摘要 |
An image processing system produces clusters of related objects for subsequent analysis. It receives a multi-level digital representation of the image and predetermined objects are identified by extraction units (4). A segmentation unit (6) drives boundary data from the multi-level digital representation, the boundary data representing boundaries between different regions of the image. A clustering unit (8) clusters the predetermined objects into groups in dependence on their proximity to each other and in dependence on the boundary data and supplies this cluster data for subsequent analysis. |