Component Based Method for Face Detection using Fuzzy Membership Functions

摘要

In this paper we address the problem of face detection in grayscale images. Before make candidates we calculate membership face components that influence to make more accurate answer and rejecting false positives in detection. The proposed method detects all candidate windows that has enough membership of face components. Eyes, eyebrows and Lip are organized to build fuzzifier membership functions using their pattern, distances of their location in 2D plan, and edginess of forehead.rnα cut is used as defuzzifier and a parameter to achieve fuzzyrnbased decision using empirical face features. A novel and simplernfuzzy method for face detection using the face structural model isrnintroduced. Due to using fuzzy membership functions for facerndetection, its rate using has increased over the original qualitativernmodel for face detection. The experiments show promisingrnresults compared to the original structural method without usingrnfuzzy.