A remote sensing coastline data fusion method

Abstract

The present application relates to a kind of remote sensing coastline data fusion method, first selected study area, study period and for extracting remote sensing data source;Then, water body index calculation is carried out to each kind of remote sensing image set of adjacent time in study period, result is assigned to the new band created;Second, using Otsu algorithm, the histogram characteristics of water body index in the whole image is realized by statistics to realize the automatic selection of global threshold, the water-land two-class division is carried out to image by threshold, and water-land boundary is obtained;Then, water-land boundary is carried out smoothing processing, and the coastline of each data source of the studied area is obtained;Finally, the obtained each coastline is integrated and fused, and the integrated average coastline is the coastline of the study area;The present application can integrate and fuse the coastline extracted from different satellite sources in selected area, obtain the integrated average coastline, to reduce the system error caused by different sensors, and obtain more accurate coastline.

Description

TECHNICAL FIELD

[0001] The present application relates to a remote sensing coastline data fusion method, belonging to the technical field of remote sensing technology and coastal zone observation application. BACKGROUND

[0002] The coastline is the boundary between the sea and the land, and is the lifeline and golden line of marine economic development, and has important ecological functions and resource values. Monitoring and quantifying the position changes of the coastline are the basis and key of current coastal management and restoration research and future coastal planning. Satellite remote sensing technology can objectively, accurately and economically and conveniently monitor the coastline and the dynamic changes of the coastline and the coastal development and utilization, and provides favorable conditions for carrying out the coastline monitoring, and can carry out the evaluation that cannot be carried out by traditional methods.

[0003] However, in the prior art, the dynamic monitoring of the coastline by using the remote sensing technology can only record the change conditions of the coastline, and cannot obtain more accurate coastline boundaries, and therefore, it is urgent to provide a coastline monitoring method based on the remote sensing technology to more accurately monitor the coastline and the coast. SUMMARY

[0004] The present application provides a remote sensing coastline data fusion method, which can integrate and fuse the coastlines extracted from different satellites in a selected area, to obtain a fused average coastline, to weaken the system errors caused by different sensors, and to obtain a more accurate coastline.

[0005] The technical solution adopted by the present application to solve the technical problems is:

[0006] A remote sensing coastline data fusion method, specifically comprising the following steps:

[0007] Step S1: selecting a research area and a research period of the coastline, and collecting satellite images of the research area in the research period by using a satellite, and performing image registration, radiation correction and atmospheric correction on the obtained satellite images;

[0008] Step S2: calculating the water body index of the satellite images obtained in step S1, and merging the calculation results to the original wave band; wherein the selected satellite images contain water body index information of each single pixel in the research area, the water body index adopts NDWI or MNDWI, and the calculation formula is as follows:

[0009]

[0010] In formula (1), ρ Green is the reflectivity of the green wave band, and ρ Nir is the reflectivity of the near-infrared wave band;

[0011] Step S3: using Otsu algorithm, realizing automatic selection of global threshold value by statistics of histogram characteristics of water body index in selected satellite image, carrying out water-land two-class division on satellite image through threshold value, and obtaining water-land boundary line;

[0012] Step S4: carrying out smoothing processing on water-land boundary line, and obtaining coastlines of each data source in research region;

[0013] Step S5: integrating and fusing each obtained coastline, and taking fused average coastline as the coastline of the research region;

[0014] As a further optimization of the application, the satellite image selected in step S1 is a multi-source satellite image, the interval between the shooting moments in the research period is within 10 days, and no extreme weather event occurs in the research period, the extreme weather event including typhoon or storm surge;

[0015] As a further optimization of the application, the satellite adopted in step S1 is Landsat 8 satellite or Sentinel 2 series satellite or Super Dove series satellite;

[0016] Among them, the sensor adopted in Landsat 8 satellite is OLI and ETM+, the sensor adopted in Sentinel 2 series satellite is MSI, and the sensor adopted in Super Dove series satellite is PS2.SD and PSB.SD;

