This paper has developed a generalised sampling strategy for the rapid location of objects in digital images. In this strategy a priori information on the possible locations of objects is used to guide the sampling process, and earlier body-based and edge-based approaches emerge automatically on applying the right a priori probability maps. In addition, the limitations of the earlier regular sampling technique have been clarified and eased - with the result that sampling patterns are better matched to the positions of the image boundaries. These methods lead to improved speeds of operation both in the cases where all the objects in an image have to be located and also where the positions of individual objects have to be updated. Finally, the method is interesting in being intrinsically able to perform full binary search tree edge location without the need for explicit programming.