Low-Computation Face Recognition Using High-Information Regions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional face recognition systems for consumer portable devices are either computationally expensive and power-intensive or ineffective in varying lighting conditions and biometric distortions, making them unsuitable for outdoor use and varying user orientations.
Innovation Solution
A low-computation face recognition method that focuses on high-information portions of the face, such as eyes and mouth, using an orange-distance filter to preprocess images, normalize face features, and calculate weighted differences for matching, thereby reducing the impact of lighting and biometric distortions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional robust face recognition systems are used, then reliability is improved, but use of energy increases and device complexity increases
Solution Approach 1:
The patent divides the face image into multiple regions of interest (eyes, mouth, nose) and processes only these high-information portions rather than the entire face image. This segmentation reduces the computational load and energy consumption while maintaining recognition reliability by focusing on discriminative features.
Solution Approach 2:
The patent applies different processing strategies to different regions of the face. High-information regions like eyes and mouth receive specialized processing with higher weight in the matching algorithm, while other regions are processed differently or given lower weight. This local quality approach optimizes energy usage by concentrating computational resources where they provide the most value.
2Reliability
If traditional robust face recognition systems are used, then reliability is improved, but device complexity increases
Solution Approach 1:
The system segments the face recognition task into distinct steps: detecting high-information regions, extracting features from these regions, normalizing based on region positions, and performing weighted matching. This segmentation simplifies the overall system architecture compared to traditional holistic approaches while maintaining reliability.
Solution Approach 2:
The patent changes the parameter space by focusing on specific facial regions and their geometric relationships rather than processing the entire face image at full resolution. This parameter transformation reduces computational complexity while preserving the reliability needed for accurate recognition.
3Use of energy by moving object
If specialized security-type face recognition systems are used, then use of energy is reduced, but adaptability worsens
Solution Approach 1:
The patent performs preliminary detection and identification of high-information regions (eyes, mouth, nose) before performing the main recognition task. This preliminary action includes detecting skin-tone regions and identifying facial features, which prepares the system to adapt to varying lighting conditions by focusing on regions less affected by illumination changes.
Solution Approach 2:
By focusing on specific high-information regions like eyes and mouth that have consistent appearance across different lighting conditions, the system achieves adaptability to varying environments. These local regions provide stable features for recognition regardless of overall lighting changes, enabling the low-power system to maintain versatility.
4Measurement precision
If traditional face recognition is used with distorted features, then measurement precision is maintained, but reliability worsens
Solution Approach 1:
The system performs preliminary normalization of facial features based on the detected positions of high-information regions before performing recognition matching. This preliminary normalization corrects for distortions caused by varying distances and camera tilts, ensuring that measurements are made on normalized features, thereby maintaining precision while improving reliability.
Solution Approach 2:
The patent transforms the coordinate system and scaling parameters based on the detected positions of eyes, mouth, and nose. By changing these parameters dynamically according to the detected feature positions, the system maintains measurement precision even when the face is captured at different distances or angles, thereby improving recognition reliability.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, are disclosed for reducing the impact of lighting conditions and biometric distortions, while providing a low-computation solution for reasonably effective (low threshold) face recognition. In one aspect, the methods include processing a captured image of a face of a user seeking to access a resource by conforming a subset of the captured face image to a reference model. The reference model corresponds to a high information portion of human faces. The methods further include comparing the processed captured image to at least one target profile corresponding to a user associated with the resource, and selectively recognizing the user seeking access to the resource based on a result of said comparing.