[0017] As a further optimization of the application, the Otsu algorithm adopted in step S3 assumes that the satellite image includes two types of pixels according to the bimodal histogram, and the Otsu algorithm calculates the best threshold value for separating the two types of pixels after degrading the gray image of the satellite image into a binary image;

[0018] Two separated best threshold values realize that the intra-class variance is minimum, the inter-class variance is maximum, and the MNDWI value presents a bimodal distribution, one side peak value is the land pixel cluster with MNDWI less than zero, and the other side is water body pixel, and a clear water-land boundary line is obtained;

[0019] As a further optimization of the application, in step S4, the forward algorithm in HMM is used to process the water-land boundary line, a random sequence of unobservable states is randomly generated by the hidden Markov model in HMM, and the process of generating observable random sequences from each state is continued, so that the step line vector is gradually smoothed into a coastline with sub-pixel accuracy smoothing;

[0020] As a further preferred embodiment of the present application, the integrated fusion of each coastline in step S5 is a fusion average coastline, specifically, the resolution of different satellite images is weighted to construct a baseline and a sample band for each coastline obtained in step S4, and the weighted average of the intersection position of each coastline on each sample band is obtained, and each sample band after weighted average is the fusion average coastline.

[0021] As a further preferred embodiment of the present application, the specific steps of constructing a baseline and a sample band for each coastline obtained in step S5 are as follows:

[0022] Step S51: the sample band is vertically drawn at equal intervals along the baseline at the intersection of each coastline, and the intersection point generated at the nth sample band is obtained, and the distance between each coastline and the sample band is x i(n) .

[0023] Step S52: the resolution of different satellite images is obtained to obtain the distance x i(n) of each coastline to the sample band in the fusion average calculation; i The weight w i in the fusion average calculation is calculated according to the following formula:

[0024]

[0025] In formula (2), p i is the resolution of each satellite image.

[0026] Step S53: the distance x 0(n) of each sample band to the sample band after fusion average is obtained, and the calculation formula is:

[0027] x 0(n) =∑x i(n) *w i #(3)

[0028] In formula (3), p i is the resolution of each satellite image.

[0029] Step S54: connecting each intersection point to obtain the coastline of the study area.

[0030] Compared with the prior art, the remote sensing coastline data fusion method provided by the present application has the following beneficial effects:

[0031] The remote sensing coastline data fusion method provided by the present application can automatically identify and extract the coastline, and integrate and fuse the extracted coastlines of different data sources to weaken the system error caused by the sensor and obtain a more accurate coastline. BRIEF DESCRIPTION OF DRAWINGS

[0032] The present application will be further described below in combination with the drawings and examples.

[0033] Figure 1is a true color satellite image schematic diagram provided by the present application;

[0034] Figure 2 is a gray scale image of water body index NDWI value obtained by the present application through a remote sensing coastline data fusion method;

[0035] Figure 3 is a threshold value schematic diagram obtained by the present application using Otsu algorithm, and 3a-3c are threshold value schematic diagrams of three embodiments;

[0036] Figure 4 is a coastline schematic diagram obtained by the present application through a remote sensing coastline data fusion method by preliminary extraction of coastlines of each satellite image;

[0037] Figure 5 is a schematic diagram of a fusion average algorithm used in the remote sensing coastline data fusion method provided by the present application;

[0038] Figure 6 is a coastline schematic diagram finally obtained by the present application through the remote sensing coastline data fusion method. DETAILED DESCRIPTION

[0039] The present application will now be further described in detail with reference to the accompanying drawings. In the description of the present application, it should be understood that the terms "left side", "right side", "upper part", "lower part" and the like indicate the orientation or positional relationship shown in the drawings, and are only used for the purpose of facilitating the description of the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and "first", "second" and the like do not represent the importance of the parts, and therefore cannot be understood as a limitation on the present application. The specific dimensions used in the present embodiment are only used to illustrate the technical solutions and do not limit the protection scope of the present application.

