Face detection system design for real time high resolution smart camera

Square patch feature based face detection architecture for high resolution smart camera

“…In this paper we present an FPGA based face detection architecture that can be used with the method in [3] on a smart camera such as what was presented in [1] to build a complete high resolution smart camera system with face detection capability. The face detection algorithm is constructed using a faster weak classifier that we called square patch feature and a response binning boosting algorithm [16].…”

“…For this algorithm, the required processing time highly depends on the number of search windows that quadratically increases when image resolution is increased. To reduce the search space of such algorithms, a method was proposed in [3] that utilizes background subtraction and skin color detection algorithm to reduce the region of interest inside an image as shown in Figure 2. With this method the region of interest in the image will be reduced to only the moving skin region, which is a good candidate for face detection.…”

“…Face detection is defined as the problem of locating a face in an image in the presence of uncontrolled lighting, facial Region of Interest Classification Method [3].…”

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

“…An appearance-based approach to real-time face detection from video frames was proposed in [8] which merely work in real-time in VGA resolution images. 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].…”

Recent Advances in Region-of-interest Video Coding

“…Such for example, (Mustafah et al, 2009) proposes a design of a face detection system for real-time high resolution smart camera, while making an emphasis on the problem of crowd surveillance where the static color camera is used to monitor a wide area of interest, and utilizing a background subtraction method to reduce the Region-of-Interest (ROI) to areas where the moving objects are located. Another work was performed by (Zhang et al, 2009), in which was presented a ROI based H.264 encoder for videophone with a hardware macroblock level face detector.…”