Neuro-fuzzy shadow filter

Abstract

In video sequence processing, shadow remains a major source of error for object segmentation. Traditional methods of shadow removal are mainly based on colour difference thresholding between the background and current images. The application of colour filters on MPEG or MJPEG images, however, is often erroneous as the chrominance information is significantly reduced due to compression. In addition, as the colour attributes of shadows and objects are often very similar, discrete thresholding cannot always provide reliable results. This paper presents a novel approach for adaptive shadow removal by incorporating four different filters in a neuro-fuzzy framework. The neuro-fuzzy classifier has the ability of real-time self-adaptation and training, and its performance has been quantitatively assessed with both indoor and outdoor video sequences.