| 摘要 |
In order to process images at a high speed without occupying much room in the memory of a data processing system used to this effect, the main points of such image called marker are first searched for. They are then arranged in a system of hierarchical queues as a function of the value of an image parameter assigned to said markers. Then, the processing is started by the point situated closer to the base of the queue having the highest priority of the system. The processing is comprised on the one hand of the attribution to the processed point of a label which may be used later as data for visualising the point, and, on the other hand, of entering in the queue the neighbours in the image of such point with priority levels which depend either on the image parameters or on the image parameters of the processed point or on a combination of both.
|
