardia].
W STUDART · PubMed · 1953
In video telephone scenes to be transmitted at low data rates, non--diffuse illumination has to be expected. Therefore, in this work the scene illumination is automatically estimated from a monocular image sequence in addition to shape and motion of moving objects. Image analysis uses the source model "moving rigid 3D objects with nonuniform local illumination". The irradiance is modelled by a discrete 3D irradiance map. Experiments report an interframe prediction error gain over a former method ignoring illumination. This gain depends on the amount of object motion and the nonuniform portion of scene illumination. A first experiment analysing a real image sequence states an interframe prediction gain of about 1dB. 1 Introduction In Object--Based Analysis--Synthesis Coding (OBASC) [2], three dimensional information has to be automatically extracted from two dimensional projections of a real 3D scene containing moving 3D objects. Working on a monocular image sequence, a 3D model based...