Facial Image Color Correction Using Reference Color Strips
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Facial recognition algorithms do not effectively utilize color information due to unreliable color correction of facial images, particularly in identification documents and access control systems, where color casts distort skin colors and reduce recognition performance.
Innovation Solution
Incorporating a reference color strip in the facial image capture process, which is used to detect and correct color deviations through a processor, allowing for color transformation and correction within a CIE L*a*b* color space.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If color information is used in facial recognition algorithms, then recognition performance can be improved, but color casts from unknown lighting and processing cause unreliable color correction
Solution Approach 1:
A reference color strip is introduced as an intermediary element in the scene during image capture. This strip contains known reference colors that serve as a mediator to measure and correct the color cast introduced by unknown lighting conditions and subsequent printing/scanning processes, enabling reliable color correction without requiring controlled lighting environments.
Solution Approach 2:
The system captures the facial image under unknown lighting conditions and then uses the reference color strip to determine color transformation parameters. These parameters are applied to transform the color values of the facial image pixels from the captured color space to a standardized color space, effectively changing the color parameters to compensate for lighting and processing variations.
2Reliability
If a reference color strip is added to the facial image capture, then color correction reliability is improved, but device complexity increases
Solution Approach 1:
The reference color strip is a simple, inexpensive element that can be temporarily placed in the scene during capture. It does not require complex integration into the imaging device itself, and can be a separate object held by the person or placed in the background, minimizing added complexity while providing reliable color reference information.
Solution Approach 2:
The color reference function is extracted from the facial image itself and placed in a separate reference color strip. This allows the reference colors to be measured and used for correction without interfering with the facial features, and enables the reference strip to be automatically detected and processed separately from the face detection algorithms.
3Ease of manufacture
If printed and re-scanned facial images are used for identification documents, then established procedures are maintained, but color casts are compounded through multiple processing steps
Solution Approach 1:
The reference color strip is included in the original facial image capture to establish a color reference before printing and scanning occur. This preliminary color reference allows subsequent color correction to compensate for the color shifts introduced by printing and re-scanning, maintaining color accuracy through the entire document production process while using established procedures.
Solution Approach 2:
The reference color strip provides feedback information about the actual color conditions during capture and after printing/scanning. By comparing the captured reference colors with the known reference colors, the system can determine the color transformation that occurred and apply corrective transformations to the facial image, creating a closed-loop color correction process.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to an apparatus (100) for reducing a colour cast of an image of the face of a person, wherein the image of the face comprises a face of the person and a reference colour stripe. The reference colour stripe comprises a reference colour field, with a predetermined reference colour being assigned to the reference colour field. The apparatus (100) comprises a memory (101) that is embodied to store the image of the face. The apparatus (100) further comprises a processor (103) that is embodied to capture the reference colour stripe in the image of the face, capture the reference colour field in the captured reference colour stripe, extract a colour from the captured reference colour field, compare the extracted colour to the predetermined reference colour in order to obtain a colour deviation and reduce the colour cast of the image of the face on the basis of the colour deviation in order to obtain a corrected image of the face.