A method for identifying river and lake ships by optical remote sensing for different water color types

Abstract

The present application relates to a kind of river and lake ship optical remote sensing identification method for different water color type, comprising: obtaining the remote sensing surface reflectivity data of river and lake water area;To the remote sensing surface reflectivity data: extract brightness component TCB by the mink coat transformation;Calculate the reflectivity ratio NIR / SWIR of near-infrared band and mid-infrared band;The sliding window is set to river and lake water area near-infrared band image, and the standard deviation of the center pixel of each sliding window is calculated;After Z-Score standardization processing is carried out to the TCB, NIR / SWIR, standard deviation of each pixel, the sum is calculated as pixel ship index;In deep color, shallow color water body area in river and lake water area, pixel ship index is used to extract ship pixel in combination with threshold segmentation method respectively;The ship pixel of deep color, shallow color water body area is fused, and the ship pixel of river and lake water area is obtained.The present application method can be suitable for inland river and lake ship remote sensing identification under different water color and optical environment conditions.

Description

Technical Field

[0001] This invention belongs to the field of environmental remote sensing technology, specifically relating to an optical remote sensing identification method for river and lake vessels with different water color types. Background Technology

[0002] Ships are among the most important carriers of human activity in rivers and lakes, and their spatial distribution and activity status directly reflect shipping intensity, resource utilization levels, and the degree of human disturbance. Remote sensing identification of river and lake vessels helps to accurately distinguish between water body pixels and non-water body pixels in water remote sensing monitoring, avoiding interference from ships in the inversion of water color and water quality parameters, thereby improving the reliability and accuracy of water environment remote sensing monitoring results. At the management and policy levels, ship remote sensing identification technology can provide important support for inland waterway shipping supervision and the monitoring of illegal sand mining and illegal sewage discharge.

[0003] In recent years, with the rapid development of high-resolution optical remote sensing and synthetic aperture radar technologies, ship remote sensing identification methods have been continuously developed and improved. Existing research largely focuses on marine ship identification, addressing the relatively uniform background and large ship size in open seas, resulting in various identification methods based on threshold segmentation, traditional machine learning, and deep learning. In the field of optical remote sensing, deep learning methods such as convolutional neural networks have been widely applied to the automatic extraction of marine targets. However, compared to marine scenarios, research on ship remote sensing identification in inland river and lake waters is significantly insufficient. Existing inland waterway ship identification methods mostly rely on land pixel masking technology, indirectly eliminating land areas through water body extraction results before target detection within the water area. However, due to the complex optical types of inland water bodies, significantly affected by factors such as suspended solids concentration, algal blooms, biological activity, and shoreline shadows, water body extraction itself has considerable uncertainty, thus affecting ship identification accuracy. Furthermore, the small size, diverse types, and dense berthing of inland waterway vessels easily lead to confusion with background targets such as bridges and docks, resulting in prominent issues of missed and false detections in complex river and lake environments, making it difficult to meet the demands of high-precision and automated applications.

[0004] In summary, existing ship remote sensing identification technologies still suffer from problems such as strong background interference and insufficient identification accuracy in complex inland river and lake scenarios. Conducting research on ship remote sensing identification tailored to the characteristics of river and lake environments can not only improve the accuracy of water body remote sensing monitoring but also provide important technical support for inland waterway management, ecological protection, and law enforcement supervision, demonstrating significant application value and promotional significance. Summary of the Invention

[0005] The primary objective of this invention is to provide an optical remote sensing identification method for river and lake vessels with different water color types. This method is based on optical remote sensing images, constructs the River and Lake Vessel Identification Optical Index (RSVI), and identifies vessels with different water body optical types.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] An optical remote sensing identification method for river and lake vessels with different water color types, the method comprising:

[0008] Obtain remote sensing surface reflectance data of rivers and lakes;

[0009] The tasseled cap transformation is performed on the remotely sensed surface reflectance data to extract the luminance component TCB;

[0010] The reflectance ratio NIR / SWIR in the near-infrared band and mid-infrared band is calculated from the remotely sensed surface reflectance data.

[0011] A sliding window was set for near-infrared images of rivers and lakes, and the standard deviation of the center pixel of each sliding window was calculated.

[0012] The TCB, NIR / SWIR, and standard deviation of each pixel are standardized and then summed to form the pixel ship index.

[0013] Ship pixels were extracted from dark and light water areas in rivers and lakes using a combination of pixel ship index and threshold segmentation method.

[0014] Ship pixels are obtained by merging dark and light water areas.

[0015] In some embodiments of the present invention, remote sensing surface reflectance data covering river and lake water areas are acquired, water body regions are extracted using water body correlation spectral indices, and a water body mask is generated. River and lake water areas are then extracted based on the water body mask. The water body correlation spectral indices are preferably modified normalized difference water indices (MNDWI).