[0040] As set forth in the background, some of the existing technologies for monitoring the coastline currently use satellite remote sensing technology, but it is not enough to rely solely on satellite remote sensing technology, because different satellites necessarily involve different sensors, and different sensors are prone to system errors, resulting in an inability to accurately obtain the coastline.

[0041] Based on this, the present application provides a remote sensing coastline data fusion method, which obtains images through different satellites, integrates and fuses the coastlines extracted from different satellites in the selected area to obtain a fusion average coastline, so as to weaken the system errors caused by different sensors and obtain a more accurate coastline.

[0042] Specifically, the following steps are included:

[0043] Step S1: select a research area of a coastline and a research period, collect satellite images of the research area in the research period by satellite, and perform image registration, radiation correction, and atmospheric correction on the collected satellite images; the selected satellite images are multi-source satellite images, and the satellite source selection is open under the condition that the interval between the shooting moments in the research period is within 10 days and no extreme weather events such as typhoons and storm surges occur in the research period to cause sudden changes in the beach morphology; this selection method can improve the monitoring accuracy while reducing the monitoring cost.

[0044] When selecting a remote sensing satellite data source, Landsat 8 satellite, Sentinel 2 satellite, and Super Dove satellite can be selected because different resolution levels and sensor types need to be considered. The sensor used in Landsat 8 satellite is OLI (Operational Land Imager) and ETM+ (Enhanced Thematic Mapper), the sensor used in Sentinel 2 satellite is MSI (Multispectral Instrument), and the sensor used in Super Dove satellite is PS2.SD and PSB.SD.

[0045] Step S2: calculate the water body index of the satellite images obtained in step S1, and combine the calculation results to the original band; the selected satellite images contain water body index information of each single pixel in the research area, and the water body index is NDWI or MNDWI, and the calculation formula is as follows:

[0046]

[0047] In formula (1), ρ Green is the reflectivity of the green band, and ρ Nir is the reflectivity of the near-infrared band.

[0048] Step S3: The Otsu algorithm is used to automatically select the global threshold value by counting the histogram characteristics of the water index of the selected satellite image, and the water-land boundary is obtained by dividing the satellite image into two categories of water and land through the threshold value. The threshold value is calculated by using the Otsu algorithm, which degrades a gray-scale image to a binary image. The algorithm assumes that the image contains two types of pixels according to the bimodal histogram, so it calculates the best threshold value that can separate the two types of pixels to minimize the intra-class variance. Since the squared distance between two points is constant, the inter-class variance is maximized, so the MNDWI value presents a good bimodal distribution. Assuming that the left peak value is the land pixel cluster with MNDWI less than zero, the right side is the water pixel, so the water-land boundary obtained by segmentation is clear.

[0049] Step S4: The water-land boundary is smoothed to obtain the coastline of each data source in the study area. The water-land boundary obtained in step S3 is a vector representation following the pixel edge, which presents a stepwise line with a step size equal to the pixel resolution of the optical satellite image. The forward algorithm of the three classical algorithms of HMM (Hidden Markov Model) is used to process the boundary, which generates a random sequence of unobservable states by a hidden Markov chain, and then generates an observable random sequence by each state, which smooths the stepwise line vector to a coastline with sub-pixel accuracy.

[0050] Step S5: The obtained coastlines are integrated and fused. Specifically, the weight of each satellite image is divided according to the resolution of the different satellite images, and the baseline and sample strip are constructed by the digital coastline calculation system for the coastlines obtained in step S4. The weighted average of the intersection position of each coastline on each sample strip is calculated, and the weighted average of each sample strip is the fused average coastline, which is taken as the coastline of the study area.

