Enhanced face detection framework based on skin color and false alarm rejection

Abstract

Fast and precise face detection is a challenging task in computer vision. Human face detection plays an essential role in the first stage of face processing applications such as recognition tracking, and image database management. In the applications, face objects often come from an inconsequential part of images that contain variations namely different illumination, pose, and occlusion. These variations can decrease face detection rate noticeably. Besides that, detection time is an important factor, especially in real time systems. Most existing face detection approaches are not accurate as they have not been able to resolve unstructured images due to large appearance variations and can only detect human face under one particular variation. Existing frameworks of face detection need enhancement to detect human face under the stated variations to improve detection rate and reduce detection time. In this study, an enhanced face detection framework was proposed to improve detection rate based on skin color and provide a validity process. A preliminary segmentation of input images based on skin color can significantly reduce search space and accelerate the procedure of human face detection. The main detection process is based on Haar-like features and Adaboost algorithm. A validity process is introduced to reject non-face objects, which may be selected during a face detection process. The validity process is based on a two-stage Extended Local Binary Patterns. Experimental results on CMU-MIT and Caltech 10000 datasets over a wide range of facial variations in different colors, positions, scales, and lighting conditions indicated a successful face detection rate. As a conclusion, the proposed enhanced face detection framework in color images with the presence of varying lighting conditions and under different poses has resulted in high detection rate and reducing overall detection time.