CN-121616972-B - Coastal tidal flat large-scale remote sensing automatic extraction method

CN121616972BCN 121616972 BCN121616972 BCN 121616972BCN-121616972-B

Abstract

The invention discloses a coastal tidal flat large-range remote sensing automatic extraction method, which belongs to the technical field of remote sensing tidal flat drawing and comprises the steps of calculating turbid water indexes pixel by pixel based on cloud removal time sequence remote sensing images, selecting comparison wave bands according to the turbid water indexes, calculating self-adaptive mixed water indexes by combining green wave bands, calculating a value range histogram of each scene image determined by the self-adaptive mixed water indexes, taking self-adaptive mixed water index values corresponding to derivative mutation points in the images as self-adaptive land segmentation threshold values, calculating improved normalized difference water indexes from scene to scene, extracting the maximum water surface range based on the improved normalized difference water indexes and the self-adaptive water Liu Fenge threshold values, extracting tidal flat exposure frequency images based on each scene image and the self-adaptive land segmentation threshold values, further extracting preliminary tidal flat spatial distribution, and generating a tidal flat spatial distribution visualization result after connectivity processing. The invention can realize the rapid, accurate and stable extraction of the tidal flat space distribution in a large-scale and complex environment.

Inventors

ZHOU BIN
SU QIANQIAN
LEI HUI
YU ZHIFENG
CHEN PENGYU
GU WENXUAN

Assignees

杭州师范大学

Dates

Publication Date: 20260508
Application Date: 20260128

Claims (9)

