Fig. 1. Workflow diagram of NC-PCA algorithm

Fig. 2. Image denoising simulation. (a) Original image; (b) image with background noise and shot noise; (c) denoised image obtained by 1--3 principal components; (d) denoised image obtained by 1--6 principal components; (e) denoised image obtained by 1--9 principal components; (f) normalized intensity curves corresponding to dotted lines in Fig.2(a)--(e); (g) PSNR and SSIM curves for denoised images with different principal component numbers; (h) CEVR for denoised im

Fig. 3. Effect of NC-PCA algorithm on images with different SNRs. (a) PSNR curves before and after NC-PCA algorithm processing; (b) E_RMS of CSSTORM localization of images

Fig. 4. Photo scintillation images before and after NC-PCA algorithm processing. (a) Original flashing image; (b) denoised image obtained by NC-PCA processing with 1--6 principal components; (c) denoised image obtained by NC-PCA processing with 1--3 principal components

Fig. 5. Comparison of super-resolution results of different algorithms for data reconstruction with different frames. (a) Reconstructed 500-frame data, scale bar: 2 μm; (b) magnified view of Fig.5(a), scale bar: 0.5 μm; (c) magnified view of Fig.5(a) with the data reconstruction using 200 frame, scale bar: 0.5 μm

Fig. 6. Photon number distribution along the dashed lines in Fig.5. (a) First column image in Fig.5(b); (b) second column image in Fig.5(b); (c) third column image in Fig.5(b)

Fig. 7. FRC spatial resolution analysis curves of methods in Fig.5(a). (a) Raw image; (b) K-factor preprocessing; (c) NC-PCA noise reduction and K-factor pre-processing

Fig. 8. Noise reduction effect of outer mitochondrial membrane. (a) Wide-field image of outer mitochondrial membrane; (b) image obtained by CSSTORM reconstruction; (c) image obtained by CSSTORM reconstruction after NC-PCA noise reduction and K-factor pretreatment; (d) magnified image of the rectangle area in Fig.8(c) obtained by CSSTORM reconstruction; (e) magnified image of the rectangle area in Fig.8(c) obtained by CSSTOR