Face Mask Identification Using Spatial and Frequency Features in Depth Image from Time-of-Flight Camera

Area of Science:

• Computer Vision
• Biomedical Engineering
• Public Health Technology

Background:

• Face masks are crucial for preventing virus transmission in high-risk areas like transit stations and hospitals.
• Accurate and rapid identification of mask-wearing conditions is essential for public health surveillance.
• Contactless recognition systems minimize human resource costs and reduce exposure risks.

Purpose of the Study:

• To develop a novel, fast, and accurate contactless method for face mask recognition.
• To utilize 3D spatial and frequency features for enhanced recognition capabilities.
• To classify both mask-wearing status and mask types efficiently.

Main Methods:

• A Time-of-Flight (ToF) camera captures depth images for 3D information.
• Facial contour extraction from depth images reduces data dimensionality for faster processing.
• A two-part classification process identifies mask presence and then mask type using spatial and frequency features.

Main Results:

• The proposed algorithm achieved a total recall accuracy of 96.21%.
• Recall accuracy for identifying individuals without masks reached 99.21%.
• The method effectively utilizes 3D spatial and frequency features for robust recognition.

Conclusions:

• The developed method offers a highly accurate and efficient solution for contactless face mask recognition.
• This technology can be deployed in various public settings to monitor mask compliance.
• The use of 3D depth data and contour analysis significantly improves recognition performance.