Deep learning-enabled single-shot fringe projection profilometry with composite coding strategy

Abstract

In recent years, due to the rapid development of deep learning technology in computer vision, deep learning has gradually penetrated into fringe projection profilometry (FPP) to improve the efficiency of three-dimensional (3D) shape measurement and solve the problem of phase/or depth retrieval accuracy. In order to measure dynamic scenes or high-speed events, the single-shot fringe projection technique, due to its single-frame measurement property that can completely overcome the motion-induced errors of the object, becomes one of the optimal options. In this paper, we introduce a deep learning-enabled single-shot fringe projection profilometry with a composite coding strategy. By combining an FPP physical model-based network architecture with a large dataset, we demonstrate that models generated by training an improved deep convolutional neural network can directly perform high-precision phase retrieval on a single fringe image.

Robust structured light 3D imaging with two fringe patterns using recurrent classification neural network

“…The process of acquiring the absolute phase is usually implemented step-by-step in both traditional and deep learningbased frameworks. Some researchers have attempted to model the process as a regression problem (see [10,15,[19][20][21]). However, using a regression network for this task can be more challenging to train and may result in less robust performance.…”

“…However, this method was proved to be effective mainly by simulated data, and it faced challenges when applied to real-world scenarios. Another approach was presented by Li et al [20,21], who developed a deep learning-based method for single-shot 3D reconstruction using a composite fringe pattern. The composite fringe pattern provides additional information that helps alleviate ambiguity, although it may lead to reduced robustness when dealing with surfaces exhibiting texture variations.…”

“…Typically, more than three frames are required for handling complex scenes. Although these single-shot FPP-based 3D reconstruction approaches [19][20][21] have achieved considerable progress, they only work for specific scenes and have significant limitations when applied to common real-world scenes. In this study, the motivation is to develop a more concise and robust 3D imaging approach for real and complex scenes.…”

Three 1-bit speckle-embedded pulse-width modulation patterns for robust absolute 3D measurement

“…Yang et al [41] proposed a single-shot phase extraction approach based on a deep convolutional generative adversarial network. Li et al [42,43] proposed composite fringe projection deep learning profilometry. The wrapped phases of different frequency patterns are extracted by deep learning from a single composite fringe pattern to obtain the absolute phase.…”

Three-dimensional measurement based on equal spacing binary fringe coding

“…With the emergence of new technologies in human society such as machine vision [1], precision manufacturing [2], biomedicine [3], and 3D printing [4], the demand for correctly measuring the three-dimensional contours of objects is increasing. Among the existing 3D shape measurement technologies, fringe projection profilometry (FPP) has been widely used because of its flexibility, speed and accuracy [5].…”