Erosion gully morphological index rapid extraction method and system

Abstract

The invention discloses a method and a system for rapidly extracting morphological indexes of an erosion gully, and belongs to the technical field of water and soil conservation. The method comprises the following steps: S1, watershed segmentation; s2, identifying an erosion gully object; comprising the following steps: performing spectral feature analysis based on an orthoimage to obtain a first principal component; selecting a texture feature variable and a topographic feature variable; performing multi-scale image segmentation on the first principal component based on the texture feature variable; identifying whether each object is an erosion gully according to the topographic feature variables; and correcting the identification result to obtain a final erosion gully object identification result. S3, channel line extraction is carried out on the erosion gully; extracting a river network water system through hydrological analysis, correcting the river network water system, and taking the corrected river network water system as a channel line of the erosion gully; and S4, extracting the morphological index of the erosion gully according to the erosion gully object identification result and the gully line. The method can effectively improve the extraction precision of the morphological characteristics of the erosion gully, and breaks through the problem of poor extraction precision of the erosion gully.

Description

Technical Field

[0001] The invention belongs to the technical field of soil and water conservation, and in particular relates to a method and system for quickly extracting erosion gully morphological indicators. Background Art

[0002] The water conservancy department has achieved full coverage of annual dynamic monitoring of soil and water loss, and has carried out factor extraction, modulus calculation and dynamic analysis and evaluation of soil and water loss by combining remote sensing monitoring, field investigation, model calculation and statistical analysis. As the monitoring scope covers the whole country, higher requirements are put forward for image resolution and time phase, and it is necessary to explore the application of images with high time and spatial resolution to carry out work. Among them, topography is one of the important factors affecting soil and water loss, including slope and slope length factors, which need to be extracted based on large-scale topographic maps or high-precision DEM. Gaofen-7 Satellite (GF-7) is an outstanding representative of the series of satellites of China's High-Resolution Earth Observation System Major Project (GF Project). It can obtain full-color stereo images with a resolution better than 0.8m, which meets the urgent demand for high-precision stereo mapping data based on remote sensing investigation and model calculation. It is very necessary to improve the quantification accuracy of soil and water loss. The high-precision stereo images obtained provide a basis for improving the calculation accuracy of terrain factors.

[0003] In recent years, the country has implemented gully control projects and achieved remarkable results in soil and water loss control. Carrying out research on gully and channel changes to provide a scientific basis for precise control of gully soil and water loss is the development direction and urgent need of high-resolution remote sensing soil and water conservation applications. In terms of spatial scale, the mapping accuracy of domestic satellite remote sensing images was low in the past, with a mapping accuracy of 1:50,000 scale, which could not meet the needs of soil and water conservation monitoring at a small spatial scale; the acquisition of terrain data such as slope, slope length, and slope shape was mainly based on digital elevation (DEM) data generated from 1:50,000 data topographic maps, which had great limitations in accuracy. In terms of time scale, the real-time acquisition of remote sensing data was difficult and the cycle was long, resulting in insufficient background data for monitoring and research objects, which seriously restricted the development and application of soil and water conservation monitoring and evaluation based on high-resolution remote sensing, and research on gully and channel changes. The soil and water conservation monitoring and evaluation work based on high-resolution remote sensing is still at a relatively low level. There is an urgent need for domestically produced GF7 high-resolution remote sensing image data with full coverage, high accuracy, fast updating, and capable of stereo mapping. We also need to explore methods for extracting gully indicators from domestically produced GF7 stereo mapping image data to promote and improve the level of soil and water conservation monitoring and evaluation work. Summary of the invention

[0004] The present invention aims to solve one of the technical problems in the above-mentioned related art at least to a certain extent.

[0005] To this end, the purpose of the present invention is to provide a method and system for quickly extracting erosion gully morphological indicators, which can effectively improve the accuracy of erosion gully morphological feature extraction and overcome the problem of poor erosion gully extraction accuracy.

