Development of a Face Recognition System and Its Intelligent Lighting Compensation Method for Dark-Field Application
read more
Citations
A cost-efficient-based cooperative allocation of mining devices and renewable resources enhancing blockchain architecture
Pig Face Recognition Based on Trapezoid Normalized Pixel Difference Feature and Trimmed Mean Attention Mechanism
Design and analysis of sinusoidally modulated substrate integrated waveguide and filter
LocalEyenet: Deep Attention framework for Localization of Eyes
Multi-Target Cross-Dataset Palmprint Recognition via Distilling From Multi-Teacher
References
Robust Face Recognition via Sparse Representation
FaceNet: A unified embedding for face recognition and clustering
Robot Vision
Emotion-Aware Connected Healthcare Big Data Towards 5G
Image-Quality-Based Adaptive Face Recognition
Related Papers (5)
Frequently Asked Questions (16)
Q2. What are the future works in this paper?
In the future, different LED layout methods and face detection or recognition techniques with better processing performance will be studied to improve the usability of proposed system.
Q3. How many vertexes of triangular patch can be selected?
When estimating the facet normal vector of LUOPs, since the 3DMM includes more than 50,000 points, some vertexes of triangular patch can be selected close to the LUOPs; then the authors can use any 3 vertexes to estimate the facet normal vector.
Q4. How can the intelligent system control be solved?
After the sequencing computations of LED radiation, ray reflection, camera response, and lighting uniformity, intelligent system control can be solved by a genetic algorithm.
Q5. What can be done to improve the processing effect of their system?
In the future, additional spatial layouts of LED units can be designed, and other machine learning methods can be used to improve the processing effect of their system.
Q6. How many facial landmarks can be used to estimate their parameters?
Sixty-eight facial landmarks can be used to estimate their parameters because they can provide the corresponding spatial point coordinates close to the face edge or the nose.
Q7. What is the simplest method of estimating the 3DMM?
In this study, the Basel face model (BFM) [25] is used as the mean face; then, the estimation of the 3DMM transforms into an iterative computation of the shape and expression factors.
Q8. What was used to assess the lighting effect between the standard and the arbitrary environment?
The DTW was employed to assess the lighting effect between the point set captured from the standard lighting environment and the set recorded from the arbitrary environment.
Q9. What is the main design concept of the rectangle shape design mode?
Considering the spatial layout constraint, the appearance design requirement, and even the government management regulation, a rectangle shape design mode was considered in this study.
Q10. What is the first step in the facial landmark extraction process?
When performing 2D facial landmark extraction, a facial landmark template is first defined; then, an ERT is used to fit the contour of the face in an image via iterative computations.
Q11. What is the effect of the shadow on the estimation of 3DMM?
The shadow will affect the estimation of 3DMM severely in dark field: if an image contrast is low, the estimations of 68 facial landmarks will be inaccurate.
Q12. What are the steps in the red dash rectangle?
In Fig. 4, contents within the red dash rectangle are their proposed steps, which can achieve intelligent lighting compensation, whereas other contents belong to the traditional processing steps of face recognition.
Q13. Why does the waitress have to stand in front of a display?
As shown in Fig. 2-(a), because the waitress requires to assess information on a computer, she has to stand in front of a display; moreover, a visitor has to stand in the left or the right side of that display such that he or she can directly face the waitress.
Q14. What is the reason why the system is not able to perform the best?
The inability of the traditional neural network and sparse representation-based methods to obtain the best performance may be attributed to the small amount of training data and the improper feature description.
Q15. What is the reason for the inability of the deep learning-based method to obtain the best?
the inability of the deep learning-based method to obtain the best result may be attributed to the small amount of training data.
Q16. How can the authors improve the accuracy of the face recognition system?
According to the experimental results, face recognition accuracy can be improved by at least 10.0%, particularly when the environment lighting is poor and complex.