Optimize unresolved implementation aspects of PyWFS denoising

Optimize the number of controlled Karhunen–Loève modes, mode-dependent regularization, additional detector-noise modelling, and the PyWFS-specific denoising representation to improve noise filtering and reconstruction fidelity.

Background

The simulations show that denoising can reduce modal variance and may permit control of additional higher-order modes, but the implementation retains the same modal basis and truncation for noisy and denoised cases. The results also indicate that the optimal regularization factor varies across modes, while the current noise model omits effects such as dark current, background noise, and defective pixels. In addition, the algorithm uses a global noise estimate and a generic wavelet basis rather than spatially adaptive noise maps and representations tailored to the four-pupil geometry of pyramid wavefront sensor images. Section 4.7 identifies these issues as the principal areas requiring further optimization.