Monitoring the Rate of Change of Vegetation Growth in Mine Rehabilitation Using Machine Learning

This research developed and evaluated methods for predicting and simulating vegetation growth in mine rehabilitation using clustering and time series techniques. The study employed the K-means clustering algorithm to partition the HSV image into seven clusters, with results evaluated through pixel shuffling and testing on additional sample images. The findings indicated that the clustering algorithm was robust and could accurately partition data in the HSV colour space. However, challenges arose when there were changes in colour, temperature or lighting conditions. A total of 248 images were tested using the clustering algorithm, and the percentage of vegetation clusters in each image was plotted on a chart to illustrate their percentages over time. The study also investigated the impact of mining activities on vegetation growth by applying four mathematical models, including linear, logarithmic, sinusoidal and overshooting, to represent the relationship between mining operations and vegetation growth.