Head Feature Measurement From a Single Image Using Facial Landmarks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for determining real-world dimensions and distances of head features require additional hardware, multiple images, or specific illumination, making them inconvenient for applications like spectacle fitting and eye examinations.
Innovation Solution
A computer-implemented method using probability distributions and machine learning techniques to estimate real dimensions and distances based on a single image, utilizing facial feature landmarks and prior knowledge from extensive data, without requiring additional hardware or multiple images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods use additional hardware objects of known size for scale reference, then measurement precision is improved, but device complexity and ease of operation deteriorate
Solution Approach 1:
The system uses facial features that are inherently present on the person's face as reference objects. The method automatically identifies multiple facial features (eyes, nose, mouth, ears) and uses their known anatomical relationships to establish scale, eliminating the need for external reference objects. The person's own face serves as the reference, making the process self-service and convenient.
Solution Approach 2:
The patent introduces probability distributions as an intermediary computational model that relates pixel dimensions to real-world dimensions. Instead of directly measuring with physical reference objects, the system uses statistical models of facial feature dimensions to mediate the conversion from image space to real space, improving both accuracy and convenience.
2Measurement precision
If conventional methods require multiple images or specific illumination, then measurement precision is improved, but productivity and ease of operation deteriorate
Solution Approach 1:
The system performs preliminary action by pre-establishing probability distributions for facial feature dimensions based on extensive training data. These pre-computed statistical models allow the system to quickly estimate real dimensions from a single image without requiring multiple captures or complex illumination setups, thus improving productivity while maintaining precision.
Solution Approach 2:
The patent changes the parameter space by working with probability distributions rather than deterministic measurements. By modeling facial feature dimensions as probabilistic quantities with known statistical properties, the system can accurately estimate real dimensions from single images without requiring multiple images or controlled illumination conditions.
3Measurement precision
If conventional methods use multiple images for scale determination, then measurement precision is improved, but loss of time and device complexity increase
Solution Approach 1:
The method uses facial features naturally present on the person as reference objects, eliminating the need for external tools or multiple image captures. The person's own facial anatomy serves as the reference framework, allowing real distance determination from a single image and reducing time loss.
Solution Approach 2:
The system performs preliminary action by pre-training probability distribution models on extensive facial data. These pre-computed statistical models enable accurate real distance estimation from single images without requiring multiple captures, thus reducing the time required while maintaining precision.
Data Source
Figure 1~2
Figure 3~4
AI summary
Computer implemented methods and devices for determining dimensions or distances of head features are provided. The method includes identifying a plurality of features in an image of a head of a person. A real dimension of at least one target feature of the plurality of features or a real distance between at least one target feature of the plurality features and a camera device used for capturing the image is estimated based on probability distributions for real dimensions of at least one feature of the plurality of features and a pixel dimension of the at least one feature of the plurality of features.