Pose Invariant Face Recognition Under Arbitrary Illumination Based on 3D Face Reconstruction

Abstract

Abstract. Pose and illumination will be quite difference between probe and gallery images in many face recognition tasks. In this paper, a novel technique for face recognition under varying poses and lightings is proposed by calibrating the pose and light to a reference condition through an illumination invariant 3D face reconstruction. First, from single facial image, the elaborate 3D shape is recovered based on a statistical deformable model regressed through 2D geometry formed by some facial landmarks. Then with the recovered 3D shape, the texture independent on the illumination is generated by spherical harmonics ratio image and finally the illumination invariant 3D face is reconstructed completely. Our method combines the strength of statistical deformable model to describe the shape information and the compact representations of the illumination in spherical frequency space, and handle both the pose and illumination variation simultaneously. This algorithm can be used to synthesize novel views and enhance the performance of general face recognition. Some experimental results on CMU PIE database are presented.

Symmetry-Aided Frontal View Synthesis for Pose-Robust Face Recognition

“…Afterwards, taking advantage of facial symmetry, we overcome problems due to self-occlusion and synthesize a virtual frontal face by sampling texture from the original image onto the new mesh, using Thin Plate Splines-based warping. There exist similarities between our method and the works of Blanz et al [8] and Xiujuan Chai et al [4], [5] as all of them try to generate frontal images. However, and among other differences, facial symmetry will be taken into account, leading to important improvements in system performance.…”

“…Blanz et al also use the 3D Morphable Model in [8] to synthesize frontal faces from non frontal views, which are then fed into the recognition system. In this same direction, other researchers have tried to generate frontal faces from non frontal views, like the works proposed by Xiujuan Chai et al in [4], via linear regression in each of the regions in which the face is divided, and in [5] where a 3D model is used. In [9], Samaras and Zhang combine the strengths of Morphable models to capture the variability of 3D face shape and a spherical harmonic representation for the illumination.…”

Active Appearance Based Pose and Illumination Variant Face Recognition From Video Sequences

“…Chai et al proposed a method to handle both pose and lighting condition simultaneously [6]. The pose and lighting condition of each image is calibrated to a pre-set reference condition through an illumination invariant 3D face reconstruction.…”

“…The two main challenges in face recognition are illumination and pose variation [2], [6], [7]. Many methods have been developed to provide feasible solution to recognize face.…”

A Comprehensive Review of Significant Researches on Face Recognition Based on Various Conditions

“…Xiujuan Chai et al [63] have described that the Pose and illumination changes from picture to picture were two main barriers toward full automatic face recognition. In their paper, a novel method to handle both pose and lighting condition concurrently was proposed.…”