| 摘要 |
A method and apparatus for recognizing objects by their contour representations being similar to the contour of a reference object. The method and apparatus includes dividing the contour representation into scanning lines, coding the lines into code line digits, determining the number of intersections of the background and contour in each line, determining transitions from lines with one number of intersections to lines with another number of intersections, automatically fixing lines at the places of transitions, dividing the contour representation by the fixed lines into sections, forming section codes of the sections from present digits of respective section line codes, detecting in the section line code being obtained code portions confined between the code digits corresponding to the intersections in at least one of the fixed lines defining the boundaries of the sections, determining closed-open contour characteristics of the contour portions between intersections in the fixed lines by the character of the code portions, and determining by the obtained closed-open contour characteristics of the contour portions whether the object being recognized belongs to one of the reference objects.
|
