A novel approach for quantifying particulate matter distribution on leaf surface by combining SEM and object-based image analysis

Most methods for assessing the loading of particles on plant leaf surfaces involve a cumbersome manual step, and hence are slow to employ. Furthermore, they yield results that are summative, representing total number of particles or total volume or weight of particles in standardized size fractions. Here, we present a novel approach that cannot only accurately quantify the number of particles, but also their size and shape. In addition, the method we present replaces the manual measurement of the particles on leaf surfaces with an automated step. We applied the well-developed object-based image analysis technique to scanning electronic microscope (SEM) micrographs of tree leaf, and tested this approach for replicate SEM micrographs of a common urban tree species. We demonstrated that: 1) this new method automatically identifies the number of particles, as well as their size and shape, in contrast to the commonly used microscopic inspection approach that can only measure the number of particles; and 2) this method achieved similar overall accuracy to that of microscopic inspection (92.17% versus 95.53%), but microscopic inspection takes fourteen times longer. It is expected that the difference in efficiency would be more significant with the increase of micrograph numbers, because micrographs can be batch processed with object-based classification. With the greatly increased efficiency and the ability of the proposed method to capture new variables about particle shape and complexity, this method can facilitate comparative research on the adsorption capacities of different plant species, and potentially identifying the source apportionment of particulate matters based on their morphological characteristics, which may provide insights for species selection for pollutant reduction.