Face Detection in Low-Resolution Color Images

In low-resolution images, people at large distance appear very small and we may be interested in detecting the subject’s face for recognition or analysis. However, recent face detection methods usually detect face images of about 20×20 pixels or 24×24 pixels. Face detection in low-resolution images has not been explicitly studied. In this work, we studied the relationship between resolution and the automatic face detection rate with Modified Census Transform and proposed a new 12bit Modified Census Transform that works better than the original one in low-resolution color images for object detection. The experiments show that our method can attain better results than other methods while detecting in low-resolution color images.