Abstract: The present disclosure provides a phase unwrapping method based on multi-view constraints of a light field and related components. The method includes: sampling main view phases at different depths in a measurement scene; performing polynomial calibration on a mapping relation from the main view phase to the pixel coordinates of the corresponding points in auxiliary view images; calculating a candidate absolute phase set of each pixel in a main view image; traversing the candidate absolute phase set, calculating an error value between the candidate absolute phase set and wrapped phases of the pixel coordinates of the corresponding points in all the auxiliary view images by utilizing a calibrated mapping relation, and taking a candidate phase corresponding to the minimum error value as an absolute phase of each pixel in the main view image. The present disclosure can stably realize accurate phase unwrapping of the structured light field.

Abstract: The present disclosure provides a multi-view single-frame phase demodulation method based on a structured light field and related components. The method includes: constructing an LFDNet neural network; collecting a data set by a structured light field system and training the LFDNet neural network to optimize the LFDNet neural network; and inputting a multi-view fringe image to be predicted into the optimized LFDNet neural network, outputting numerators and denominators of multiple views, and calculating the numerators and denominators of the multiple views via an arc tangent function and obtaining wrapped phases of multiple views.