| 摘要 |
PURPOSE:To easily and highly precisely detect plural light and dark edge positions from a picture to which plural pieces of edge information are included by extracting edge position information from space data based on plural and rough edge position information detected by means of a rough edge detection means. CONSTITUTION:When the picture of a measured object, which includes two pieces of edge information, is inputted to a picture input means 1, sampled picture data is transformed into a quantity in the spatial frequency domain by a Fourier transformation means 3. A filter 4 cuts the spatial frequency component of 0-th light and an inverse Fourier transformation means 5 inversely Fourier-transforms data and it is returned to original space information. An edge extraction means 7 simultaneously separates two edges corresponding to appropriate positions from plural drops in space information by using space information obtained by the inverse Fourier transformation means 5 and rough edge position information detected by the rough edge detection means 6 so as to detect the edge positions. |
