Evaluation of Morphological Edge Detector Algorithms for Passive Millimeter Wave Imaging

Passive millimeter-wave (PmmW) imaging presents challenges in edge detection due to low contrast and high noise, which impact object recognition in security applications. Here, a selection of conventional edge detection techniques is cross-evaluated for their applicability to mmW imaging. Special attention is given to the size of the sliding window or box filter applied that takes advantage of the spatial oversampling obtained as a result of diffraction-limited resolution of the generated mmW images. Performance is assessed based on visual clarity, computational efficiency, and detection accuracy. While several algorithms have notable difficulties with imaging technologies at millimeter wavelengths, it has been found that simple standard deviation, when paired with adaptive mean thresholding, produces relatively lower clutter, and higher-quality edge maps compared to other methods.