Thresholding for edge detection using fuzzy reasoning technique

摘要

Threshold values for extracting effective sketch features arendifferent according to images. Moreover, a global threshold cannotnextract sketch features effectively because of the fact that there arenvarious types of objects and regions having many different gray levelsnwithin a single image, and the human visual characteristics are morensensitive to the sketch features in dark regions than in bright regions.nTherefore, in order to extract sketch features in a manner based on thenhuman visual characteristics, this paper suggests a method for decidingnwhether a pixel is a feature or not, deduced through fuzzy reasoningnutilizing a contrast and a local brightness as its input measures. To donthis, a contrast and a local brightness are proposed and thesenmembership functions are generated subject to the histograms which arenconstructed from applying these measures to the input image. In order tondecide the degree of threshold values according to image context, annoutput membership function is generated subject to the standardndeviation of input image. Several experimental results are presented tonsupport the validity of proposed method