Image Analysis for Vision-Based Skimming

Infestation with invasive aquatic weeds is a major problem being faced while maintaining inland water bodies. Cleaning the surface pollution ideally requires a fully autonomous trash skimmer with capabilities for detecting the presence and finding the physical location of nearby free- floating weed clusters, collection of weed clusters using cutting of matted weeds, if required, moving the skimmer close to the cluster, collecting the cluster, and moving the collected weeds to the bank. Vision enabled trash skimmer consists of a trash skimmer connected to a vision sensor with capability to capture individual video frames and analyze them to extract desired information. Under these conditions, providing visual information of the surroundings to the trash skimmer will make its operations more convenient, accurate and improve its overall performance. This book focuses on improving the performance of the image processing steps namely horizon line detection techniques for segmenting water regions, object detection algorithm for segmenting floating objects on the water surface, color based edge detection algorithm to estimate green-colored weeds in inland water bodies.