Extraction Study Of Aquaculture Ponds In Zhuhai Based On High-Resolution Images

With the modernization of aquaculture development,pond culture,as the most important kind of aquaculture method,increasingly needs more effective modern monitoring means.In the past,aquaculture information extraction was generally based on low-resolution or mediumresolution remote sensing images.What's more,the extraction was carried out with the whole aquaculture waters as the object,which was not able to extract single aquaculture ponds in a fine method.In this study,based on 0.5m high-resolution remote sensing images,I use Objectoriented classification methods and improved U-Net deep learning model method to finely extract aquaculture ponds.Work on this thesis is expected to provide scientific methodological support for more effective monitoring of aquaculture farming system development.Based on the 0.5m high-resolution remote sensing image of Zhuhai city,this thesis firstly uses the local variance method to determine the optimal segmentation scale of aquaculture ponds under the 0.5m resolution image to solve the problem that the optimal segmentation scale is difficult to be determined.Then,an improved U-Net deep learning model is proposed: based on the original U-Net model structure,the Dense Net model structure is fused into the coding part of U-Net to increase the extraction capability of aquaculture ponds under complex background.The following conclusions were eventually drawn.(1)In the object-oriented method,the optimal segmentation scale for extracting aquaculture ponds from 0.5m resolution images is 85,and the aquaculture ponds can be effectively extracted finely based on 0.5m resolution images;the improved U-Net model can be effectively used for the fine extraction of aquaculture ponds from 0.5m resolution remote sensing images.(2)Through comparison,it is found that the improved U-Net model method in this thesis can effectively extract aquaculture ponds finely with individual aquaculture ponds as the target,and the extracted aquaculture ponds have clear edges and strong independence.The method is more automated and applicable,and can be applied to finer extraction of aquaculture ponds in a larger range,providing scientific method support for finer extraction of aquaculture ponds.(3)The aquaculture ponds are mainly distributed in the west of Zhuhai city,with an overall "?"-shaped distribution.The district with the largest area of aquaculture ponds in Zhuhai is Doumen District,and the town(street)with the largest area of aquaculture ponds in Zhuhai is Baijiao Town,Doumen District.