We consider the problem of detecting manmade objects in range data in the presence of extensive clutter. Such situations arise in, for example, the detection of small structures or vehicles beneath a leaf canopy in range data collected from an airborne platform. This problem calls for an extremely robust detection scheme, and it should be fast. Since most manufactured objects comprise large low-order piecewise smooth surfaces (often planes), we focus on detecting locally planar surfaces. We propose a novel technique we call Distribution Weighted Histograms (DWH), which exploits the inherent geometric distributions of man-made objects versus the random occlusions such as those due to an overhanging leaf canopy. The DWH algorithm performs well under heavy occlusion while being computationally inexpensive (linear complexity). We present extensive experimental results.