Microplastics are defined as plastic particles less than 5mm in size. Sources of microplastics include synthetic clothing textiles, scrubbing agents in toiletries, plastic packaging, tires, etc. They can be found in various shapes such as fragments, films, fibers, and beads. Microplastics present significant challenges as their size makes it difficult to remove from the environment. There are ongoing studies regarding the detection and quantification of microplastics in the ocean. However, a lot of the methods that these studies apply are manual and very time consuming. It typically involves visual sorting, physical and chemical characterization as well as manual counting. The purpose of this study is to explore and assess the effectiveness of using scanning electron microscope (SEM) imaging, energy dispersive x-ray spectroscopy (EDS) and image processing techniques to identify and automatically count microplastics.
For the purpose of this study, wet and dry microplastic samples were analyzed. Wet microplastic samples are microplastics that were obtained from dilute water samples. Dry microplastic samples were collected directly from a source of plastic For SEM-EDS studies, a Hitachi SU 3500 SEM is used. During analysis, these non-conductive microplastic samples are left uncoated and are imaged in a 100Pa variable pressure environment with an accelerating voltage of 10kV. The SEM images are then analyzed using an image processing algorithm. The current version of the algorithm involves using Local Otsu’s method to convert an RGB image into a binary image. The foreground of the binary image is then organized, and a contour is drawn around objects of interest using an OpenCV function called ‘cv2.drawContours’. After, the algorithm circles the objects of interest, in this case the microplastics, and then counts the amount.
SEM provides images of the microplastics that highlight its surface morphology and structural details, which serve as markers for identification. EDS is not helpful in identifying the specific type of microplastic, but it can identify a microplastic from an inorganic substance. It also tends to detect a lot of elements from the substrate. The image processing algorithm was then used to analyze the SEM images of the microplastics. The resulting algorithm count outputs were skewed towards the higher side due to processing issues with larger-sized objects, object entanglement and background noise.
Medhavi Patel 