Moving Shadow and Object Detection in Traffic Scenes

We present an algorithm for segmentation of traffic scenes that distinguishes moving objects from cast shadows. Three image features at each pixel site are considered: brightness, and normalized red and blue color components. Each feature is analyzed by an a posteriori probability estimator that computes membership probabilities for the three classes: background, foreground and shadow. The algorithm iterates between the three estimators by using the output of one as the a priori probability input to the next. This approach is used in iterative decoders, a component of turbo codes. Thus, we have named the algorithm turbo segmentation. A fading memory estimator calculates mean and variance of features for background pixels. Given the statistics for a background pixel, simple rules for calculating its statistics when covered by a shadow are used. In addition to the color features, we examine the use of neighborhood information to produce smoother classification. We also propose the use o...

• Type: Conference
• Year: 2000
• Where: ICPR
• Authors: Ivana Mikic, Pamela C. Cosman, Greg T. Kogut, Mohan M. Trivedi
• Keywords: Background Pixel, ICPR 2000, Posteriori Probability Estimator, Priori Probability Input