System and Method for Off Angle Three-Dimensional Face Standardization for Robust Performance

Abstract

A system uses range and Doppler velocity measurements from a lidar system and images from a video system to estimate a six degree-of-freedom trajectory of a target. The system utilizes a two-stage solution to obtain 3D standardized face representations from non-frontal face views for a statistical learning algorithm. The first stage standardizes the pose (non-frontal 3D face representation) to a frontal view and the second stage uses facial symmetry to fill in missing facial regions due to yaw face pose variations (i.e. rotation about the y-axis).

Description

CROSS-REFERENCE TO RELATED APPLICATIONS[0001]This application claims priority to U.S. Provisional Patent Application No. 61 / 699,430, which was filed on Sep. 11, 2012, and is incorporated herein by reference as if reproduced below in its entirety.FIELD OF THE INVENTION[0002]The invention is generally related to combining lidar (i.e., laser radar) measurements and video images to generate three dimensional images of targets, and more particularly, to obtaining standardized three-dimensional (“3D”) face representations.BACKGROUND OF THE INVENTION[0003]One of the purported advantages of three-dimensional (“3D”) face recognition is the ability to handle large variations in pose (e.g., orientation) of a subject. This advantage is attributed to the consistent 3D geometric structure that is exhibited between facial scans acquired from significantly different angles. However, several problems have to be addressed to achieve robust 3D face recognition performance in a statistical learning fra...

AI Technical Summary

Problems solved by technology

However, several problems have to be addressed to achieve robust 3D face recognition performance in a statistical learning framework across a variety of poses.

This process is complicated when face recognition is needed for faces acquired from significantly different views due to the following: 1) 3D facial scans, acquired using a 3D camera system from a largely non-frontal view, will exhibit missing regions or holes compared to the frontal view due to object self-occlusion; and 2) 3D face alignment, of two scans acquired from dramatically different angles, is complicated by the large angle variation and the missing regions between the two significantly different views.

Images