Face Region Detection Using Nose Reference Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing face region detection systems face challenges in accurately detecting eye positions when the face moves significantly, particularly when using previous eye positions or nose positions as references, leading to incomplete detection or complex processing requirements.
Innovation Solution
A face region detecting device that captures images of a person's face, detects the position of the nose or the region between eyebrows, computes the relative position of the eyelids based on past positions, and determines the eye position using the vertical and left-right boundaries of the eyelids, allowing for simple and accurate detection even when the face moves.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If eye search regions are set based on previous eye positions, then detection speed is improved, but detection accuracy deteriorates when the face moves greatly
Solution Approach 1:
The patent performs preliminary detection of nose or eyebrow positions to establish reference points before determining eye search regions. This preliminary action creates a stable foundation that remains valid even when the face moves, allowing the eye search regions to be accurately positioned without requiring complex tracking of previous eye positions.
Solution Approach 2:
The patent introduces nose or eyebrow positions as intermediary reference points that mediate between the camera frame and the eyes. These intermediaries provide a stable reference system that is less sensitive to face movement than direct eye position tracking, thereby maintaining detection accuracy while enabling efficient search region definition.
2Adaptability or versatility
If eye search regions are set using detected nose position as reference, then adaptability to face movement is improved, but processing complexity increases when multiple eye candidates are found
Solution Approach 1:
The patent extracts and utilizes only the essential geometric relationship between the nose/eyebrow and the eyes, ignoring irrelevant details. By focusing solely on the relative position and orientation derived from nose or eyebrow detection, the system achieves adaptability to face movement without the complexity of processing multiple eye candidates or performing sophisticated pattern matching.
3Ease of operation
If only vertical direction distance between nose and upper eyelid is investigated, then measurement simplicity is improved, but detection accuracy deteriorates when the face moves left or right
Solution Approach 1:
The patent transitions from one-dimensional vertical measurement to two-dimensional relative position measurement by incorporating both vertical and horizontal coordinates. This dimensional expansion allows the system to maintain measurement simplicity while achieving accurate eye position detection even when the face moves left or right, as the relative position relationship captures movement in both directions.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An image capturing section captures a face of an observed person. A position detecting section detects, from a face image expressing the face captured by the image capturing section, one of a position expressing a characteristic of a nose and a position expressing a characteristic of a region between eyebrows. A computing section computes a past relative position of a predetermined region of the face that is based on the position detected in the past by the position detecting section. A region position detecting section detects a position of the predetermined region on the basis of the past relative position computed by the computing section and the current position detected by the position detecting section.