Face detection, tracking and recognition is still actual field of human centered technologies used for developing more natural communication between computing artefacts and users. Analyzing modern trends and advances in this field, two approaches for face sensing and recognition have been proposed. The first color/shape-based approach uses sets of fuzzy saturated color regions providing face detection by Fourier descriptors and recognition by SVM. The second approach provides fast face detection by adaptive boosting algorithm, and recognition based on SIFT key point extraction into eye-nose-mouth regions has been improved using Bayesian approach. Designed systems have been tested in order to evaluate capability of the proposed approaches to detect, trace and interpret faces of known individuals registered into facial standard databases providing correct recognition rate in range of 94.5-99.0% with recall up to 46%. The conducted tests ensure that both approaches have satisfactory performance achieving less than 3 seconds for human face detection and recognition in live video streams.
TopIntroduction
Face sensing and recognition is still open problem despite the fact that there are many publications and scientific reports in this research field. Currently, automatic face recognition is used in affective computing and usability testing, robot guidance, distance education, security monitoring, automatic environment inspection, marketing studies and consumption, control and surveillance systems, training people for more effective interpersonal communication, and others (Zhang, 2010; Garcia-Amaro, 2012; Singla, 2014; Granger, 2014; Nappi, 2015). In the mobile autonomous systems particularly, if this process is carried out in real time, the ability of a robot to recognize a human in video stream and track continuously his face is the most important feature of human-machine interaction (Parmar, 2013; Unar, 2014; Vezzetti, 2014; Huang, 2015).
After analysis of known relevant systems for face detection, recognition and tracking, the standard set of used approaches may be classified into a generalized architecture as it is shown in Figure 1.
Figure 1. Generalized architecture of a systems for face detection, recognition, and tracking
As it is depicted in Figure 1 after image acquisition, noise reduction, scaling and enhancement the global or local feature extraction is applied. These processes as well as image segmentation and features representation in particular structures define the precision of the detection of regions of interest and simplify the face recognition. Each process is controlled by knowledge base (KB) that a priory contains all the necessary data and rules. On the preprocessing step, for example, the knowledge base operates with appropriate parameters for used filters under specific conditions of input images (noise, contrast, color distribution, etc.) or in case of feature extraction and segmentation, KB provides appropriate method or transform and then defines how and which region of interest must be segmented. Particular importance the KB has for classification and interpretation of features providing recognition of regions of interest according to used classification approaches. Therefore, the generalized architecture provides standard reference for developing new systems ensuring their high performance.
The particular requirements for designing systems for face detection and recognition may be subdivided in the following way:
- •
A system must have a capacity to extract relevant features from images as well as to represent and encode them in suitable manner for fast and precise processing;
- •
Used classifier must realize inferences providing accurate recognition on base of incomplete information;
- •
A software for decision making must interpret the obtained data and learn on examples to be used in the future circumstances (Gonzalez, 2009; Meva, 2014; Starostenko, 2015b).