The extraction of seafloor substrate information based on remote sensing is the inevitable requirement of carrying out the survey of the basic habitat elements around islands and islands ecological environment protection. However, due to the influence of sun glint and water attenuation, the extraction accuracy of seafloor substrate information based on remote sensing has been poor. Therefore, the purpose of this study is to explore the extraction method of coral reefs based on remote sensing data, aiming to improve the feasibility and accuracy of detection. High resolution WorldView- 2 remote sensing images were used in this study as data source and waters around one of the Xiasha Islands were selected as study area Our research can be summarized into two part, one is to obtain bottom radiance from remote sensing data and another one is to identify different species from bottom radiance ,such as coral reefs, sand and so on. The following results have been obtained.(1) The overall accuracy of the classification results was 80.6%, and the consistency coefficient, also called Kappa coefficient, was 71.5%.(2) Among all the bottom sediment classification results, coral reefs has the highest production accuracy, which is 88.1%, mud has the lowest production accuracy, which is 59.0%, sand was the bottom sediment type whose user accuracy was the highest, which is 94.1%, and the mud has the lowest user accuracy, which is 65.1%.
|