[0006] In order to solve the above-mentioned technical problems, the present invention is achieved as follows:

[0007] The embodiment of the present invention provides a method for quickly extracting erosion gully morphology indicators, the method comprising:

[0008] S1. Segment the watershed based on DEM data to obtain several watershed units;

[0009] S2. Identify gully objects in watershed units; the identification content includes:

[0010] Based on the orthophoto of the watershed unit, spectral feature analysis is performed to extract the red, green and blue band images and construct the erosion gully interpretation signs;

[0011] Perform principal component transformation on the extracted image, and select the component with the most three-band information as the first principal component;

[0012] Selecting a number of texture feature variables and a number of terrain feature variables required for subsequent processing;

[0013] Based on the selected texture feature variables, multi-scale image segmentation is performed on the first principal component;

[0014] According to the segmented terrain characteristic variables, identify whether each object in the watershed unit is an erosion gully;

[0015] Correct the recognition result to obtain the final erosion gully object recognition result;

[0016] S3, extracting the channel line of the erosion gully;

[0017] Based on the DEM data of the current watershed unit, the hydrological analysis is carried out, the river network is extracted and then corrected, and the final river network is used as the channel line of the erosion gully;

[0018] S4, extracting morphological indicators of erosion gullies;

[0019] According to the erosion gully object recognition result of S2 and the gully line of S3, the morphological index of the erosion gully is extracted.

[0020] In addition, the method for quickly extracting erosion gully morphology indicators according to the present invention may also have the following additional technical features:

[0021] In some implementations, the first principal component accounts for more than 99% of the information of the three bands of red, green and blue.

[0022] In some of the embodiments, the selected texture feature variables include contrast, dissimilarity, homogeneity, angular second moment, entropy, mean, variance, and correlation.

[0023] In some of the embodiments, the selected terrain characteristic variables include elevation, slope, sunlight simulation, and runoff accumulation.

[0024] In some embodiments, the rule for identifying whether each object in a watershed unit is an erosion gully is:

[0025] Objects with a negative terrain ratio greater than a first threshold are identified as erosion gullies, and objects with a slope less than a second threshold are identified as non-erosion gullies, or

[0026] Objects in the shaded area are identified as erosion gullies.

[0027] In some implementations, performing hydrological analysis on DEM data to extract river network contents includes:

[0028] Extract river network based on D8 algorithm and Bumin algorithm;

[0029] Based on the extraction results, the Strahler classification method is used to classify the extracted river network system, and the buffering analysis is carried out on the rivers at all levels. The difference between the average elevation within the river line and the average elevation in its buffer zone is calculated respectively. The channels with a difference greater than the third threshold are identified as erosion channels, and the final river network system is obtained by combining the extracted river network system.

[0030] In some of the embodiments, the watershed threshold is set to 800 when extracting river network water systems based on the D8 algorithm and the Burn in algorithm.

[0031] In some of the embodiments, gully morphology indicators are extracted from three scales: watershed, gully, and channel.

[0032] In some of the embodiments, the morphological indicators at the watershed scale include: gully area, gully length, gully density, and gully fragmentation;

[0033] The morphological indicators of gully scale include: the main gully to branch gully ratio and the longitudinal gradient of the main gully; the morphological indicators of gully scale include:

[0034] The morphological indicators of channel scale include: channel area, channel length and channel depth.

[0035] The embodiment of the present invention further provides a system for quickly extracting erosion gully morphology indicators, characterized in that it can implement any one of the methods for quickly extracting erosion gully morphology indicators described above; the system comprises:

[0036] The watershed segmentation module is used to segment the watershed based on DEM data to obtain several watershed units;

[0037] The erosion gully object recognition module is used to identify erosion gully objects in the watershed unit; the recognition content includes: based on the orthophoto of the watershed unit, spectral feature analysis is performed to extract the red, green and blue three-band images and construct the erosion gully interpretation mark; the extracted image is transformed into a principal component, and the component with the most three-band information is selected as the first principal component; a number of texture feature variables and a number of terrain feature variables required for subsequent processing are selected; based on the selected texture feature variables, the first principal component is segmented at multiple scales; according to the segmented terrain feature variables, each object in the watershed unit is identified as an erosion gully; the recognition result is corrected to obtain the final erosion gully object recognition result;

