Background subtraction is widely used in multimedia applications, such as traffic monitoring, video surveillance, and object tracking. Several methods with different advantages in different applications have been proposed. The advent of cloud computing also has made possible of the combination of various background subtraction techniques and the processing of large amounts of images. In this paper, an integrated algorithm for background subtraction is implemented and analyzed. The proposed AdaBoost algorithm combines weak classifiers: pixel-based background subtraction methods, block-based background subtraction methods, and graph-cut segmentation methods. After training, the program adjusts the weight of each weak classifier. The algorithm is accelerated using Hadoop cloud-computing architecture. By using a MapReduce framework, this system can parallel-processing on multiple servers in order to reduce computing time. When the system completes its task, the user can see the combined results on the screen and then choose the preferred result. The system can obtain user feedback and tune the combination mechanism.