Figure 4: Regularization, using the bootstrapping method to choose the log-likelihood penalty. | Scientific Reports

Figure 4: Regularization, using the bootstrapping method to choose the log-likelihood penalty.

From: Image reconstruction from photon sparse data

Figure 4

(a) ground truth image (b) synthesised data with 50,000 photons (c) Gaussian-smoothed image with a pixel radius of 1.5 (d) reconstructed image using bootstrapping, optimizing Total Curvature with squared terms. Bootstrapping gives a better RMS residual than the naively smoothed image.

Back to article page