A bottom-up audio attention model for surveillance | IEEE Conference Publication | IEEE Xplore