Figure 3. Effects of compressive sensing on improving image resolution.

(a) The fluorescent image captured by w-SCOPE, under 0.3X magnification. (b) Three cropped subsections of the raw image with ~20 μm resolution. (c) shows the results of compressive sensing on the cropped images. (d) displays images of the same subsections as seen through a 10 × objective lens (N.A. 0.3) acquired via a fluorescent microscope for comparison. The cells in (c,d) have the same pixel spacing, indicating that through image processing of compressive sensing, we can achieve resolving power similar to a 10 × objective lens of a conventional fluorescent microscope. It should be noted that the compressive sensing algorithm is less effective in dealing with dense objects, therefore the cell cluster in Fig. (d-2) cannot be revealed up in Fig. (c-2) from the original Fig. (b-2). (e) shows the resolution test on sub-pixel fluorescent particles (average size ~2μm). The raw images acquired under the 0.3X magnification are presented in the left 2 columns, e-1 and e-2. A 6X bi-cubic interpolation is applied to the raw image and shown in the middle column, e-3. The right column, e-4, shows the final result by compressive sensing computation. The intensity distributions are plotted along the labelled yellow lines, for indicating the resolving power under each configuration.