PARTIAL FACE RECOGNITION USING DYNAMIC FEATURE MATCHING AND CNN
R Bharath, K. Arunkumar, Shyam Diwakar, P. Dineshkumar · Journal of Emerging Technologies and Innovative Research · 2020
As the computer generation grows, security and safety are being an active topic. The most active and most advanced problem is face detection and face recognition. Nowadays, detection and recognition of frontal face images are achieved in all computer fields, from identification to authorization and authentication. Still, the problem is to deal with a partial face. Sometimes the obtained picture may be occlusions, large-viewing angles, or out-of-view. Identifying a partial face and recognition isn't an easy task. We propose a technique in this paper to detect the partial face and recognize them using various steps. Our method helps to identify the partially appearing faces with very low computational complexity. We couple the detected facial segment, which allows detecting partially visible faces. Once a partial face is detected, the facial sections can be used in many applications. The partially detected probe is sent to a face recognition model to discover the person. We use FCNs (Fully Convolutional Networks) and SRC (Sparse Representation Classification) combined using DFM (Dynamic Feature Matching) to handle the recognition of partial face regardless of different sizes.