Sea ice detection method based on collaborative active learning and direct-push type support vector machine

Abstract

The invention discloses a sea ice detection method based on collaborative active learning and a direct-push type support vector machine. The invention relates to the field of remote-sensing sea ice detection. The defects that time is consumed for sea ice remote sensing image labeling, and a large number of label-free samples in the image contain rich information but are difficult to fully utilizeare overcome. Key points of technical scheme of the method are as follows: reliable label samples can be searched through cooperative active learning and a direct-push type support vector machine; a classification model established using rich information contained in unlabeled samples, and redundancy of samples is reduced.According to the sea ice detection method based on collaborative active learning and the direct-push type support vector machine, the performance of the classification model can be improved, and the workload of manual marking can be reduced.

Description

technical field [0001] The invention relates to remote sensing sea ice detection, in particular to a sea ice detection method of cooperative active learning and direct push type support vector machine. Background technique [0002] Sea ice has become one of the most prominent marine hazards in high latitudes. For the research on disaster prevention and mitigation and disaster assessment of sea ice disasters, it is necessary to obtain detailed information such as the outer edge line and category distribution of sea ice in a timely manner. Traditional visual and instrumental sea ice detection methods are difficult to effectively detect and obtain rich and detailed information such as the density and category distribution of large-scale sea ice. Remote sensing technology can provide all-weather, large-area, and accurate sea ice remote sensing image information, and has been widely used in sea ice detection, becoming an efficient method for sea ice detection. Existing sea ice ...

AI Technical Summary

Problems solved by technology

[0003] 1) Judging from the characteristic criteria used in the classification, the threshold segmentation method mainly determines the spectral parameters (such as band difference and ratio) that are easy to distinguish between sea ice and seawater, and distinguishes the edge line or calculation of sea ice and seawater by threshold segmentation. Sea ice density, the accuracy of this method is low, and it is difficult to obtain more detailed sea ice category information

[0004] 2) From the perspective of training samples, the research on unsupervised classification and supervised classification methods mainly focuses on selecting several different band combinations, and adopts unsupervised or supervised classification methods to classify sea ice. Unsupervised classification methods do not need to first The method is easy to implement based on empirical knowledge, but the classification accuracy is often lacking; the supervised classification strategy can achieve better results, but requires a certain number of labeled samples, otherwise it will easily lead to the Hush phenomenon, especially for high-dimensional feature data

Images