A method for detection and classification of ships at sea in dual-polarization spaceborne SAR images

Abstract

The invention discloses a method for detecting and classifying ships at sea in a dual-polarization space-borne SAR image. The dual-polarization SAR image data is first read in in the form of a GeoTIFF file. At the same time, necessary files such as configuration parameters and masks are read. Then a decision-level dual-polarization SAR data fusion detection is performed. The detection algorithm adopts a simple, stable and reliable two-parameter CFAR algorithm, and the fusion algorithm independently judges and selects logical "or" and logical "and" according to the situation, and suppresses false alarms at the same time. Finally, the target type is judged according to the length and width scale, scattering characteristics, etc. Generally speaking, the technical solution of the present invention is to use the most efficient and reliable detection algorithm and fusion method known from the original image for operational detection, and at the same time carry out special false alarm suppression for two common special areas, and further extract After the target attribute, the target type is judged.

Description

technical field [0001] The invention relates to the field of marine remote sensing, in particular to a method for detecting and classifying ships at sea in a dual-polarization spaceborne SAR image. Background technique [0002] Synthetic Aperture Radar (SAR) is a powerful remote sensing tool, which has the characteristics of all-day and all-weather work, and is widely used in many fields such as resource detection and ocean research. Traditional single-polarization SAR adopts a fixed polarization method to transmit and receive electromagnetic waves, and can only measure a certain component of the echo scattering vector, while the propagation and scattering of electromagnetic waves are obviously described by vectors, so under the traditional technology, it is included in the scattering The information in the wave polarization characteristics will be lost, so the ability of single-polarization SAR to obtain target information is theoretically limited. Polarization synthetic a...

AI Technical Summary

Problems solved by technology

[0004] ① Unable to run commercially

Existing similar types of programs are mainly used for scientific research and teaching. Relative to operating efficiency and input and output standardization, they generally use complex algorithms with theoretical precision far greater than actual requirements because they focus on verifying a certain scientific theory, and some even use more than one algorithm. calculations, which will almost certainly lead to relative inefficiency

[0005] At the same time, in terms of input and output, because the existing programs mostly serve certain scientific research purposes, the data to be processed is some representative data, and generally there are not many subsequent uses, so the input form is relatively fixed, and the output form is relatively free , that is, the degree of standardization is low

[0006] ② High false alarm rate in low wind speed area

In some low wind speed areas, a large area shows low scattering values, but there are still some differences between these low values, so when using statistical methods, such as two-parameter CFAR method for detection, the whole Although the values ​​in the bay window are all such low values, a certain amount of relatively strong scattering blocks are still detected, resulting in non-negligible false alarms

Existing similar types of programs are mostly used for scientific research and teaching, and the test data are mostly representative but in a small range, and such phenomena are rarely encountered, so generally there is no plan to suppress such false alarms

[0007] ③The false alarm rate is high when there is strong background interference such as Enteromorpha

Existing similar types of programs or methods are mostly used for scientific research and teaching, and the goals are relatively clear, and generally do not integrate the corresponding target type judgment module

Images