Face Recognition via Pupil Coordinates and Template Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is a need for an efficient system and method to automatically recognize facial images within digital images, addressing the challenge of identifying whether a face belongs to the same person across varying conditions and implementing access protection effectively.
Innovation Solution
The method involves detecting facial images, calculating pupil coordinates, converting images to black and white, identifying a face rectangle, reducing external lighting effects, and generating image templates using mathematical functions to measure similarity between templates, allowing for accurate face recognition and access control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automatic face recognition is implemented in digital images, then identification efficiency is improved, but accuracy under varying lighting and angle conditions deteriorates
Solution Approach 1:
The patent transforms the face image into a template representation by calculating mathematical functions at multiple points across the image. This parameter transformation converts pixel-based recognition into a normalized mathematical model, making the recognition invariant to lighting conditions and angles while maintaining high accuracy for automated identification.
2Measurement precision
If face recognition accuracy is improved by using detailed image analysis, then measurement precision is improved, but computational complexity increases
Solution Approach 1:
The patent divides the face recognition task into discrete steps: detecting face location, extracting facial features, calculating mathematical functions at specific points, and comparing templates. This segmentation breaks down the complex recognition process into manageable operations that can be efficiently computed while maintaining high precision through the mathematical template matching.
Data Source
AI summary
A system, method and computer program product for face recognition in digital images containing portraits or images of human faces. The facial images are detected and pupil coordinates are calculated. A facial image is converted into a black and white image. A rectangle containing the face is identified. Then, pupil coordinates are determined. A rectangle of a pre-defined size is cut out from the image so the pupils are located at pre-defined coordinates. External lighting effects are reduced and an image template is generated by calculating sets of values of different image points.