1. The coastal tidal flat large-scale remote sensing automatic extraction method is characterized by comprising the following steps of: Calculating a muddy water index pixel by pixel and selecting a contrast wave band based on the muddy water index, and calculating a self-adaptive mixed water index based on the contrast wave band and the GREEN light wave band based on the cloudy time sequence remote sensing image of a research area, wherein the muddy water index TWI=RED-SWIR 1 is calculated pixel by pixel, SWIR1 is the reflectivity of a first wave band of short wave infrared, pixels are classified according to a preset muddy threshold TWI th , when TWI is more than or equal to TWI th , the muddy water pixels are land and non-muddy water pixels when TWI is more than TWI th ; For each image determined by the adaptive mixed water index, calculating a value range histogram, smoothing the value range histogram, then solving a first derivative, and taking the adaptive mixed water index value corresponding to the derivative mutation point as an adaptive land and water segmentation threshold of the image; Calculating an improved normalized difference water body index for the cloud-removed time sequence remote sensing image on a scene-by-scene basis, extracting a maximum water surface range based on the improved normalized difference water body index and an adaptive water Liu Fenge threshold, extracting a tidal flat exposure frequency image based on each scene image and an adaptive land and water segmentation threshold, and extracting preliminary tidal flat spatial distribution by combining the tidal flat exposure frequency image and the maximum water surface range; And generating a tidal flat space distribution visualization result after connectivity treatment on the preliminary tidal flat space distribution.
2. The method for automatically extracting the coastal tidal flat in a large scale according to claim 1, wherein the construction of the cloud-removed time sequence remote sensing image comprises the following steps: Screening a Sentinel-2 surface reflectivity image covering a research area in a GEE cloud platform, and performing preliminary cloud mask processing based on a quality control wave band of the image; further correlating the Cloud score+S2_ HARMONIZED dataset, and carrying out refined elimination on the residual Cloud and Cloud shadows by utilizing the Cloud detection quality assurance wave band; And counting the number of effective pixels of the image after cloud removal by scene, removing the image with the number of effective pixels lower than a preset threshold value, and constructing to obtain the high-quality cloud removal time sequence remote sensing image.
3. The method for automatically extracting the coastal tidal flat in a large-scale remote sensing manner according to claim 1, wherein the adaptive mixed water body index value corresponding to the derivative mutation point is used as the adaptive land and water segmentation threshold of the local image, and the method comprises the following steps: And positioning a first significant peak point in an interval of which the self-adaptive mixed water body index value is greater than zero in the histogram, searching from the peak point along the decreasing direction of the self-adaptive mixed water body index value, and determining the self-adaptive mixed water body index value corresponding to the first position of the first derivative from negative to positive as the self-adaptive water Liu Fenge threshold of the current image.
4. The method for automatically extracting the coastal tidal flat in large scale according to claim 1, wherein the calculating the improved normalized difference water body index for the cloud-removed time sequence remote sensing image scene by scene, extracting the maximum water surface range based on the improved normalized difference water body index and the adaptive water Liu Fenge threshold value comprises: For the cloud-removed time sequence remote sensing image, calculating an improved normalized difference water index mNDWI = (GREEN-SWIR 1)/(green+swir1) by each scene, wherein GREEN is the reflectivity of a GREEN light wave band, SWIR1 is the reflectivity of a short wave infrared first wave band, and maximum value synthesis is carried out on mNDWI of all time phases by each pixel to generate a mNDWI maximum value image; And applying the self-adaptive amphibious segmentation threshold value to mNDWI maximum value images to carry out amphibious binary segmentation, wherein the areas with pixel values larger than or equal to the self-adaptive amphibious segmentation threshold value are classified as water bodies, the areas with pixel values smaller than the self-adaptive amphibious segmentation threshold value are classified as non-water bodies, and the segmented water body areas are converted into vectors and are subjected to fusion treatment to obtain a complete and continuous maximum water surface range.
5. The method for automatically extracting the coastal tidal flat in a large-scale remote sensing mode according to claim 1, wherein the extracting the tidal flat exposure frequency image based on each view image and the adaptive land and water segmentation threshold value comprises the following steps: For each image determined by the adaptive mixed water body index, carrying out amphibious binary segmentation by using the corresponding adaptive amphibious segmentation threshold value to generate a time sequence amphibious mask image; Counting the times that the time sequence amphibious mask image is identified as a non-water body in all effective observations pixel by pixel, calculating the ratio of the times to the total effective observations as the tidal flat exposure frequency, and assigning the tidal flat exposure frequency back to the corresponding pixel position to generate a tidal flat exposure frequency image.
6. The method for automatically extracting the coastal tidal flat in a large-scale remote sensing manner according to claim 1, wherein the method for extracting the preliminary tidal flat spatial distribution by combining the tidal flat exposure frequency image and the maximum water surface range comprises the following steps: And extracting a region with the tidal flat exposure frequency higher than a preset threshold value from the tidal flat exposure frequency image as a potential tidal flat region, and performing mask treatment on the potential tidal flat region by utilizing the maximum water surface range to exclude a stable land region which is not submerged under high tide level, so as to obtain the preliminary tidal flat spatial distribution.
7. The coastal tidal flat large-range remote sensing automatic extraction device is realized by the coastal tidal flat large-range remote sensing automatic extraction method according to any one of claims 1-6 and is characterized by comprising an index calculation module, a threshold determination module, a tidal flat extraction module and a result optimization module; The index calculation module is used for calculating turbid water indexes pixel by pixel based on cloud removal time sequence remote sensing images of the research area, selecting a comparison wave band according to the turbid water indexes, and calculating a self-adaptive mixed water index based on the comparison wave band and a green light wave band; the threshold determining module is used for calculating a value domain histogram of each scene image determined by the adaptive mixed water index, obtaining a first derivative after smoothing, and taking an adaptive mixed water index value corresponding to a derivative mutation point as an adaptive land and water segmentation threshold of the scene image; the tidal flat extraction module is used for calculating an improved normalized difference water body index for the cloud-removed time sequence remote sensing image in a scene-by-scene manner, extracting a maximum water surface range based on the improved normalized difference water body index and a self-adaptive water Liu Fenge threshold value, extracting a tidal flat exposure frequency image based on each scene image and the self-adaptive water-land segmentation threshold value, and extracting preliminary tidal flat space distribution by combining the tidal flat exposure frequency image and the maximum water surface range; The result optimization module is used for generating a tidal flat space distribution visualization result after connectivity processing of the preliminary tidal flat space distribution.
8. An electronic device comprising a memory and one or more processors, the memory being configured to store a computer program, wherein the processor is configured to implement the coastal tidal flat wide area remote sensing automated extraction method of any one of claims 1-6 when the computer program is executed.
9. A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a computer, implements the coastal tidal flat large-scale remote sensing automatic extraction method of any one of claims 1 to 6.