[0051] The specific steps of constructing the baseline and sample strip of the coastline are as follows:

[0052] Step S51: The sample strip is drawn vertically along the baseline at equal intervals, and the intersection point generated at the nth sample strip is obtained, and the distance between the coastline and the sample strip is x i(n) ;

[0053] Step S52: The resolution of each satellite image is obtained according to the resolution of the different satellite images, and the resolution of each satellite image is obtained according to the resolution of the different satellite images. i(n) The weight w i in the fused average calculation is calculated by the following formula:

[0054]

[0055] In formula (2), p i is the resolution of each satellite image.

[0056] Step S53: Obtain the fused average coastline and the distance x of each sample belt 0(n) The calculation formula is:

[0057] x 0(n) =∑x i(n) *w i #(3)

[0058] In formula (3), p i is the resolution of each satellite image;

[0059] Step S54: Connect each intersection to obtain the coastline of the study area.

[0060] In order to better illustrate the superiority of the above fusion method, the present application gives relevant embodiments to describe in detail the extraction of the coastline of a certain place in Rizhao, China in October 2022. First, select the coastline of Rizhoa in the study area and the study period October 2022, and perform image registration, radiation correction and atmospheric correction on the selected satellite images to obtain the corrected satellite images shown in Figure 1 The Landsat 8 satellite image was taken on October 7, 2022, and the sensor used was OLI (Operational Land Imager, Land Imager), ETM+ (Enhanced Thematic Mapper, Enhanced Thematic Mapper), with a resolution of 30 m. The Sentinel 2 satellite image was taken on October 13, 2022, and the sensor used was MSI (Multispectral Instrument, Multispectral Sensor), with a resolution of 10 m. The Super Dove satellite image was taken on October 12, 2022, with a resolution of 3 m. The selection of this group of satellite data sources covers different resolution levels and sensor types.

[0061] Next, the obtained images are subjected to water body index calculation, and the gray scale diagram of the water body index NDWI value is as shown in Figure 2

[0062] Secondly, using Otsu algorithm, the histogram characteristics of the water body index in the whole image are counted to realize the automatic selection of global threshold value, as shown in Figure 3 ​The threshold values calculated from the Landsat 8 satellite, the Sentinel 2 satellite and the Super Dove satellite images are 0.3363397768731364, 0.30078125 and 0.1757931343448877 respectively.

[0063] Next, the water-land boundary is smoothed to obtain the satellite coastlines in the study area, as shown in Figure 4

[0064] Finally, the obtained coastlines in the region are integrated and fused, a reference line on land is drawn based on the three coastlines extracted in the foregoing steps to construct a baseline, sample bands are drawn vertically along the baseline at intervals of 5 m, the positions of the fused coastlines on each sample band are calculated, as shown in Figure 5 The average fused coastline is taken as the coastline of the study area, as shown in Figure 6

[0065] It can be seen that, first, the study area, the study period and the remote sensing data source used for extraction are selected; next, the water body index of each set of remote sensing images at adjacent time points in the study period is calculated, and the result is assigned to a new band created; secondly, the Otsu algorithm is used to realize automatic selection of the global threshold by counting the histogram characteristics of the water body index in the entire image, the image is divided into two categories of water and land through the threshold, and the water-land boundary is obtained; next, the water-land boundary is smoothed to obtain the coastlines of each data source in the study area; finally, the obtained coastlines are integrated and fused, and the average fused coastline is taken as the coastline of the study area; that is, the coastlines extracted from different data sources are integrated and fused to reduce the systematic error caused by the sensor and obtain a more accurate coastline.

[0066] Those skilled in the art can understand that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art to which the present application belongs. It should also be understood that terms such as those defined in general dictionaries should be understood as having meanings consistent with those in the prior art, and should not be interpreted in an idealized or overly formal sense unless otherwise defined.

[0067] The meaning of "and / or" described in the present application means that each single existence or both existences are included.

