Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: LOCAL BACKGROUND SPRITE GENERATION Andreas Krutz 1 , Alexander Glantz 1 , Michael Frater 2 , and Thomas Sikora 1 1 Communication Systems Group, Technische Universit¨at Berlin, EN-1, Einsteinufer 17, 10587 Berlin, Germany, { krutz,glantz,sikora } @nue.tu-berlin.de 2 School of Information Technology and Electrical Engineering, University of New South Wales, Canberra ACT 2600, Australia, m.frater@adfa.edu.au ABSTRACT Background modeling of video sequences can be used in many different applications. For video object segmenta- tion it is often applied in a background subtraction method. When conventional sprites like single or multiple sprites are used a background sequence has to be reconstructed from the model. The double mapping into the coordinate system of the sprite and back can lead to severe distortion of the background model and therefore to erroneous seg- mentation masks. We present a novel background model- ing approach that lessens distortion. These so-called local background sprites are built for every reference frame in- dependently and fit its original size. Experimental results show that this new approach clearly outperforms conven- tional background sprites in terms of PSNR. 1. INTRODUCTION Background modeling – meaning the description of the background of a video sequence – is an important research field. Application scenarios range from video surveillance systems to video coding. A so-called background sprite or single sprite gener- ated over a certain number of frames can be such a back- ground model. A background sprite is an image that typ- ically is of large size and contains only the background pixels of a video sequence. This approach can be extended to so-called multiple sprites or super-resolution sprites. In case of multiple sprites the video sequence is divided into parts and for each part one background sprite is generated independently. In case of super-resolution a higher quality background sprite is generated for a sequence. The potential for using background sprites for object- based video coding has been summarized in [1]. Fur- thermore, background modeling is an efficient means for video object segmentation. Various approaches using sin- gle or multiple background sprites in a background sub- traction method have demonstrated that this kind of back- ground model is very promising [2], [3]. However, the mapping of pixel content from various frames in a scene into a single sprite or a collection of multiple sprites may cause severe geometrical distorion of the background. For reconstruction of the background of a single frame a sec- ond mapping needs to be performed which causes addi- tional distortion. Object segmentation is one application Figure 1. Single sprite, sequence “Stefan”, reference frame 253 that is degraded by such distortion....
View Full Document

This note was uploaded on 04/22/2010 for the course MI IP taught by Professor Vladbalan during the Spring '10 term at Universidad del Rosario.

Page1 / 10


This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online