Description

Coastal tidal flat large-scale remote sensing automatic extraction method Technical Field The invention belongs to the technical field of remote sensing tidal flat mapping, and particularly relates to a coastal tidal flat large-range remote sensing automatic extraction method. Background Tidal beaches are areas which are regularly submerged under the action of tides, are submerged under water in high tide and are exposed to the water surface in low tide, and comprise the types of mud beaches, sand beach, stone beaches and the like, and are important components of coastal zones. As a biodiversity hot spot area, tidal beaches provide key habitats for coastal organisms and play an important role in wind prevention, shore fixation, resource supply, water quality purification, carbon sequestration and the like. However, as one of the most vulnerable ecosystems worldwide, tidal beaches are facing a dual threat of natural factors (such as sea level rise and climate change) and artificial activities (such as industrialization and coastal zone development). With the continuous reduction of tidal flat range, the linkage environmental problems of ecological function decline, biodiversity loss and the like are caused. Therefore, the accurate monitoring of the tidal flat range is of great significance to the protection and repair of the tidal flat wetland and the sustainable development of the coastal zone. The tidal flat has the characteristics of wide range, quick change and periodical inundation, the traditional field investigation method is difficult to develop in a large range, and the remote sensing technology has become an effective means for monitoring the tidal flat due to the characteristics of large range, low cost and real-time property. The current common tidal flat mapping scheme mainly comprises the following categories of a tidal model-based method, a machine learning-based classification method, a threshold-based method, a simple and convenient implementation, high efficiency and general adoption of an empirical threshold or an Otsu self-adaptive threshold, wherein the spatial-temporal resolution of satellite images and the tidal level difference of different areas are subject to uncertainty of observation results, the machine learning-based classification method is high in classification accuracy but depends on a large number of training samples, and the threshold-based method is high in cost. In recent years, research trends gradually evolve towards multi-source remote sensing data fusion and automation threshold optimization directions so as to improve the robustness and precision of tidal flat identification. For example, patent document with publication number CN119048852a provides an automatic classification method and device for a coastal ecosystem based on multi-source remote sensing fusion, cloud layer filtering is performed by integrating Landsat and MODIS images and adopting NSPI algorithm, a regression-based space-time fusion model is introduced, fusion images with small cloud amount and continuous time are reconstructed, and further, the automatic distinction of tidal beaches, mangroves and salt and biogas is realized by combining a spectrum index curve. The patent document with publication number CN116152655A provides a coastal wetland classification method, a device, equipment and a storage medium, which are used for carrying out primary classification by synthesizing the image of the flood tide, the multiple times of summer and winter and utilizing the vegetation index of the enhanced mangrove forest and the OTSU algorithm, automatically extracting random sample points based on tide characteristics and physical characteristics and calculating a classification threshold value, thereby improving the objectivity and the repeatability of classification. Although the above method has advanced in terms of multi-source fusion and automation, the empirical threshold is highly subjective, whereas the Otsu method has unstable segmentation effect when the gray histogram does not form a typical bimodal distribution. Therefore, how to realize the rapid and accurate extraction of the tidal flat range is still a research target to be broken through in the current remote sensing monitoring field. Disclosure of Invention In view of the above, the invention aims to provide a coastal tidal flat large-scale remote sensing automatic extraction method, which is based on a Google EARTH ENGINE cloud platform to efficiently process mass remote sensing images, and is suitable for application fields of coastal zone resource monitoring, ecological assessment, protection management and the like by constructing an adaptive mixed water index capable of automatically adapting to the turbidity of a water body and combining an adaptive threshold segmentation mode based on derivative mutation point detection, overcoming the dependence of a traditional method on an ideal bimodal histogram, further fusing time sequence freq