| 摘要 |
PURPOSE:To extract the form of a model from picture information within the limit of parallel processing operation by putting the model and input picture information one over the other with respect to feature elements of the form defined as direction faces. CONSTITUTION:A model storage means 601, a direction face generating means U, a couner face storage means C, a moving means 615, an adding means 616, and a repeating means are provided, and the means 601 expresses the model of the form to be extracted as a sequence of evaluation points and defines the value related to the directional feature of each evaluation point and the extent of movement from the just preceding evaluation point and preliminarily stores defined values related to direction features and extents of movement. A means 622 is provided which detects the form and the position of the model in accordance with the distribution of the total sum of direction faces at each point of the counter face. Thus, the model and input picture information are put one over the other with respect to feature elements defined as direction faces to extract the form of the model from picture information within the limit of parallel processing operation. |
