3D Face Image Co-Registration Using Projected Landmark Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The selection of initial correspondences or landmarks in automated co-registration of two face images is time-consuming and prone to errors, affecting the accuracy and efficiency of geometric alignment in face recognition processes.
Innovation Solution
A method involving orthogonal projection of three-dimensional face images onto a two-dimensional plane for initial correspondence point identification, followed by reverse projection to determine an initial transformation matrix, and subsequent refinement using the Iterative Closest Point algorithm to enhance co-registration accuracy and automation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If automated co-registration of two face images is performed by selecting initial correspondences or landmarks, then geometric alignment can be achieved, but the process becomes time-consuming and error-prone
Solution Approach 1:
The patent applies preliminary action by pre-identifying correspondence points on a reference face image before the actual co-registration process. The system pre-processes the reference image to establish a set of correspondence points that can be quickly matched against the target image, eliminating the need for time-consuming real-time selection during co-registration. This preparatory step significantly reduces processing time while maintaining alignment accuracy.
Solution Approach 2:
The patent uses an intermediary approach by introducing a reference face image as a mediator between the target face image and the co-registration process. The correspondence points are first established on the reference image, then used to guide the alignment of the target image. This intermediary reference structure simplifies the complex direct matching problem into a two-step process: establish reference points, then match target points to reference points.
2Productivity
If automated co-registration is performed without proper initial correspondence selection, then processing speed may improve, but accuracy and reliability of alignment deteriorates
Solution Approach 1:
The system performs preliminary identification of correspondence points on a reference face image before the actual co-registration operation. By pre-establishing these reference points during an initial processing phase, the system enables rapid matching during production co-registration tasks without sacrificing precision. The pre-processed reference data serves as a template for quick, accurate alignment.
Solution Approach 2:
The patent segments the co-registration process into distinct phases: first establishing correspondence points on a reference image, then using those points to guide alignment of the target image. This segmentation allows the system to perform computationally intensive point identification once during reference preparation, then reuse those results for multiple rapid co-registration operations, improving overall productivity while maintaining precision.
Data Source
AI summary
A computer program product provides for obtaining first and second three-dimensional point clouds from first and second three-dimensional face images. The first three-dimensional point cloud is orthogonally projected onto a two-dimensional plane to form a first two-dimensional point cloud. A first set of initial correspondence points are identified in the first two-dimensional point cloud before orthogonally projecting the identified first set of initial correspondence points from the first two-dimensional point cloud back to the first three-dimensional point cloud to identify a first set of initial correspondence points in the first three-dimensional point cloud. An initial transformation matrix is determined that, when applied to the identified first set of initial correspondence points in the first three-dimensional point cloud, co-registers the first set of initial correspondence points and/or the first point cloud with a second set of initial correspondence points associated with the second three-dimensional point cloud and/or the second point cloud.