Evaluation of Canny Edge Detection and Muti-Threshold Segmentation Technique for Precise Mapping and Inventory of Aquaculture Ponds in Coastal Areas

摘要

The demand for a pond-based aquaculture system has increased over the past years and it resulted into uncontrolled competition of fish farmers and commercial pond owners. This system of fish production has rapidly increase the dynamic changes occurring in our coastal areas. One significant change is the conversion of mangrove areas into fishponds aquaculture system. Monitoring and Inventory of Aquaculture ponds is critical to the sustainable management and planning of coastal areas. Coastal aquaculture mapping is an integral role in providing a baseline data that can aid in decision making, regulatory laws and environmental protection towards an ecologically sustainable management of coastal areas. Thus, in this study, mapping of coastal aquaculture fishponds is aimed by implementing a methodology that maps precisely an aquaculture pond as a single semantic object. This extraction is aimed using the successful integration of Edge Detection Algorithm and Multi-threshold Segmentation techniques. This study made used of two available datasets: WorldView-2 Satellite Imagery and LiDAR Data, covering a coastal area in the Philippines, to test the efficiency of the proposed methodology. Results show that the Canny Edge Detection Algorithm in the Near Infrared Region of WorldView-2 and LiDAR-derived Data in the form of Hillshade sharpened the edges of fishponds and increased the contrast between the water and fishpond embankments. Multi-threshold Segmentation was employed whereas a threshold value was optimized until a final selected value resulted to a precise delineation of an aquaculture pond as a single semantic object. This algorithm also classified the resulting objects on their respective classes based on the defined threshold value. The techniques presented in this paper can be a framework for precise mapping and inventory of aquaculture ponds in coastal areas.