[0038] The channel line extraction module is used to extract the channel line of the erosion gully. The specific contents include: performing hydrological analysis based on the DEM data of the current watershed unit, extracting the river network and then correcting it, and using the final river network as the channel line of the erosion gully;

[0039] The morphological index extraction module is used to extract the morphological index of the erosion gully according to the erosion gully object recognition result and the channel line extraction result.

[0040] Compared with the prior art, the present invention has at least the following beneficial effects:

[0041] In the embodiment of the present invention, the provided method for rapid extraction of erosion gully morphological indicators uses principal component analysis to reduce the redundancy of Gaofen-7 image information data and improve algorithm efficiency; the first principal component after dimensionality reduction processing using principal component analysis is used for orthophoto segmentation, and the classification algorithm set and rule set are constructed by combining the terrain characteristics of the erosion gully and Gaofen-7 stereo mapping data, and the positive and negative terrain method, slope threshold method and spectral feature analysis are used to extract the erosion gully position. Compared with traditional erosion extraction methods, this method has the characteristics of high extraction accuracy, fast speed, strong robustness, etc.

[0042] In the embodiment of the present invention, the provided method for rapid extraction of erosion gully morphological indicators uses hydrological analysis to extract river network and channel lines from the DEM produced by the Gaofen-7 stereoscopic mapping remote sensing data, uses the D8 algorithm and the Burn in algorithm for correction, performs buffer zone analysis on the primary, secondary and tertiary river channels, and sets thresholds to eliminate pseudo channels. Compared with the traditional erosion gully line extraction method, the erosion channel line extraction method has a significant improvement in extraction accuracy;

[0043] In an embodiment of the present invention, a method for quickly extracting gully morphological indicators is provided. Based on the extraction structure of gully and gully channel lines, gully morphological indicators such as gully area, gully length, gully density, gully fragmentation, main channel longitudinal gradient, channel area, channel length, and channel depth are extracted from three scales: watershed, gully, and channel. This method can be applied in a business and engineering manner in the field of soil and water conservation.

[0044] The erosion gully morphology index rapid extraction system of the present invention can implement the erosion gully morphology index rapid extraction method, and thus at least has all the features and advantages of the erosion gully morphology index rapid extraction method, which will not be described in detail here. Additional aspects and advantages of the present invention will be partially given in the following description, and partially will become obvious from the following description, or will be understood through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A flowchart of a method for quickly extracting erosion gully morphology indicators disclosed in an embodiment of the present invention;

[0046] Figure 2 The present invention discloses an object-oriented erosion gully extraction flow chart according to an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] The embodiments of the present invention are described in detail below through specific embodiments and application scenarios in conjunction with the accompanying drawings.

[0049] See also Figure 1 As shown, in some embodiments of the present invention, a method for quickly extracting erosion gully morphology indicators is provided, and the steps of the method include:

[0050] Step 1: Watershed division

[0051] A watershed is a natural geographical boundary with strict geographical significance and has multiple levels of characteristics. In addition, erosion gullies often show different characteristics in different watersheds, such as quantity, spatial shape, and distribution pattern. Therefore, we first divide the watershed according to the DEM data and set the 2.5km 2 The watershed unit is divided into approximately 2.5 km 2 Small watershed unit.

[0052] Step 2: Characteristic analysis of erosion gully objects

[0053] 1) Spectral characteristics

[0054] Based on the high-resolution orthophoto obtained after preprocessing by GF-7, the red (R), green (G), and blue (B) band images are extracted. The image features are analyzed to construct the erosion gully interpretation mark. The interpretation mark is the basis for the following steps and serves as the reference truth value.

