Face Direction Estimation via Perturbed Region Integration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for estimating the direction of a line of sight and face orientation from a single image are prone to inaccuracies due to poor imaging conditions or shielding, leading to failed estimations.
Innovation Solution
An estimation device that generates multiple extraction regions by applying perturbations to the extraction region of a partial image, estimates directions and reliability for both face and line of sight based on these regions, and integrates the directions based on their reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single extraction region is used for estimation, then the device complexity is low, but the measurement precision deteriorates under poor imaging conditions
Solution Approach 1:
The patent divides the estimation process into multiple segments by generating multiple extraction regions through perturbation. Instead of relying on a single extraction region, the system creates several candidate regions and processes each independently, then integrates the results. This segmentation approach improves measurement precision by ensuring that at least some of the extracted regions contain sufficient information for accurate estimation even when imaging conditions are poor.
Solution Approach 2:
The patent applies partial action by generating multiple extraction regions beyond the single required region. By creating more extraction regions than the minimum one needed, the system increases the probability that at least some regions will contain sufficient information for accurate estimation. The integration process then selects and combines results from these multiple regions, improving overall estimation accuracy without requiring all regions to be perfectly valid.
2Reliability
If multiple extraction regions are generated through perturbation, then the reliability of estimation improves, but the loss of time increases due to multiple estimations
Solution Approach 1:
The patent applies preliminary action by generating multiple extraction regions in advance through perturbation before the actual estimation is performed. This preliminary generation of multiple candidate regions ensures that when estimation is needed, the system already has multiple prepared regions to work with, reducing the time penalty. The perturbation-based generation is computationally efficient and prepares the data structure ahead of time for rapid estimation and integration.
Solution Approach 2:
The patent uses copying by creating multiple copies of the extraction region through perturbation. Instead of processing one original region repeatedly, the system generates multiple slightly varied copies of the extraction region. Each copy is then used for independent estimation, and the results are integrated. This copying approach distributes the computational work across multiple simpler estimations rather than requiring one complex repeated estimation, improving reliability while managing time consumption.
3Measurement precision
If feature points are extracted from the original image, then the ease of operation is high, but the measurement precision deteriorates when the image state is unsuitable
Solution Approach 1:
The patent applies dynamics by making the extraction region generation process adaptive and flexible. Instead of using a fixed extraction region based solely on original feature points, the system dynamically generates multiple perturbed extraction regions that can adapt to varying image conditions. This dynamic approach allows the estimation process to respond to different image states, improving measurement precision when the original image state is unsuitable while maintaining reasonable processing simplicity through automated perturbation.
Data Source
AI summary
An estimation device according to one aspect of the present disclosure includes: at least one memory storing a set of instructions; and at least one processor configured to execute the set of instructions to: generate a plurality of extraction regions by adding a perturbation to an extraction region of a partial image determined based on positions of feature points extracted from a face image; estimate a plurality of directions of at least one of a face and a line of sight and a reliability of each of the plurality of directions based on a plurality of partial images in the plurality of extraction regions of the face image; and calculate an integrated direction obtained by integrating the plurality of directions based on the estimated reliability.