Generic saturation-induced phase-error correction algorithm for phase-measuring profilometry

Abstract

Intensity saturation causes partial incorrect intensities in captured images, leading to obvious phase errors in high dynamic range (HDR) phase measuring profilometry. Most of the existing methods require numerous projected patterns or additional hardware equipment to retrieve the 3D shape. This paper proposes a comprehensive saturation-induced phase error correction method by combining the average-phase compensation method applying the four-step phase-shifted(PS) patterns with phase repairment employing a total variation minimization(TVM) model. We first analyzed the periodic characteristic of saturation-induced phase error. The phase error can be efficiently compensated by averaging the initial phase and the auxiliary phase, which was calculated utilizing a set of PS patterns with an initial phase offset of /4. Furthermore, a judgment condition was provided to detect invalid points in overexposed shiny areas whose initial calculated phases were wrong. The corrected phases were repaired utilizing the TVM model from the compensated phase information around the invalid points. Simulations and experiments demonstrated that the proposed method could simultaneously correct the phase for non-uniform high-reflectivity scenes and shiny areas on the surface with high accuracy and relatively few images. The phase error is reduced to nearly 80%.

Comparative study of variant shifting-phase coding method based on noise error model

“…Among the different FPP techniques, phase shifting profilometry (PSP) stands out as an important and extensively utilized method, offering merits such as high-accuracy, high-resolution and pixel-independently calculation [6]. In the past several decades, researchers have presented numerous phase shifting algorithms to address diverse measurement challenges, including discontinuous or isolated objects [7]; as well as to address intensity saturation [8][9][10]. Wang and Yang [11] developed a triple N-step phase shifting approach to minimize the nonliear errors of images cuaused by the nonlinear response of measurement system.…”

PSNet: A Deep Learning Model-Based Single-Shot Digital Phase-Shifting Algorithm

“…In optical interferometry [1] as well as in 3D reconstruction using fringe projection profilometry (FPP) [2,3], phase retrieval is a common but fundamental step that often consists of two stages: wrapped or relative phase retrieval and phase unwrapping. While extensive efforts have been made to reduce phase errors induced via gamma nonlinear error elimination [4], color cross-talk [5], highlights [6], interreflection [7] and motion [8], which leads to improved phase accuracy, the efficiency is still unsatisfactory for real-time applications due to the projection of multiple fringe patterns with different frequencies and phase shifts [9]. For instance, during relative phase retrieval with a standard N-step phaseshifting (PS) algorithm, at least N ≥ 3 fringe patterns with different phase shifts are needed, which limits the algorithm to a relatively narrow range of applications.…”