Quantitative phase image stitching guided by reconstructed intensity images in one-shot double field of view multiplexed digital holographic microscopy

In digital holographic microscopy (DHM), achieving large field of view (FOV) imaging while maintaining high resolution is critical for quantitative phase measurements of biological cell tissues and micro-nano structures. We present a quantitative phase image stitching guided by reconstructed intensity images in one-shot double FOV multiplexed DHM. Double FOVs are recorded simultaneously through frequency division multiplexing; intensity feature pairs are accurately extracted by multi-algorithm fusion; aberrations and non-common baselines are effectively corrected by preprocessing. Experimental results show that even if phase images have coherent noise, complex aberrations, low overlap rate and large size, this method can achieve high-quality phase stitching.


Supplementary Material
. Starting from a noise-free model with a signalto-noise ratio (SNR) of 29dB, and gradually increasing the noise level, the hologram SNR gradually decreases in 5dB intervals.In these experiments, the overlap areas of the two fields of view (FOVs) are further reduced to 13% to test the limitation of the proposed method.
SNR can be expressed as where M and N are the image size, g(i, j)and f (i, j) are the original noise-free image and the noisy image, respectively.From Table S1, the higher hologram SNR, the higher the intensity image stitching quality and phase image stitching accuracy.Without using denoising algorithms (such as median filtering), our method can achieve high-quality image stitching with SNR of  S2.
From Fig. S1 and Table S2, when the overlap rate of two FOVs is set to 13%, the phase stitching results obtained by DPIS, PCHCD and SURF are severely deformed, while the phase stitching quality of the proposed method is acceptable, with SSIM above 98%, and the imaging FOV is laterally expanded by 1.87 times.

S2. Simulation experiment of two FOV stitching with minimum overlap rate
To further test the limitations of the proposed method on the overlap area and amplitude image features, we conduct two simulated stitching experiments with the overlap rate of 10% and the hologram SNR set to 28dB and 24dB respectively, as shown in Figs.S3 and S4.Table S3 lists the evaluation data of the phase stitching results obtained by the proposed method under two hologram SNRs.From Table S3 and the red rectangle in Fig. S3(c This shows that our method is unable to perform accurate affine transformation matrix calculation and image stitching on high SNR amplitude image features with the overlap rate below 10%. The above two sets of experiments (S1 and S2) prove that when the FOV overlap rate is larger than 13%, the hologram SNR is higher than 9dB, and the overlapping area has sufficient sample texture features (the number of feature matching point pairs is not less than 6), our method can achieve high-quality phase image stitching.FOVs and six FOVs obtained by the proposed method.By stitching the intensity and phase images of six FOVs using our method, the imaging FOV can be expanded by 4.5 times, and the SSIM remains above 98%.With the increase of multiple overlap images, our method is capable of further expanding the imaging FOV.

S3. Simulation experiment of four FOV stitching
GAO, LIU HUANG, * LIPING YAN, YINGTIAN LOU, AND XIAPING FU Precision Measurement Laboratory, Zhejiang Sci-Tech University, Hangzhou 310018, China *huangliu@zstu.edu.cnS1.Simulation experiment of two FOV stitching under five hologram SNRs S2.Simulation experiment of two FOV stitching with minimum overlap rate S3.Simulation experiment of four FOV stitching S4.Simulation experiment of six FOV stitching S1.Simulation experiment of two FOV stitching under five hologram SNRs To analysis the impacts of hologram noise on intensity image feature extraction and phase image stitching, we simulate the phase stitching experiments under five hologram SNRs, as shown in Table 14~29dB.The simulation results of two FOV stitching with the overlap rate of 13% and the hologram SNR of 19dB are shown in Fig. S1.The quantitative evaluation data of the phase stitching results obtained by the four methods are listed in Table Fig. S1 Simulation results of two FOV stitching with the hologram SNR of 19dB: (a1) and (a2) FOV1 and FOV2 intensity images; (b1) and (b2) feature matching point pairs extracted from (a1) and (a2); (c1) and (c2) phase stitching preprocessing results of FOV1 and FOV2; (d1)−(d4) phase stitching images obtained by the proposed method, DPIS, PCHCD, and SURF, respectively.
Fig. S4 Simulation results of two FOV stitching with the overlap rate of 10% and hologram SNR of 24dB: (a) intensity stitching image, and (b) phase stitching image.
), even though the hologram has noise-free and high SNR of 28dB, the stitched images exhibit slight distortion due to the lack of sufficient sample texture features.When the hologram SNR drops to 24dB, the linearly transformed FOV1 intensity image and phase image are completely deformed, resulting in large stitching errors, as shown in the red rectangles in Fig. S4.

Figure
Figure S5 is the simulation result of four FOV stitching with the overlap rate of 25%.The hologram SNR is 19dB (the noise mean and variance set to 1 and 4.5).

Table S2
Quantitative evaluation of two FOV phase image stitching results (overlap rate of 13% and hologram SNR of 19dB)

Table S3
Quantitative evaluation of two FOV phase stitching results (overlap rate of 10%)

Table S4
Quantitative evaluation of four and six FOV phase stitching results (hologram SNR of 19dB)Table S4 lists the evaluation data of the phase stitching results of the above four