Partial face recognition: An alignment free approach

Abstract

Abstract-Numerous methods have been developed for holistic face recognition with impressive performance. However, few studies have tackled how to recognize an arbitrary patch of a face image. Partial faces frequently appear in unconstrained scenarios, with images captured by surveillance cameras or handheld devices (e.g. mobile phones) in particular. In this paper, we propose a general partial face recognition approach that does not require face alignment by eye coordinates or any other fiducial points. We develop an alignment-free face representation method based on Multi-Keypoint Descriptors (MKD), where the descriptor size of a face is determined by the actual content of the image. In this way, any probe face image, holistic or partial, can be sparsely represented by a large dictionary of gallery descriptors. A new keypoint descriptor called Gabor Ternary Pattern (GTP) is also developed for robust and discriminative face recognition. Experimental results are reported on four public domain face databases (FRGCv2.0, AR, LFW, and PubFig) under both the open-set identification and verification scenarios. Comparisons with two leading commercial face recognition SDKs (PittPatt and FaceVACS) and two baseline algorithms (PCA+LDA and LBP) show that the proposed method, overall, is superior in recognizing both holistic and partial faces without requiring alignment.

People Identification and Tracking Through Fusion of Facial and Gait Features

“…An occluded query face is reconstructed by a linear combination of gallery images before being assigned to the class with the minimal reconstruction error. 2) local matching based approaches [52], [67], [66], [48], [77], [78], [75] that extract features from the local areas of a face (e.g., patches), such that the affected and unaffected parts of the face can be analyzed separately. To minimize matching errors of the occluded parts, several strategies can be used such as local space learning [66], [52], [67], multi-task sparse representation learning [48] or voting [75].…”

“…2) local matching based approaches [52], [67], [66], [48], [77], [78], [75] that extract features from the local areas of a face (e.g., patches), such that the affected and unaffected parts of the face can be analyzed separately. To minimize matching errors of the occluded parts, several strategies can be used such as local space learning [66], [52], [67], multi-task sparse representation learning [48] or voting [75]. 3) occlusion-insensitive feature based approaches [10], [70], [88] that utilize features such as line segments [10], image gradient orientation (IGO) difference [70] and the Gabor phase (GP) difference [88] which were shown to be robust to occlusion.…”

Active Self-Paced Learning for Cost-Effective and Progressive Face Identification

“…To interact with these photos, there have been increasing demands of developing intelligent systems (e.g., content-based personal photo search and sharing from either his/her mobile albums or social network) with face recognition techniques [1], [2], [3]. Thanks to several recently proposed pose/expression normalization and alignment-free approaches [4], [5], [6], identifying face in the wild has achieved remarkable progress. As for the commercial product, the website "Face.com" once provided an API (application interface) to automatically detect and recognize faces in photos.…”

A Novel Pose and Light Invariant Face Recognition System in Videos Using Advanced Local Directional Patterns

“…For face based biometric authentication scheme, traditionally face recognition has been considered where the responsibility has been to match a captured face against sets of faces stored in the database [1]. In 1991 it was proposed [2], Starting from popular Eigen face based face recognition technique, many algorithms are being proposed which focuses on solving specific challenges in face recognition like pose invariance [3], illumination invariance [4], partial face recognition [5], expression invariance [6], face recognition in night [7] and so on. Focus on unconstrained face recognition [8] most of the studies were towards practical face recognition system, which takes into consideration of one or more practical variations mentioned above.…”