Moving object detection based on improved Gaussian mixture model

In order to solve the problem of background modeling by Gaussian mixture model(GMM) with constant K components, such as computational complexity and being not able to perfectly represent the status of pixels in the region with complex movement. The algorithm of moving objects detection method with improved Gaussian mixture model was proposal. The proposed algorithm improves the traditional GMM in the aspects of adaptively adjusting the model parameters learning rate and the number of components with the advantage of being less computational complex, more reliable and higher robust. The experiment result with actual surveillance video shows lower computational complexity and higher robustness.