We introduce an algorithm of morphological filters and propose its use to classic polarization metrics for applications requiring passive longwave-infrared, polarimetric remote sensing and real-time anomaly detection. The approach significantly augments the daytime and nighttime detectability of weak-signal manmade objects immersed in a predominant natural background scene. A tailored sequence of signal-enhancing filters is featured, consisting of basic and higher level morphological operators to achieve a desired goal. Qualitatively, the goal is to effectively squeeze the variance of pixel values representing the natural clutter background, while simultaneously spreading the pixel variance within the manmade object class and separating the pixel mean averages between the two classes of objects. Using real data, the approach persistently detected with a high confidence level three mobile military howitzer surrogates (targets) from natural clutter, during a 72-h coverage. Targets were posed at three aspect angles (range 557 m), yielding a negligible false alarm rate. Performance was invariant to diurnal cycle and mild atmospheric changes.