| 摘要 |
PURPOSE:To attain recognition of a form which is set in a free posture by obtaining coincidence between a picture memory and the feature picture data to produce a silhouette picture and then comparing the result of a product/sum operation carried out between picture memories with the area of a picture data group. CONSTITUTION:A picture is supplied through a TV camera 1 for teaching the form of a subject to be recognized and stored to an input picture memory 4. The picture stored in the memory 4 is processed by a CPU5 to obtain the picture center, the polygonal apex coordinates of an area as well as the area coordinate vector which are stored to a main memory 6 to be used as the dictionary pattern data. The form to be recognized is supplied visually and processed by the CPU5 to obtain a picture center, etc. as the input pattern data. A taught form is regenerated so as to obtain coincidence with the feature of the input pattern data and then converted into a silhouette picture containing a painted-out closed pattern based on the polygonal contour data. Then a result is obtained from a product/sum arithmetic circuit 11. Thus an area where both dictionary and input patterns overlap with each other is obtained, and the coincidence between both patterns is decided when the difference of areas between both patterns is less than the threshold value.
|
