An improved multilayer self-organizing background subtraction algorithm for microorganism detection in sewage

In this paper, a new image sequence model, obtained by learning in a multilayer self-organizing neural network, is proposed for moving microorganism detection in sewage treatment system. The model is able to handle diverse challenging scenarios accurately, such as dynamic background, gradual illumination variations, shadows cast and so on, which are robust against false detections for different types of micro-videos. Experimental results demonstrate its effectiveness compared with other state-of-the-art methods.