A Quick Self-Adaptive Background Updating Algorithm Based on Moving Region

Background subtraction is a common way of real-time motion detection. Owing to lacking the right approach to update the background, it was confined in certain aspects. In this paper, we will present a quick selfadaptive background updating algorithm based on moving region, and use label method to de-noise the moving objects. The algorithm can reduce the calculation and accelerate the background subtraction. It also can update the background exactly along with the variance of illumination, and meet the real-time processing. While the label method can de-noise all the noises in the moving objects, the moving objects can be segmented from the background perfectly. The images in our experiment are obtained from the traffic scenes, and the experimental results show that this builds a good foundation for further shadow elimination and object recognition.