[0016] In some embodiments of the present invention, ship pixels are extracted from dark water areas using OTSU threshold segmentation combined with the pixel ship index.

[0017] For light-colored water areas, ship pixels are extracted using a preset fixed threshold combined with the pixel ship index.

[0018] In some embodiments of the present invention, the extraction of ship pixels from dark water regions and light water regions in rivers and lakes using a pixel ship index combined with a threshold segmentation method includes:

[0019] Threshold segmentation of river and lake waters is performed using the OTSU method to obtain candidate ship pixels, and the candidate regions are labeled with connected components.

[0020] Extract connected patches whose connected region area exceeds a preset threshold, and mark the remaining connected regions as ship pixels in dark water areas;

[0021] The extracted connected patches are marked as light-colored water areas, and ship pixels in the light-colored water areas are extracted using a pixel ship index combined with a threshold segmentation method.

[0022] In some embodiments of the present invention, the TCB, NIR / SWIR, and standard deviation of each pixel are Z-score standardized and then summed.

[0023] In some embodiments of the present invention, the method further includes performing consistency processing on the fused ship pixels to obtain the final ship pixels.

[0024] In some embodiments of the present invention, the pixel consistency processing includes:

[0025] Connectivity analysis is performed on the fusion results to remove connected patches whose number of pixels is outside the preset range;

[0026] Morphological post-processing is performed on the connected pixels to obtain the final set of ship pixels.

[0027] The present invention further provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above method.

[0028] The present invention further provides a computer-readable storage medium having a computer program / instructions stored thereon, which, when executed by a processor, implement the steps of the above-described method.

[0029] The present invention further provides a computer program product, including a computer program / instruction that, when executed by a processor, implements the steps of the above-described method.

[0030] This invention proposes an optical remote sensing identification method for river and lake vessels with different water color types. This method enables rapid extraction and reliable differentiation of vessel pixels under complex inland water background interference conditions, thereby improving the stability and accuracy of vessel target identification. This method helps improve the accuracy of water body information extraction in water remote sensing monitoring and provides crucial data support for inland waterway management, ecological protection, and law enforcement supervision.

[0031] It should be understood that all combinations of the foregoing concepts and the additional concepts described in more detail below may be considered part of the inventive subject matter of this disclosure, provided that such concepts do not contradict each other. Furthermore, all combinations of the claimed subject matter are considered part of the inventive subject matter of this disclosure.

[0032] The foregoing and other aspects, embodiments, and features of the teachings of the present invention will be more fully understood from the following description in conjunction with the accompanying drawings. Other additional aspects of the invention, such as features and / or beneficial effects of exemplary embodiments, will become apparent from the following description or may be learned through practice of specific embodiments according to the teachings of the present invention. Attached Figure Description

[0033] The accompanying drawings are not intended to be drawn to scale. In the drawings, each identical or nearly identical component shown in the various figures may be denoted by the same reference numeral. For clarity, not every component is labeled in each figure. Embodiments of various aspects of the invention will now be described by way of example and with reference to the accompanying drawings, wherein:

[0034] Figure 1 This is the overall flowchart of the method of the present invention.

[0035] Figure 2 This is an example of a true-color remote sensing image of the middle reaches of River A (November 13, 2021).

[0036] Figure 3 This is a map showing the extracted water body area of ​​the main channel of river A.

[0037] Figure 4 This is a diagram showing the calculation results of the ship's multidimensional optical characteristic indicators.

[0038] Figure 5 Spatial distribution results of the ship index RSVI.

[0039] Figure 6 This is a map showing the results of dividing water bodies into different color types (dark water bodies and light water bodies).

[0040] Figure 7 This is a graph showing the results of ship pixel recognition in the dark water area.

[0041] Figure 8 This is a graph showing the results of ship pixel recognition in a light-colored water area.

[0042] Figure 9 This is the final ship detection result image after fusing the ship identification results for deep and shallow water bodies.

[0043] Figure 10 This is a distribution map of the sample points for accuracy verification; in the map, blue circles represent non-ship sample points, and red triangles represent ship sample points.

[0044] Figure 11 This is a comparison chart of RSVI, MNDWI, and NDWI ship identification results, where the blue area represents the identified ship pixels.

[0045] In the aforementioned Figures 1-11, the coordinates, symbols, or other representations expressed in English are all well-known in the field and will not be elaborated upon in this example. Detailed Implementation

[0046] To better understand the technical content of the present invention, specific embodiments are described below in conjunction with the accompanying drawings.

[0047] Various aspects of the invention are described in this disclosure with reference to the accompanying drawings, in which numerous illustrative embodiments are shown. The embodiments of this disclosure are not necessarily intended to encompass all aspects of the invention. It should be understood that the various concepts and embodiments described above, as well as those described in more detail below, can be implemented in any of many ways, as the concepts and embodiments disclosed herein are not limited to any particular implementation. Furthermore, some aspects of the invention disclosed may be used alone or in any suitable combination with other aspects of the invention disclosed.

