A device for detecting oil spills on the sea surface using radar images that combine spatiotemporal spectrum analysis and time stability characteristics.

Abstract

This invention provides a device for detecting oil spills on the sea surface using radar images, combining spatiotemporal spectrum analysis and time stability characteristics. It acquires a sequence of radar images of the sea surface within a specific time period using a radar system. A spatiotemporal spectrum analysis module performs spatiotemporal spectrum conversion to obtain a wavenumber-frequency spectrum representation. A wave signal module identifies dynamic spectral components related to wave propagation based on the dispersion characteristics of sea surface waves in the spectrum representation, separating these dynamic spectral components from a non-wave signal. An image remodeling module inversely converts the separated spectral data back to the image domain, generating a calibrated image sequence. A time stability analysis module performs time-series statistical analysis on the echo intensity of each pixel in the calibrated image sequence within the specified time period to calculate a time stability index. An oil spill determination module identifies a target area with persistently low echo characteristics based on a comparison between the time stability index and a feature threshold, classifying it as an oil spill-covered area. This invention effectively reduces interference from wave dynamics and distance attenuation, thereby improving the accuracy and stability of oil spill detection on the sea surface.

Claims

1. An apparatus for detecting oil spills on the sea surface using radar images that combine spatiotemporal spectrum analysis and time stability characteristics, comprising: a radar system having an image acquisition module for acquiring a sequence of radar images of the sea surface within a time period; a spatiotemporal spectrum analysis module connected to the image acquisition module for receiving the sequence of radar images of the sea surface and performing a spatiotemporal spectrum conversion to obtain a corresponding wavenumber-frequency spectrum representation; a wave signal module connected to the spatiotemporal spectrum analysis module for identifying a dynamic spectral component related to wave propagation based on a dispersion characteristic of sea surface waves in the wavenumber-frequency spectrum representation, thereby separating the dynamic spectral component from a non-wave signal in the wavenumber-frequency spectrum representation; and an image remodeling module connected to the wave signal module for inversely converting the separated and processed spectral data back to the image domain to generate a calibrated image sequence. A time stability analysis module, which is connected to the image remodeling module, performs time series statistical analysis on the echo intensity of each pixel in the adjusted image sequence within the time period to calculate a time stability index; and an oil spill determination module, which is connected to the time stability analysis module, identifies a target area with a continuous low echo characteristic based on the comparison result of the time stability index and a feature threshold, and determines the target area as an oil spill covered area on the sea surface.

2. The apparatus for detecting oil spills on the sea surface using radar images combining spatiotemporal spectrum analysis and time stability characteristics as described in claim 1, wherein, This spatiotemporal spectrum analysis module performs spatiotemporal spectrum conversion through three-dimensional Fourier transform.

3. The apparatus for detecting oil spills on the sea surface using radar images combining spatiotemporal spectrum analysis and time stability characteristics as described in claim 1, wherein, The dispersion characteristics of the wave signal module are established based on a sea surface wave dispersion formula, which is expressed as shown in Equation 1 below: (Equation 1); where ω is the angular frequency, k is the wave number, g is the gravitational acceleration, and d is the water depth; the dispersion characteristics are used to define the dynamic spectral components within a specific range in the wave number-frequency spectrum representation.

4. The apparatus for detecting oil spills on the sea surface using radar images combining spatiotemporal spectrum analysis and time stability characteristics as described in claim 1, wherein, The wave signal module separates the dynamic spectrum component from the non-wave signal by using a spectrum mask or a band filter to suppress the dynamic spectrum component.

5. The apparatus for detecting oil spills on the sea surface using radar images combining spatiotemporal spectrum analysis and time stability characteristics as described in claim 1, wherein, The time stability index calculated by the time stability analysis module is defined based on the time series statistical characteristics of the echo intensity of each pixel.

6. The apparatus for detecting oil spills on the sea surface using radar images combining spatiotemporal spectrum analysis and time stability characteristics as described in claim 5, wherein, The time stability index is defined based on the combination of the time median and the absolute deviation of the median echo intensity of the pixel. When the time median is lower than an intensity threshold and the absolute deviation of the median is lower than a variation threshold, the pixel is determined to have a sustained low echo characteristic.

7. The apparatus for detecting oil spills on the sea surface using radar images combining spatiotemporal spectrum analysis and time stability characteristics as described in claim 1, wherein, The time stability index further includes the proportion of time during which the echo intensity of the pixel is lower than a preset intensity threshold value within the time period, and when the proportion of time is higher than a proportion threshold, the pixel is determined to have a continuous low echo characteristic.

8. The apparatus for detecting oil spills on the sea surface using radar images combining spatiotemporal spectrum analysis and time stability characteristics as described in claim 1, wherein, The oil spill determination module includes a classification model. The oil spill determination module inputs the time stability index into the classification model for analysis. The classification model is selected from one of the support vector machine, random forest or neural network models.

9. The device for detecting oil spills on the sea surface using radar images that combine spatiotemporal spectrum analysis and time stability characteristics as described in claim 1 further includes an output display module connected to the oil spill determination module for outputting and displaying the determination result of the oil spill coverage area on the sea surface.

10. The apparatus for detecting oil spills on the sea surface using radar images combining spatiotemporal spectrum analysis and time stability characteristics as described in claim 8, wherein, The input to this classification model must include at least one of the following: time median, median absolute deviation, low echo time proportion, area of ​​spatially connected regions, region shape features, or region boundary gradient, or a combination thereof.

Images