Image Averaging for Digital Holography Simulation

In this experiment, we employ image averaging techniques to enhance image quality through pixel intensity normalization. The implementation typically involves calculating the mean intensity values across multiple image frames or regional kernels using functions similar to MATLAB's mean2() or OpenCV's blur() with a specified kernel size. The averaged image then serves as the input for digital holography simulation, where we model wavefront propagation using angular spectrum methods or Fresnel diffraction algorithms. To evaluate different reconstruction approaches, we implement two distinct hologram reconstruction methods: the frequency-domain phase retrieval method utilizing Fourier transforms (fft2/ifft2 operations) and the spatial-domain constrained iterative method employing Gerchberg-Saxton algorithms. Each method demonstrates unique characteristics in reconstruction accuracy, computational efficiency, and noise resilience. We conduct detailed comparisons of their performance metrics including peak signal-to-noise ratio (PSNR) calculations and structural similarity index (SSIM) measurements. Through this experiment, we gain deeper insights into digital holography techniques and develop methodologies for selecting optimal reconstruction approaches based on specific application requirements.