Three-dimensional image authentication from multi-view images.

Area of Science:

• Information Security
• Computer Vision
• Optical Engineering

Background:

• Modern information security relies heavily on 3D optical authentication.
• Current methods using integral imaging require complex calibration and high transmission overhead.
• There is a need for streamlined 3D information acquisition and authentication.

Purpose of the Study:

• To introduce a novel 3D optical authentication approach.
• To simplify the acquisition and transmission of 3D information for authentication.
• To enhance the efficiency and accuracy of 3D optical authentication.

Main Methods:

• Constructing 3D data from multi-view images.
• Generating fixed-viewpoint elemental images to simplify 3D projection.
• Utilizing compressed sensing for data compression during transmission.
• Employing a deep learning network for enhanced 3D reconstruction.

Main Results:

• The proposed method effectively streamlines 3D projection and data acquisition.
• Elimination of additional viewpoint information reduces transmission overhead.
• Compressed sensing and deep learning enhance 3D data recovery and authentication efficiency.
• Experimental results validate the approach across diverse datasets.

Conclusions:

• The novel multi-view image-based 3D optical authentication method is efficient and effective.
• This approach offers a simplified and high-performance alternative to existing integral imaging techniques.
• The method shows promise for enhancing modern information security through robust 3D authentication.