The PRC segmentation algorithm is a graph-based method that relies heavily on the concept of ``minimum spanning trees'' (Zahn, 1971). In short, the algorithm makes comparisons between pairs of nearby pixels in a migrated image, and determines whether the pixels belong in the same image segment, or in separate regions. See Halpert (2010) for full details on how this method is adapted for seismic images. This algorithm is very efficient compared to other segmentation techniques such as NCIS; for example, the 1000 x 1000 pixel 2D result here was computed in less than a minute, while the 3D result (1000 x 800 x 40) required 10 minutes on a single processor. Parallelization of this scheme is made possible by operating simultaneously on different chunks of the input cube, although additional pre- and post-processing is necessary. Segmentation results (for example, Figures 2(a) or 5(a)) are shown as random-color segments overlying the image. Very few parameters are required; the user controls the minimum segment size, and can quickly choose segments to merge together in two or three dimensions. All results shown here required merging 5-10 segments within the salt body.