Efficient real-time face detection for high resolution surveillance applications

Abstract

This paper presents an efficient face detection method suitable for real-time surveillance applications. Improved efficiency is achieved by constraining the search window of an AdaBoost face detector to pre-selected regions. Firstly, the proposed method takes a sparse grid of sample pixels from the image to reduce whole image scan time. A fusion of foreground segmentation and skin colour segmentation is then used to select candidate face regions. Finally, a classifier-based face detector is applied only to selected regions to verify the presence of a face (the Viola-Jones detector is used in this paper). The proposed system is evaluated using 640 × 480 pixels test images and compared with other relevant methods. Experimental results show that the proposed method reduces the detection time to 42 ms, where the Viola-Jones detector alone requires 565 ms (on a desktop processor). This improvement makes the face detector suitable for real-time applications. Furthermore, the proposed method requires 50% of the computation time of the best competing method, while reducing the false positive rate by 3.2% and maintaining the same hit rate.

Real-time acquisition of high quality face sequences from an active pan-tilt-zoom camera

“…Then the geometric center of face region is detected and the distance between the center of region and center of mass is calculated by: (10) The final single score for each face is calculated by linearly combining the abovementioned 4 quality parameters with empirically assigned weight factor, as shown in (11) (11) Where, w i are the weight associated with P i , and P i are the score values for the parameters pose, sharpness, brightness and resolution, consecutively.…”

“…For real-time face detection in high resolution image frames, Mustafah et al proposed two high resolution (5 MP) smart camera-based systems by extending the works of [9][10]. Cheng et al also proposed a face detection method in high resolution imagery by employing a grid sampling and a skin color based approach [11]. For dealing with too low-resolution faces in surveillance videos in [3] and [12], networks of two active cameras were proposed to detect faces in a wideview camera and later extract high resolution face images in a narrow-view camera by using PTZ control.…”

Face detection with a Viola–Jones based hybrid network

“…Further efficiencies have been achieved by limiting the Viola-Jones search window to pre-selected regions based on segmentation by skin colour and the approximation of minimum face size [11][12][13]. Viola-Jones has also been combined with a skin colour filter within a hybrid arrangement [14].…”

Constructing Facial Expression Log from Video Sequences using Face Quality Assessment

“…Facial analysis systems now-a-days have been employed in many applications including surveillance, medical diagnosis, biometrics, expression recognition and social cue analysis (Cheng, 2012). Among these, expression recognition received remarkable attention in last few decades after an early attempt to automatically analyse facial expressions by (Suwa, 1978).…”