Noise model and suppression algorithm for high-precision wavefront detection

Wavefront aberration is a crucial metric for evaluating the imaging quality of an optical system and enhancing the accuracy of wavefront detection is of significant importance. Noise is a critical factor that affects detection accuracy. Simulating and suppressing noise can help explore the theoretical limit of wavefront detection and improve the actual measurement accuracy. We develop a comprehensive noise model where the input is a simulated, noise-free image in units of photons, and the output is a noisy digital signal. The model considers external disturbance noise, speckle noise, and camera noise. Speckle noise is selectively added based on the light source’s coherence. Camera noise is modeled using real camera parameters and includes photon shot noise, dark shot noise, readout noise, and quantization noise. Additionally, a noise suppression algorithm based on frame averaging is designed. We introduce the concept of a noise suppression factor, calculate this factor based on the noise characteristics and system properties, and apply it to the frame-averaged noisy image on a pixel-by-pixel basis, achieving effective noise reduction. Using the established noise model, we calculate the theoretical peak-to-valley (PV) and root mean square (RMS) limit determined by noise for two typical high-precision wavefront aberration detection systems: the Ronchi lateral shearing interferometry (LSI) system and the phase-diverse phase retrieval (PDPR) system. With our proposed noise suppression algorithm, the theoretical RMS limit can be reduced to 10% of the previous value, demonstrating its effectiveness in noise suppression. Our model provides a definitive standard for the theoretical accuracy limit of optical metrology, guiding the selection of hardware and the design of wavefront detection algorithms for subsequent research.