Intrinsic Image Segmentation for Face Recognition Shadow Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Varying illumination conditions in digital images can interfere with accurate face recognition tasks, causing shadows that prevent computer systems from properly identifying facial features.
Innovation Solution
The method involves using tone mapping and spatio-spectral information to segregate intrinsic images, separating material and illumination aspects, which are then processed to generate intrinsic images that enhance face recognition accuracy by removing strong shadowing effects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional face recognition methods are used on images with varying illumination conditions, then the system can process images quickly, but the accuracy of facial feature identification deteriorates due to shadows
Solution Approach 1:
The patent segments the image into two separate intrinsic images: a material image representing reflectance properties and an illumination image representing lighting conditions. This segmentation allows the system to process the material image for face recognition while using the illumination image to correct or remove shadow effects, thereby improving identification accuracy without sacrificing processing speed
Solution Approach 2:
The patent introduces an illumination image as an intermediary component that captures lighting information. This intermediary is used to compute correction factors or masks that are applied to the material image, effectively removing shadow interference while preserving the underlying facial features for accurate recognition
2Measurement precision
If illumination correction techniques are applied to remove shadows, then face recognition accuracy improves, but the complexity of the image processing system increases
Solution Approach 1:
The patent applies segmentation by dividing the image processing task into distinct components: extracting the material image and illumination image separately. This modular approach allows each component to be processed independently using optimized algorithms, managing system complexity while achieving accurate shadow removal and face recognition
Solution Approach 2:
The patent changes the parameter representation by working in a transformed space (intrinsic images) rather than the original image space. This parameter transformation simplifies the mathematical operations required for illumination correction, as the separation of material and lighting properties enables more efficient computation despite the increased processing steps
Data Source
AI summary
In an exemplary embodiment of the present invention, an automated, computerized method is provided for processing an image. According to a feature of the present invention, the method comprises the steps of providing an image file depicting an image, in a computer memory, identifying information in the image file relevant to a logical deduction regarding material and illumination aspects of an image and selected from information relevant to spatio-spectral aspects of an image and constituents of color; defining a constraint as a function of the information; and utilizing the constraint in an image segregation operation.