[0048] Example 1

[0049] This embodiment takes the middle reaches of river A as an example to further describe the technical method of the present invention.

[0050] The flowchart of the optical remote sensing identification method for river and lake vessels with different water color types shown in the embodiment is as follows: Figure 1 As shown, the image data source used in this embodiment is Sentinel-2 optical remote sensing imagery, and the specific steps include the following:

[0051] Step 1: Obtain Sentinel-2 MSI surface reflectance products, perform cloud removal and invalid pixel elimination, then use MNDWI to extract water areas and generate a Water Mask, as detailed below:

[0052] First, Sentinel-2 MSIL2A surface reflectance product imagery data were obtained from the European Space Agency website (https: / / dataspace.copernicus.eu / ). Images with cloud cover less than 5% were selected within the study area, and the acquisition date and product number of the selected images were recorded. True-color composites were then performed on the selected images to conduct preliminary interpretations of the spatial distribution of river and lake bodies, channel width, water color differences, and visibility characteristics of ship targets within the study area. Figure 2 Subsequently, cloud and cirrus cloud masking was performed based on the Sentinel-2 image quality band QA60 to remove pixels contaminated by clouds, obtaining basic remote sensing image data for subsequent water body extraction and ship identification.

[0053] After obtaining the effective image data after cloud removal, the Modified Normalized Difference Water Index (MNDWI) is calculated, and a fixed threshold of 0.25 is used to identify river water bodies. The calculation formula is as follows:

[0054]

[0055] In the formula, and These are the reflectance values ​​for the green light band and the near-infrared band, respectively.

[0056] At this point, some non-water pixels in the image are removed, and the extracted water area contains holes, fragmented patches, and irregular edges, requiring further optimization. This includes filling holes, removing large areas of non-water (such as river islands and bare land), and removing small connected water areas. The final result is a water mask used for subsequent ship identification, which is then used to crop the main river body area. Figure 3 ).

[0057] Step 2: Construct multidimensional optical characteristic variables for river and lake vessels. These characteristic variables include three indices: Brightness, NIR / SWIR ratio, and Local_STD. Figure 4 The features are then standardized to calculate their corresponding Z-scores, as follows:

[0058] (1) Brightness: The tasseled cap transformation is performed using MSI L2A surface reflectance (SR). The SR of each band is linearly weighted and summed according to the sensor coefficients. The first component brightness (TCB) is taken to characterize the overall reflectance brightness of the scene.

[0059] (2) NIR / SWIR ratio: The ratio of reflectance in the near-infrared band (842nm) to that in the mid-infrared band (1610nm).

[0060] (3) Local Standard Deviation (Local_STD): Local texture anomaly detection is performed using the near-infrared band (842nm). A 3×3 window is slid across the image to calculate the local standard deviation of each central pixel. The water surface is usually smooth. The brightness is lower, and sudden changes in brightness caused by ships / wake waves, etc. Increase.

[0061] The three feature indicators mentioned above are standardized using Z-scores, and the calculation method is as follows:

[0062]

[0063] In the formula, It is a variable. It is the mean. It is the standard deviation.

[0064] The ship index RSVI is obtained by summing the Z scores of each feature.

[0065] Step 3: Use the OTSU threshold segmentation method to identify ship pixels (D) in dark-colored water bodies, and use the threshold method to identify ship pixels (L) in light-colored water bodies, as detailed below:

[0066] Calculate the RSVI of the waters in the study area to obtain its spatial distribution characteristics (e.g. Figure 5 (As shown). Subsequently, the Otsu method (OTSU) was used to perform adaptive threshold segmentation on the RSVI image. Given that ship targets exhibit high-value characteristics in the RSVI image, the segmentation rule was set as follows: regions with pixel values ​​greater than the optimal threshold T were identified as candidate ship pixels, while those with values ​​less than T were considered as water background. This step effectively extracted potential candidate regions for ship targets.

[0067] Because light-colored water bodies may exhibit high optical response values, candidate regions often contain large connected patches formed by light-colored water bodies. To reduce the impact of false positives, this embodiment labels the candidate regions as connected components and filters them based on a connected component area threshold. Connected patches with an area greater than 300 pixels are extracted and removed from the candidate regions. Connected components with an area less than 300 pixels are retained as the set of dark-colored water body ship pixels (D), and this set (D) is saved separately. Figure 7 ).

[0068] The extracted pixels are used as light-colored water body regions. When extracting candidate pixels within these regions, a fixed threshold of 0.1 is used to distinguish RSVI based on the calculated Normalized Ship Index (RSVI). Pixels with RSVI values ​​greater than this threshold are considered as ship pixels, ultimately obtaining the set of ship pixels (L) for the light-colored water body region. Figure 8 ).

