Satellite remote sensing extraction of aquaculture areas

Aquaculture (or aquaculture) is a type of aquaculture that utilizes natural water surfaces or artificial ponds to stock seedlings of high-value fish, shellfish, crustaceans, and algae, providing them with feed, controlling diseases, and promoting their rapid growth through artificial reproduction—a planned production operation. my country is the world's largest aquaculture nation and the only country in the world whose aquaculture production exceeds its catch, and the scale of aquaculture continues to grow rapidly.

Remote sensing has advantages such as wide detection range, fast data acquisition speed, short cycle, high timeliness, low cost, and high economic benefits. Remote sensing images can be used to quickly extract thematic information for aquaculture, helping farms select sites, determine aquaculture species, and monitor stocking density and water pollution.

△ Remote sensing monitoring of aquaculture area

The main object-oriented information extraction method for aquaculture areas first involves segmenting the image into several segments, i.e., objects, based on the spectral characteristics of the segments. Information about aquaculture areas is then extracted based on these segments.

First, initialize seed points (cluster centers): evenly distribute seed points within the image according to the set number of superpixels, i.e., the number of segmentation volumes; reselect seed points in the neighborhood of the already assigned seed points and calculate the gradient values ​​of all pixels in the neighborhood; reassign the pixel with the smallest gradient as a seed point; assign a class label to each pixel in the neighborhood around each seed point; for each searched pixel, calculate its distance to the seed point and iteratively optimize.

△ Schematic diagram of image segmentation

After the segmented regions are generated, remote sensing indices such as vegetation index, water index, and reflectance value of each band are calculated for each segmented region. Based on the characteristics of the aquaculture area, such as NDVI, NDWI, and band reflectance, corresponding thresholds are set. Segments that meet the threshold conditions are assigned a value of 1, and those that do not are assigned a value of 0. Finally, the regions marked as 1 are vectorized to obtain the extracted aquaculture area range.

The aquaculture area obtained by inversion based on this method is shown in the figure below.

△ Results extracted from aquaculture area