Rotational fusion and extended field of depth for a single cell layer in DIC microscopic images

Abstract

This work proposes a new image fusion method that is well suited for merging angle information and extending the field of depth of Differential Interference Contrast (DIC) microscopic images. Many other extend field of depth methods only regularize their solution in a small neighborhood. Such approaches are not suitable for DIC images where reliable depth estimates are only possible at cell walls, which are sparse in the overall images. Therefore, we developed a novel fusion method that first accurately estimates the depth of the cell layer of interest and in a second step extends the field of depth using this depth estimation. The proposed method only fuses information specifically linked to the epidermal cell layer of Arabidopsis leaves, thus avoiding clutter related to underlying cell layers. In addition, our method fuses images with different shear angles, thus overcoming the lack of contrast in certain directions intrinsic for DIC microscopy. By both improving the focal depth and contrast, the proposed method is a useful step towards fully automated analysis of the cellular content of the leaf's epidermal cell layer.