Remote sensing ship detection technology based on DoG preprocessing and shape features

In order to detect ship targets in remote sensing images, we have made research on the detection in feature extraction and target detection. A method about Remote Sensing Ship Detection based on DoG Preprocessing and Shape Features was proposed. Firstly, the original image is processed by DoG filter, which made the blocks more obvious. Then, Otsu is used to segment the image after preprocessing, and a two-value result is obtained. Finally, we analyzed three types of the segmentation results are standardized including the density, the degree of the rectangle and the degree of circularity. The results of the experiment show that the proposed DoG preprocessing coupled with shape feature model can achieve 94.27% in accuracy, at the same time, the recall rate is as high as 87.70%, F1 was up to 90.87%. The research could provide important technical support for monitoring of the sea. It has important reference to the research of remote sensing target detection in the future.