Partial Face Recognition: Alignment-Free Approach

Abstract

Many approaches have been developed for holistic face recognition with impressive performance. However, few studies have addressed the question of how to recognize an arbitrary image patch of a holistic face. In this paper we address this problem of partial face recognition. Partial faces frequently appear in unconstrained image capture environments, particularly when faces are captured by surveillance cameras or handheld devices (e.g. mobile phones). The proposed approach adopts a variable-size description which represents each face with a set of keypoint descriptors. In this way, we argue that a probe face image, holistic or partial, can be sparsely represented by a large dictionary of gallery descriptors. The proposed method is alignment free and we address large-scale face recognition problems by a fast filtering strategy. Experimental results on three public domain face databases (FRGCv2.0, AR, and LFW) show that the proposed method achieves promising results in recognizing both holistic and partial faces.

3D Face Recognition Based on Multiple Keypoint Descriptors and Sparse Representation

“…Inspired by [41], [50], we adopt the following multi-task SRC to determine the identity of the probe face scan:where δ c (·) is a function which selects only the coefficients corresponding to class c . Eq.…”

“…Hence, we adopt a similar idea as Liao et al . [41] to derive a fast approximate solution. For each probe descriptor y i , we first compute: …”

“…In such an approach, we require neither the alignment of facial range images nor the detection of meaningful fiducial points. Our approach is highly motivated by the success of a recent work designed for 2D partial face matching, namely MKDSRC ( M ultiple K eypoint D escriptors and S parse R epresentation based C lassification) [41]. MKDSRC proposed by Liao et al .…”

“…MKDSRC proposed by Liao et al . [41] is an alignment-free 2D partial face matching approach, in which each face is represented by a set of descriptor vectors extracted from keypoints and a multi-task SRC is used for classification. Such a method can address the problem of 2D partial face matching pretty well.…”

Access Control in Smart Homes by Android-Based Liveness Detection

“…Light conditions, facial expressions, face rotations, age, physical radical changes, as well as the presence of partial occlusions, such as glasses, facial hair, or hair, covering part of the face, are just some of the elements which make the automatic face recognition challenging. Despite these issues, algorithms that allow to obtain satisfactory results of personal identification have been proposed in the literature [19][20][21].…”