Offshore sea area culture pond area extraction method under cooperation of spectral features and spatial convolution

Abstract

The invention provides a method for extracting an offshore area culture pond area under the cooperation of spectral features and spatial convolution. The method comprises the following steps: S1, delimiting an aquaculture coverage area in a remote sensing image under the support of a DEM (Digital Elevation Model); step S2, based on morphological and spectral characteristics, removing non-water body information in the aquaculture coverage area to obtain a water body coverage area in the aquaculture coverage area; s3, based on spatial convolution, removing non-aquaculture water body information in the water body coverage area to identify an aquaculture water body; s4, performing post-processing on an identification result in the step S3; the method can automatically extract the water body area of the culture pond in the offshore area from the remote sensing image.

Description

Extraction method of aquaculture pond area in offshore sea area under the synergy of spectral features and spatial convolution technical field The present invention relates to the technical field of remote sensing images, especially the near-shore sea culture under the coordination of spectral features and spatial convolution. Extraction method in the pond area. Background technique According to the United Nations Food and Agriculture Organization (FAO), fish currently provides approximately 17% of white matter consumption. In 2016, global fish production peaked at around 171 million tonnes, 47% of which was provided by aquaculture. (FAO, 2018). Therefore, aquaculture plays an important role in addressing global food security and nutrition problems. ocean Breeding ponds in the riparian zone are the main bearing form of aquaculture. With the rapid development of aquaculture, the scattered coastal In addition to bringing considerable economic benefits, marine aqua...

AI Technical Summary

Problems solved by technology

Visual interpretation is a classic method for extracting aquaculture areas from remote sensing images, but its operation is highly subjective, and the prior knowledge of the interpreter has a great influence on the extraction accuracy, resulting in low efficiency.

Some scholars start with spectral features and try to automatically extract aquaculture areas by using spectral classification or constructing spectral feature indexes. The index is more built for a certain area or a certain sensor, with low portability and robustness

In recent years, object-oriented extraction methods have also been widely used. Scholars usually use multi-scale segmentation, and combine shape, spectrum, texture and other information to establish feature rule sets to achieve automatic extraction. The performance and remote sensing image quantity and quality requirements are more stringent

Therefore, an efficient, reliable, and accurate extraction scheme for aquaculture ponds remains a major challenge

Images