[0055] 2) Principal Component Analysis

[0056] First, the remote sensing image is transformed into principal components, and then principal component analysis (PCA) is performed. Then, the main component factors are extracted to enhance the spectral information of the image. From the detailed introduction table of the components in the principal component analysis (i.e., Table 1), it can be seen that the first principal component includes 99.2% of the information of the three bands of red, green, and blue. Selecting the first principal component for later erosion gully extraction without considering the second and third principal components can effectively reduce data redundancy and improve algorithm efficiency.

[0057] Table 1 Image texture feature calculation formula and meaning

[0058]

[0059] 3) Texture features

[0060] Texture features are also commonly used in the research of object classification and extraction. They contain relatively rich image information and are a global feature that reflects the slowly changing or periodic surface tissue structure arrangement properties of the surface of the object. Eight texture features of the first principal component based on principal component analysis are selected for subsequent analysis and extraction. The eight texture features are contrast, difference, homogeneity, angular second-order moment, entropy, mean, variance and correlation.

[0061] 4) Topographic features

[0062] Erosion gullies have obvious terrain features. Therefore, when performing image segmentation for erosion gully extraction, terrain features will contain more useful information. Four terrain feature variables were selected: elevation, slope, illumination simulation and runoff accumulation for subsequent analysis and extraction.

[0063] 5) Object-oriented erosion gully extraction method

[0064] Firstly, the first principal component extracted by principal component analysis and dimensionality reduction is subjected to multi-scale image segmentation, and then the classification algorithm set and rule set are constructed in combination with the terrain characteristics of the erosion gully object, such as Figure 2As shown, the object whose negative terrain ratio is greater than a certain threshold in the segmented object is identified as an erosion gully, and the object whose slope is less than a certain threshold is identified as a non-erosion gully object. Finally, the shadow area is identified as an erosion gully object in combination with the spectral characteristics of the erosion gully object, that is, if the negative terrain ratio is greater than a certain threshold, the slope is not less than a certain threshold, and it is determined to be a shadow area, it can be identified as an erosion gully object if one of the three conditions is met. The threshold of negative terrain depends on the specific research area, and the specific judgment method of the shadow area can adopt the existing technology, and the present invention does not make too many statements.

[0065] Secondly, there will be omissions and errors in the objects extracted by GF-7 based on the negative terrain method, which require manual interpretation and correction. The following is an explanation of the basis for manual correction. In order to intuitively express the color tone, vividness and brightness between different objects and increase the contrast, the GF-7 RGB orthophoto is converted into an HSV image, and the interpretation is based on the HSV image. For roads, houses and other areas in the flat area connected to the erosion gully area, non-erosion is mistakenly extracted as erosion gullies. This can be corrected by the slope threshold method, and the area with a slope less than 15° is deducted. For (I) the shadow area of ​​the shady slope that is orange in the HSV image, the shadow area of ​​the sunny slope gully that is blue in the HSV image, and (III) the area where the RGB orthophoto is gray and the HSV image is green, and the non-erosion gully is mistakenly extracted as an erosion gully, the index VI=HV is constructed, and the area with VI<0 is extracted to correct omissions and errors. For the areas where the shadow areas between the sunny slope gullies appear blue in the HSV image, the areas where the RGB orthophoto appears gray, the HSV image appears green, and the areas where non-erosion gullies are mistakenly mentioned as erosion gullies, the spectral characteristics cannot be well classified with other land features, and human-computer interaction is used to make corrections.

[0066] Step 3: Extraction of erosion gully lines

[0067] The river network extracted through hydrological analysis based on DEM data is highly consistent with the channel line of the erosion gully. Therefore, the river network can be used to extract the erosion gully with the help of DEM. However, there are differences between the river network produced directly based on DEM and the real channel line. Different confluence thresholds need to be set for different types of channels. If the threshold is set too large, the extracted water system often cannot reach the gully head, resulting in a shorter extracted channel. In addition, it will also cause the loss of erosion gullies. If a smaller threshold is set, parallel and pseudo channels will appear. By combining D8 and Bumin algorithms to correct the original DEM, parallel and pseudo channels can be reduced.

