| The past few years have seen a growing interest in the application of three-dimensional image processing. With the increasing demand for 3-D spatial information for tasks of passive navigation, automatic surveillance, aerial cartography, and inspection in industrial automation, the importance of effective stereo analysis has been made quite clear. A particular challenge in this area is to provide reliable and accurate depth data for input to object or terrain modelling systems (such as ACRONYM). This thesis describes an algorithm for such stereo sensing. It is founded on an edge-based line-by-line stereo correlation scheme--one which provides this three dimensional analysis in a fast, robust, and parallel implementable way. Its processing consists of extracting edge descriptions of a pair of images, linking these edges to their nearest neighbors to obtain the connectivity structure of the images, correlating the edge descriptions on the basis of local edge measures, and cooperatively removing those edge correspondences formed by the correlation which violate the connectivity structure of the two images. A further correlation process, using a technique similar to that used for the edges, is done on the image intensity values within intervals defined by the preceding correlation. The result of the processing is a full image array depth map of the scene viewed. |