[0068] The meaning of "connection" described in the present application can be a direct connection between components or an indirect connection between components through other components. ​​

[0069] With the above ideal embodiments according to the present application as the inspiration, through the above description, relevant staff can make various changes and modifications without deviating from the scope of the technical idea of the present application. The technical scope of the present application is not limited to the content of the specification, and must be determined according to the scope of the claims.

Claims

1. A method for fusing remote sensing coastline data, characterized in that: Specifically, the following steps are included: Step S1: Select the study area and study period of the coastline, and acquire satellite images of the study area during the study period. The acquired satellite images are then image registered, radiometrically corrected, and atmospherically corrected. Step S2: Calculate the water index on the satellite imagery acquired in Step S1, and merge the calculation results into the original bands. The selected satellite imagery contains water index information for each individual pixel within the study area. The water index uses NDWI or MNDWI, and the calculation formula is as follows: In formula (1), p Green is the reflectivity of the green wave band, p Nir is the reflectivity of the near-infrared wave band; Step S3: Using the Otsu algorithm, the global threshold is automatically selected by statistically analyzing the histogram characteristics of the water index in the selected satellite images. The satellite images are then divided into land and water categories based on the threshold to obtain the land-water boundary line. Step S4: Smooth the water-land boundary line to obtain the coastline of each data source within the study area; Step S5: Integrate and fuse the acquired coastlines, and take the fused average coastline as the coastline of the study area. Specifically, according to the resolution of different satellite images, weight the satellite images, construct the baseline and transects of the coastlines acquired in step S4 through the digital coastline calculation system, and perform a weighted average of the intersections of the coastlines on each transect. The weighted average of each transect is the fused average coastline. The specific steps for constructing baselines and transects based on the coastline obtained in step S5 are as follows: Step S51: Draw transects perpendicularly along the baseline at equal intervals, intersecting each shoreline. The intersection point generated at the nth transect gives the distance x between each shoreline and the transect. i(n) ; Step S52: Obtain the x-axis of each coastline based on the resolution of different satellite images. i(n) Weight w in fusion average calculation i The formula for calculating the weights is: In formula (2), p i The resolution of each satellite image; Step S53: Obtain the averaged distance x between the shoreline and the transect for each transect. 0(n) The calculation formula is: x 0(n) =∑x i(n) *w i #(3) In formula (3), p i The resolution of each satellite image; Step S54: Connect the intersection points to obtain the coastline of the study area.

2. The remote sensing coastline data fusion method according to claim 1, characterized in that: The satellite imagery selected in step S1 is multi-source satellite imagery. The time interval between the images taken during the study period is 10 days. No extreme weather events occurred during the study period. The extreme weather events include typhoons or storm surges.

3. The remote sensing coastline data fusion method according to claim 2, characterized in that: The satellite used in step S1 is a Landsat 8 satellite, a Sentinel 2 series satellite, or a Super Dove series satellite; The Landsat 8 satellite uses OLI and ETM+ sensors for its data, the Sentinel 2 series satellites use MSI sensors for their data, and the Super Dove series satellites use PS2.SD and PSB.SD sensors for their data.

4. The remote sensing coastline data fusion method according to claim 2, characterized in that: The Otsu algorithm used in step S3 assumes that the satellite image includes two types of pixels according to the dual-mode histogram. The Otsu algorithm degrades the grayscale image of the satellite image into a binary image and then calculates the optimal threshold for separating the two types of pixels. Two separate optimal thresholds are used to achieve the minimum intra-class variance and the maximum inter-class variance. The MNDWI value exhibits a bimodal distribution, with one peak representing the land pixel clusters where MNDWI is less than zero and the other peak representing the water pixels, resulting in a clearly segmented land-water boundary.

5. The remote sensing coastline data fusion method according to claim 4, characterized in that: In step S4, the forward algorithm in HMM is used to process the land-water boundary line. The unobservable random sequence of states is randomly generated by the Hidden Markov Model in HMM, and the process of generating observable random sequences from each state continues, gradually smoothing the line vector into a coastline with sub-pixel precision smoothing.

Images