A ROTATION INVARIANT FACE DETECTION METHOD USING SUB-WINDOW ROTATION IN ADABOOSTING ALGORITHM

摘要

A method for detecting an inclined face by sub-window rotation in an adaboosting algorithm is provided to detect an inclined face in real time by using an adaboosting algorithm to which sub-window rotation is applied, thereby detecting multiple faces from real time images in a short time with a high detection rate. A method for detecting an inclined face by sub-window rotation in an adaboosting algorithm includes the steps of setting a sorter on an off-line by a sub-window serving as a basic window for detecting faces in a predetermined picture by the adaboosting algorithm to which Haar-like wavelet is employed(S_1), mounting the set sorter to a control computer(S_2), and generating a plurality of rotation sub-windows by rotating the sub-window of the sorter by a predetermined angle(S_3), and detecting faces from image data stored in a database by the basic window and the rotation sub-windows and displaying the rotation sub-windows used for the face detection to the image data(S_4).