Region extraction for real image based on fuzzy reasoning

Abstract

An approach to extracting regions of a natural object by binarization using fuzzy reasoning is proposed. An advantage of this approach is that the threshold is obtained by fuzzy reasoning based on the shapes instead of the user having to determine the threshold by trial and error. This approach was applied to extracting the image of a real flower as an example of a natural object. In experiments, the image of a real flower was used whose color was the same as the background. It was demonstrated that the approach was more effective than statistical methods for extracting objects whose shape is almost known.<>