Lensless 3D sensing technology with Fresnel zone aperture based light-field imaging

摘要

Lensless light-field imaging with Fresnel zone aperture (FZA) is a technology that enables lensless cameras with beneficial features such as light calculation load in the image reconstruction processing, and refocusing, i.e. post-capturing focus adjustment. The ability to refocus indicates that the lensless camera s 3D information of the scene and the depth information can be extracted through comparison of contrasts among reconstructed images at different focus distances. This feature is promising since monocular 3D sensing can be achieved without using special light sources. However, since exhaustive search is needed to extract depth information, application of the contrast method is unsuitable for purposes that require real-time depth sensing. To overcome this limitation, we aimed to establish a new fast 3D sensing algorithm for lensless cameras with FZA. The proposed algorithm is based on our findings that reconstructed images of the lensless camera are complex images that have values in real and imaginary parts, and the sign of the imaginary part is reversed when the focus position of the lensless camera crosses the in-focus position. Since this "zero-crossing point" can be calculated through simple interpolations, 3D sensing can be achieved without exhaustively searching over numbers of reconstructed images with various focus positions. We have verified the effectiveness of the algorithm through simulation and experiments with developed prototypes. The algorithm enables real-time compact 3D sensors suitable for various applications, e.g. smartphones, robots, and vehicles.