[0069] Step 4: Perform fusion and consistency processing on the ship pixel sets D and L to output the final ship pixel (ShipMask).

[0070] The ship pixel set D and L are fused; then connected component analysis is performed on the fused result, and connected patches with fewer than 7 pixels and more than 300 pixels are deleted. Post-processing is then performed through dilation and erosion morphological operations to finally obtain the ship target pixel set. Figure 9 ).

[0071] Furthermore, to further compare this invention with traditional methods, this embodiment selects the commonly used water body identification indices MNDWI and NDWI for vessel identification, calculates them within the main river channel generated in step one, and uses the OTSU method to set a threshold, comparing it with the method of this invention. Simultaneously, a total of 225 sample points are selected based on true-color imagery (135 for vessels and 90 for non-vessels, e.g., ...). Figure 10 The accuracy of the ship extraction results from the three methods was verified and compared (as shown in Table 1 and Table 2).

[0072] Table 1. Confusion matrix used to evaluate ship classification results

[0073]

[0074] Table 2. Accuracy of ship remote sensing extraction using different methods

[0075]

[0076] The validation accuracy results of the three methods show that the proposed method has better accuracy, with its overall classification accuracy (OA=0.880) and Kappa coefficient (0.751) being significantly higher than those of the MNDWI and NDWI methods. Compared with NDWI, RSVI's overall classification accuracy is improved by 0.093, and compared with MNDWI, it is improved by 0.080, indicating that RSVI has stronger distinguishing ability and higher consistency level for ship targets in complex river surface backgrounds.

[0077] In terms of category accuracy, RSVI achieved the highest mapping accuracy (PA=0.889) and user accuracy (UA=0.909) among the three methods, improving upon MNDWI by 0.059 and 0.073 respectively, and upon NDWI by 0.111 and 0.055 respectively. Simultaneously, RSVI exhibited the lowest omission error (OE=0.111) and misclassification error (CE=0.091), further demonstrating its significant advantages in reducing ship omissions and suppressing non-ship false detections. Furthermore, in terms of image performance, RSVI more effectively enhances the contrast between the ship and the water background, suppresses high-reflectivity bright spots and wave interference on the water surface, and makes the ship outline clearer. Figure 11 ).

[0078] While the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the invention. Those skilled in the art to which this invention pertains can make various modifications and refinements without departing from the spirit and scope of the invention.

Claims

1. A method for optical remote sensing identification of river and lake vessels with different water color types, characterized in that, The method includes: Obtain remote sensing surface reflectance data of rivers and lakes; The tasseled cap transformation is performed on the remotely sensed surface reflectance data to extract the luminance component TCB; The reflectance ratio NIR / SWIR in the near-infrared band and mid-infrared band is calculated from the remotely sensed surface reflectance data. A sliding window was set for near-infrared images of rivers and lakes, and the standard deviation of the center pixel of each sliding window was calculated. The TCB, NIR / SWIR, and standard deviation of each pixel are standardized and then summed to form the pixel ship index. Ship pixels were extracted from dark and light water areas in rivers and lakes using a combination of pixel ship index and threshold segmentation method. Ship pixels are obtained by merging dark and light water areas.

2. The method according to claim 1, characterized in that, Remote sensing surface reflectance data covering river and lake water areas are obtained, water body regions are extracted using water-related spectral indices, and water body masks are generated. River and lake water areas are then extracted based on the water body masks.

3. The method according to claim 1, characterized in that, For dark water areas, ship pixels are extracted by combining OTSU threshold segmentation with the pixel ship index. For light-colored water areas, ship pixels are extracted using a preset fixed threshold combined with the pixel ship index.

4. The method according to claim 1, characterized in that, The extraction of ship pixels from dark and light water areas in rivers and lakes using a pixel ship index combined with a threshold segmentation method includes: Threshold segmentation of river and lake waters is performed using the OTSU method to obtain candidate ship pixels, and the candidate regions are labeled with connected components. Extract connected patches whose connected region area exceeds a preset threshold, and mark the remaining connected regions as ship pixels in dark water areas; The extracted connected patches are marked as light-colored water areas, and ship pixels in the light-colored water areas are extracted using a pixel ship index combined with a threshold segmentation method.

5. The method according to claim 1, characterized in that, The TCB, NIR / SWIR, and standard deviation of each pixel are Z-score standardized and then summed.

6. The method according to claim 1, characterized in that, The method further includes performing consistency processing on the fused ship pixels to obtain the final ship pixels.

7. The method according to claim 6, characterized in that, The pixel consistency processing includes: Connectivity analysis is performed on the fusion results to remove connected patches whose number of pixels is outside the preset range; Morphological post-processing is performed on the connected pixels to obtain the final set of ship pixels.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 7.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 7.

Images