Detecting Faces in Noisy Images using Hit-Miss Transform (HMT)

Abstract

The object of this paper is to detect faces in noisy images. All sample images show homogeneous background faces. The transformation of the Hit-Miss is used to detect object boundaries and remove the noise effect. Two filters are cascaded to handle high noise levels in a special way. An application for face detection in noisy conditions is presented. The algorithm is implemented on faces taken from the Manchester face database after adding Gaussian noise to them.