An efficient scanning based learning free algorithm for face detection

Efficient and effective detection of face area is prerequisite for a fast face recognition application. In contrast to the conventional face detection process which requires a substantial learning time due to the training of large face database, this paper presents a simple learning free face detection algorithm based on scanning. In our proposed algorithm after converting the input image into binary form, the scanning process outputs the total contour of the face. With the head point and most frequent length of that final contour we can detect the face area by our method. While most of the existing approaches can detect only frontal faces, our proposed algorithm can detect faces from side views as well as traditional frontal views. The algorithm has been extensively tested with the critical MIT face database and has shown significant performance, specifically 98.3% success rate and 91.5% accuracy.