Time-reduction imaging method for scanning-probe microscopy using a compressed sensing algorithm based on sequential reconstruction method

Abstract

The surface image measurement time was reduced by data completion using a compressed sensing algorithm in the image-acquisition of scanning probe microscopy. In particular, by reducing the number of acquisition points of the surface-roughness image and reconstructing the image using estimated values, acquiring the surface image at one-eighth of the data-measurement time became possible. The reduced measurement time was shown to reduce the influence of thermal drift during image capture. The real time surface image demonstrated that reconstruction is possible during surface image acquisition.

Feature correlation method for image reconstruction evaluation in under-sampled scanning probe microscopy

“…3(a), right]. This under-sampling and reconstruction were performed using a previously developed algorithm, 31) which involves measuring one line and skipping three lines repeatedly. The evaluation used the area within the red square in Fig.…”

“…For experimentally undersampled images, we employed a three-line skip approach. 31) In this method, data is collected in the fast scan direction, skipped by three lines in the slow scan direction with no data acquisition, and repeated to obtain an under-sampled image. The same algorithm was applied to create an under-sampled image by compositing from normal mode.…”

“…Both under-sampled images were completed using the algorithm from the three-line skip method. 31) The under-sampled scanning tunneling microscopy (STM) experiments were conducted using a home-built SPM system, operated in ultra-high vacuum and at room temperature, developed in LabVIEW and Python with PXIe-7857R. The measurement data were saved in a Python-readable format.…”

“…To simulate thermal drift, an under-sampled SPM image was generated from a single measured SPM image [Fig.3(a), left and center] and then reconstructed [Fig.3(a), right]. This under-sampling and reconstruction were performed using a previously developed algorithm,31) which involves measuring one line and skipping three lines repeatedly. The evaluation used the area within the red square in Fig.3(a) as the reference image.…”