Gamma nonlinear error correction method based on deep learning

Abstract

The invention discloses a gamma nonlinear error correction method based on deep learning. The method comprises the following general steps of establishing a model based on a convolutional neural network,after training, obtaining a molecular denominator item for calculating a phase,and substituting the two items into an arc tangent function to calculate the phase of the object. Compared with a multi-step phase shift method, the method has the advantages that the image acquisition quantity is greatly reduced, the image acquisition time consumption is reduced, and the calculation amount is reduced; compared with mathematical transformation modes such as Fourier transformation, a large number of complex operations are avoided, the calculation cost is low, and the speed is high; compared with amethod for calibrating a gamma value, complex operations such as calibration are not needed.

Description

technical field [0001] The invention relates to the technical field of optical measurement, in particular to a gamma nonlinear error correction method based on deep learning. Background technique [0002] As a main non-contact optical measurement method, fringe profilometry is widely used in 3D modeling, engineering practice, science, education, culture, health and other fields. The main method to obtain phase in fringe profilometry is phase shift method, the most important of which is N Step phase shift method (literature "Automated phase-measuring profilometry of 3D diffuse objects", author Srinivasan V, etc.). The fringe profilometry based on the phase-shift method has main sources of errors in object phase measurement: phase shift errors, gamma nonlinear errors of projectors, light source stability, vibration errors, and quantization errors. Because the stripes are generated by software, with the development of digital raster display technology, commercial digital raste...

AI Technical Summary

Problems solved by technology

Since the gamma nonlinear error is represented as a high-order harmonic, the method of mathematical transformation mainly uses the Fourier transform (document "Afast and accurate gamma correction based on Fourier spectrum analysis for digital fringe projection profilometry", author Ma S, etc.), Hilbert transform, wavelet transform, etc., but there are problems with complex mathematical calculations and large amounts of calculations

The method of calibrating the gamma value (literature "Phase error compensation for a 3-d shape measurement system based on the phase-shifting method", author S. Zhang, etc.) does not require much calculation, but sometimes the calibrating operation of the gamma value is more complex. complex

In addition, there are phase shifting methods using high steps such as 8 steps, 12 steps or even 20 steps. This method requires a large number of images, takes a long time, and requires a lot of calculations.

Therefore, there is still a lack of a gamma correction method with a small amount of calculation, high speed and easy operation.

Images