[0068] The river network water system is extracted based on D8 and Burn in algorithms, and the river network water system is classified using the Strahler classification method. In this embodiment, the water collection threshold is set to 800, which can ensure that the extracted river channel can reach the head of the erosion ditch, but there will be some pseudo-rivers. For the first-level, second-level, and third-level rivers, a suitable radius is set for buffering analysis, and the difference between the average elevation in the river line and the average elevation in its buffer zone is calculated respectively, and the ditch with a difference greater than the set threshold is identified as an erosion ditch. Buffering analysis can be implemented using GIS software. Although some pseudo-ditches can be eliminated based on this method, there is still a part that cannot be identified and needs to be manually corrected.

[0069] Step 4: Morphological indicator extraction

[0070] The erosion gully morphology index is mainly a statistical index based on the extraction of erosion gullies and channel lines. The erosion gully morphology index is extracted from three scales: watershed, gully and channel. The specific extraction indexes are shown in Table 2.

[0071] Table 2 Calculation formula and meaning of erosion gully morphology index

[0072]

[0073]

[0074] 1) Basin scale

[0075] Taking small watersheds as analysis units, the area and length of erosion gullies, erosion gully density, and erosion gully fragmentation are statistically analyzed. The area of ​​erosion gullies is directly based on the erosion gully spatial distribution data product and is statistically analyzed with watersheds as units. The length of erosion gullies is calculated by combining the watershed network with the range of erosion gullies. First, taking small watersheds as units, a smaller watershed threshold is set based on DEM data to ensure that the watershed grid contains the actual boundary of the erosion gully. The watershed network is corrected using the boundary of the erosion gully to ensure the adaptability of the gully head area to the watershed network. Finally, the length of erosion gullies is statistically analyzed with small watersheds as units. Based on the length and area of ​​erosion gullies, the density and fragmentation of erosion gullies are calculated. The density of erosion gullies is the ratio of the length of erosion gullies to the watershed area, and the fragmentation of erosion gullies is the ratio of the area of ​​erosion gullies to the watershed area.

[0076] 2) Ravine scale

[0077] Taking the erosion gully range as the analysis unit, the main gully to branch gully ratio and the longitudinal gradient of the main gully are calculated. The main gully to branch gully ratio is the ratio of the main gully length to the total gully length.

[0078] Based on DEM, the watershed network is extracted, and the source point of the main channel is determined in combination with the direction of water flow, and the main channel and branch channel are identified. Then, based on the conversion of the watershed grid grid to vector, the main channel branch channel ratio is counted and calculated, which can be achieved using ArcGIS software. The longitudinal gradient of the main channel of the erosion ditch is the ratio of the elevation drop of the main channel of the erosion ditch to its length. Based on the main channel extracted from the watershed grid, the head and tail of the main channel are identified, the elevation difference is calculated, and the longitudinal gradient of the channel is calculated.

[0079] 3) Channel scale

[0080] Based on the extraction of erosion gullies, the erosion gullies are taken as the research object, the area of ​​the gullies is extracted, and regional analysis is carried out in combination with DEM. ArcGIS software can be used to count the corresponding highest and lowest elevations in the erosion gully unit, and calculate the difference to extract the gully depth.

[0081] Parts of the present invention that are not described in detail are well known to those skilled in the art.

[0082] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. A method for quickly extracting erosion gully morphological indicators, characterized in that: The method comprises: S1. Segment the watershed based on DEM data to obtain several watershed units; S2. Identify gully objects in watershed units; the identification content includes: Based on the orthophoto of the watershed unit, spectral feature analysis is performed to extract the red, green and blue band images and construct the erosion gully interpretation signs; Perform principal component transformation on the extracted image, and select the component with the most three-band information as the first principal component; Selecting a number of texture feature variables and a number of terrain feature variables required for subsequent processing; Based on the selected texture feature variables, multi-scale image segmentation is performed on the first principal component; According to the segmented terrain characteristic variables, identify whether each object in the watershed unit is an erosion gully; Correct the recognition result to obtain the final erosion gully object recognition result; S3, extracting the channel line of the erosion gully; Based on the DEM data of the current watershed unit, the hydrological analysis is carried out, the river network is extracted and then corrected, and the final river network is used as the channel line of the erosion gully; S4, extracting morphological indicators of erosion gullies; According to the erosion gully object recognition result of S2 and the gully line of S3, the morphological index of the erosion gully is extracted.

