Affordable Access

Shadow Detection in Dynamic Scenes Using Dense Stereo Information and an Outdoor Illumination Model

Publication Date
  • Mathematics


Shadow Detection in Dynamic Scenes Using Dense Stereo Information and an Outdoor Illumination Model Claus B. Madsen1, Thomas B. Moeslund1, Amit Pal2, and Shankkar Balasubramanian2 1 Computer Vision and Media Technology Lab Aalborg University, Aalborg, Denmark [email protected] 2 Department of Electronics and Communication Engineering Indian Institute of Technology Guwahati, Assam, India Abstract. We present a system for detecting shadows in dynamic out- door scenes. The technique is based on fusing background subtraction operations performed on both color and disparity data, respectively. A simple geometrical analysis results in an ability to classify pixels into foreground, shadow candidate, and background. The shadow candidates are further refined by analyzing displacements in log chromaticity space to find the shadow hue shift with the strongest data support and rul- ing out other displacements. This makes the shadow detection robust towards false positives from rain, for example. The techniques employed allow for 3Hz operation on commodity hardware using a commercially available dense stereo camera solution. Keywords: Shadows, stereo, illumination, chromaticity, color. 1 Introduction Shadows are an inherent part of images. Especially in outdoor vision applica- tions shadows can be a source of grave problems for the processing and analysis of video data. For humans, though, shadows represent a significant cue to un- derstanding the geometry of a scene, and to understanding the illumination con- ditions, which in turn helps processing the visual data. In this paper we present an approach to accurately identifying shadow regions in outdoor, daylight video data in near real-time (presently around 3 Hz, with potential for significant im- provement). The main contributions of this work lie in utilizing a combination of color and dense depth data from a stereo rig for an initial, rough shadow de- tection, combined with a model-based chromaticity ana

There are no comments yet on this publication. Be the first to share your thoughts.