Classifying Physiological States of Hair-Cap Moss Based on Digital Images

Mosses play important roles in soil erosion control and nutrient cycling in the ecosystem. The hair-cap mosses are desiccation-tolerant and have a hydrated state and a desiccated state. The physiological features and growth rates of mosses differ in different states. Monitoring the physiological states of hair-cap mosses will be helpful in predicting the growth of mosses and assessing the vegetation condition in boreal forests. The initiative of this project is to classify the physiological states of the mosses based on digital images through a machine learning approach. We take images of moss canopies in fields and run an image processing program to extract features that are important for human eyes to distinguish the two states. We use the WEKA suite of machine learning algorithms to identify the algorithm that classifies the most accurately. Ultimately, we turn the field images into binary images that indicates the location of hydrated or desiccated spots. The results are compared to manual classification to evaluate the accuracy.