Passive millimeter-wave imaging systems can play a significant role in security applications. Especially, the detection of hidden threats for border security is a growing field. In this paper we propose a novel approach for automatic threat detection using multiple 94 GHz passive millimeter-wave images. Herein, we discuss four steps essential to solving the task: pre-processing, region-of-interest extraction, threat extraction in each frame and, finally, intelligent fusion of the results from all frames. Besides, showing that the proposed method works reliably for the data-set at hand, we also discuss the advantages of using this method in contrast to state-of-the-art methods.