2. The method for rapidly extracting erosion gully morphology indicators according to claim 1, characterized in that: The first principal component accounts for more than 99% of the information of the three bands of red, green and blue.

3. The method for rapidly extracting erosion gully morphology indicators according to claim 1, characterized in that: The selected texture feature variables include contrast, dissimilarity, homogeneity, angular second moment, entropy, mean, variance, and correlation.

4. The method for rapidly extracting erosion gully morphology indicators according to claim 1, characterized in that: The selected terrain characteristic variables include elevation, slope, sunlight simulation, and runoff accumulation.

5. The method for rapidly extracting erosion gully morphology indicators according to claim 1, characterized in that: The rules for identifying whether each object in a watershed unit is an erosion gully are: Objects with a negative terrain ratio greater than a first threshold are identified as erosion gullies, and objects with a slope less than a second threshold are identified as non-erosion gullies, or Objects in the shaded area are identified as erosion gullies.

6. The method for rapidly extracting erosion gully morphology indicators according to any one of claims 1 to 5, characterized in that: The contents of river network extracted from DEM data through hydrological analysis include: Extract river network based on D8 algorithm and Burn in algorithm; Based on the extraction results, the Strahler classification method is used to classify the extracted river network system, and the buffering analysis is carried out on the rivers at all levels. The difference between the average elevation within the river line and the average elevation in its buffer zone is calculated respectively. The channels with a difference greater than the third threshold are identified as erosion channels, and the final river network system is obtained by combining the extracted river network system.

7. The method for rapidly extracting erosion gully morphology indicators according to claim 6, characterized in that: The water catchment threshold for river network extraction based on the D8 algorithm and Burn in algorithm is set to 800.

8. The method for rapidly extracting erosion gully morphology indicators according to claim 1, characterized in that: Erosion gully morphological indicators are extracted from three scales: watershed, gully and channel.

9. The method for rapidly extracting erosion gully morphology indicators according to claim 8, characterized in that: The morphological indicators at the watershed scale include: gully area, gully length, gully density and gully fragmentation; The morphological indicators of gully scale include: the main gully to branch gully ratio and the longitudinal gradient of the main gully; the morphological indicators of gully scale include: The morphological indicators of channel scale include: channel area, channel length and channel depth.

10. A rapid extraction system for erosion gully morphology indicators, characterized in that: A method for quickly extracting erosion gully morphology indicators as claimed in any one of claims 1 to 9 can be implemented; the system comprises: The watershed segmentation module is used to segment the watershed based on DEM data to obtain several watershed units; The erosion gully object recognition module is used to identify erosion gully objects in the watershed unit; the recognition content includes: based on the orthophoto of the watershed unit, spectral feature analysis is performed to extract the red, green and blue three-band images and construct the erosion gully interpretation mark; the extracted image is transformed into a principal component, and the component with the most three-band information is selected as the first principal component; a number of texture feature variables and a number of terrain feature variables required for subsequent processing are selected; based on the selected texture feature variables, the first principal component is segmented at multiple scales; according to the segmented terrain feature variables, each object in the watershed unit is identified as an erosion gully; the recognition result is corrected to obtain the final erosion gully object recognition result; The channel line extraction module is used to extract the channel line of the erosion gully. The specific contents include: performing hydrological analysis based on the DEM data of the current watershed unit, extracting the river network and then correcting it, and using the final river network as the channel line of the erosion gully; The morphological index extraction module is used to extract the morphological index of the erosion gully according to the erosion gully object recognition result and the channel line extraction result.

Images