Multiple‐valued image‐contour extraction method using a Laplacian– Gaussian filter
Noriaki Muranaka, Shinya Kudoh, Takanori Ashida, Masataka Tokumaru, Shigeru Imanishi · Systems and Computers in Japan · 2007
Abstract As a preparation for image recognition and character recognition in image processing, it is often necessary to separate the target region from other regions. To extract the edges of an object, binarization with a Laplacian filter or a Laplacian– Gaussian filter has been widely used. Such a method has problems, however: for example, the extraction of noise rather than the real edges of an object; failure to extract parts of a visually noticeable edge; and difficulty in setting thresholds for edge extraction. To solve these problems, this paper proposes a multiple‐valued image‐contour extraction method using a 3‐valued L‐G filter or a 5‐valued full‐color extraction method. The effectiveness of the proposed method and the realization of a multiple‐valued logic circuit are examined. © 2007 Wiley Periodicals, Inc. Syst Comp Jpn, 38(8): 61– 71, 2007; Published online in Wiley InterScience ( www.interscience.wiley.com ). DOI 10.1002/scj.10467