Scene Adaptive Shadow Detection Algorithm

Robustness is one of the primary performance criteria for an Intelligent Video Surveillance (IVS) system. One of the key factors in enhancing the robustness of dynamic video analysis is, providing accurate and reliable means for shadow detection. If left undetected, shadow pixels may result in incorrect object tracking and classification, as it tends to distort localization and measurement information. Most of the algorithms proposed in literature are computationally expensive; some to the extent of equalling computational requirement of motion detection. In this paper, the homogeneity property of shadows is explored in a novel way for shadow detection. An adaptive division image (which highlights homogeneity property of shadows) analysis followed by a relatively simpler projection histogram analysis for penumbra suppression is the key novelty in our approach. Keywords—homogeneity, penumbra, projection histogram, shadow correction