Multi-Position Face Recognition Using Spatial Position Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current face recognition technologies for mobile devices have low recognition rates and speeds, particularly due to the need for precise positioning of the user's face, which can be circumvented by photo attacks and result in poor usability and security.
Innovation Solution
A face recognition method that captures and stores face images at multiple preset positions, allowing for recognition at any of these positions, using offset parameters to determine the correct position and improve matching data, thereby enhancing recognition probability and speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If face recognition is performed using a single specific position (frontal face), then the recognition process is simple, but the recognition rate is low and security is poor (vulnerable to photo attacks)
Solution Approach 1:
The patent transitions from single-position (2D) face recognition to multi-position (3D spatial) face recognition. By capturing face images at multiple preset positions around the user's head and incorporating spatial position data into the recognition process, the system creates a more robust authentication mechanism that is resistant to photo attacks while maintaining operational simplicity through automatic position detection.
Solution Approach 2:
The system pre-captures and stores face images at multiple preset positions during a setup phase before actual authentication is needed. When recognition is required, the system simply retrieves and compares against these pre-prepared images from the appropriate position, eliminating the need for complex real-time position adjustment and speeding up the recognition process.
2Reliability
If face images are captured at multiple preset positions, then recognition probability and security are improved, but device complexity and operation complexity increase
Solution Approach 1:
The system automatically detects the user's head position and orientation using the device's sensor suite (accelerometer, gyroscope, magnetometer) and autonomously selects which preset position image to retrieve for comparison. This self-service approach eliminates the need for manual position selection or complex coordination between multiple cameras, simplifying the system while maintaining multi-position recognition capabilities.
Solution Approach 2:
The patent uses a single camera that can capture images from multiple preset positions by moving the device or rotating the camera, rather than requiring multiple fixed cameras. The same image processing and comparison algorithms are used for all positions, creating a universal recognition system that handles multiple positions without proportionally increasing complexity.
3Measurement precision
If face images are captured at multiple preset positions, then recognition accuracy is improved, but recognition speed may decrease due to more data to process
Solution Approach 1:
Face images at all preset positions are captured and stored in advance during a setup phase. During actual authentication, the system only needs to retrieve the appropriate pre-captured image based on detected head position and perform a single comparison, rather than capturing and processing multiple images in real-time. This preliminary preparation maintains high accuracy while ensuring fast recognition speed.
Solution Approach 2:
The recognition process is segmented into two distinct phases: a setup phase where images are pre-captured at multiple positions, and an authentication phase where only a single image retrieval and comparison is performed. This segmentation allows the system to gather comprehensive multi-position data without impacting the speed of the actual recognition operation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A face recognition method, device and a computer-readable storage medium are provided. The face recognition method includes: capturing face images of a user at a plurality of preset positions respectively; storing the face images and the preset positions corresponding to the face images; determining, in face recognition, one of the preset positions where the image is to be captured, and capturing a current face image of the user at the determined preset position; searching a corresponding face image in a storage device according to the determined preset position; and determining whether the current face image is matched with the searched face image.