CN111695530A - River water replenishing effect intelligent monitoring and evaluation method based on high-resolution remote sensing - Google Patents

Abstract

The invention discloses a river water replenishing effect intelligent monitoring and evaluating method based on high-resolution remote sensing, which utilizes multi-source high-resolution satellite cooperative observation, adopts a preprocessing process to obtain a river region high-resolution remote sensing image set with space-time consistency, further constructs high-precision training sample points and a deep neural network model to intelligently extract a water surface range of multi-temporal river water replenishing, and provides quantitative evaluation indexes for river water replenishing effects. The intelligent monitoring and evaluation method for the river water replenishing effect is high in applicability, can be suitable for monitoring and evaluating the river water replenishing effect of different basin scales, and can be expanded to business applications such as strong supervision of rivers and lakes.

Description

An intelligent monitoring and evaluation method for river water replenishment effect based on high score remote sensing

technical field

The invention belongs to the technical field of intelligent identification of remote sensing images, in particular to the design of an intelligent monitoring and evaluation method for river water replenishment effects based on high-resolution remote sensing.

Background technique

Rivers are an important link connecting different geographical units and ecological units, and also an important carrier of material flow and information flow. Affected by factors such as climate change and human social development, river water resources are over-exploited and utilized, and it is difficult to be replenished in time, resulting in a large number of rivers drying up and stopping, seriously affecting the ecological health of river water, connectivity of rivers and lakes, and groundwater recharge. Inevitably caused the deterioration of the river and its surrounding ecological environment.

Cross-basin and cross-regional water replenishment is the most direct and effective solution to the ecological and environmental problems caused by the drying up of rivers. By taking measures such as river remediation and river replenishment, the river surface can be restored to continuity, which is helpful for the gradual restoration of river water. Lake ecological function. At present, river water replenishment has been piloted in some areas, and the river has restored connectivity and achieved certain results. However, a set of intelligent monitoring and evaluation methods for the effect of river water replenishment has not yet been formed. Traditional hydrological monitoring stations are an important means to monitor river water resources, but The limited number of hydrological monitoring stations makes it difficult to reflect the overall water replenishment situation of the entire river. Although manual inspections are highly accurate, they are time-consuming and labor-intensive and have a long period of time, making it difficult to monitor and evaluate the effect of water replenishment in large-scale rivers.

Satellite remote sensing can obtain surface information macroscopically, accurately and truly, and is an important means to carry out large-scale surveys of water resources, water environment and water ecology. With the implementation of the high-resolution earth observation system major project in my country, a number of high-resolution satellites such as Gaofen-1 to Gaofen-7 satellites have been successfully launched, and my country's high-resolution multi-source satellite collaborative observation system has been initially formed. Monitoring provides a rich, timely and high-quality remote sensing data source, which is helpful for dynamic monitoring and evaluation of the water replenishment effect of open rivers.

SUMMARY OF THE INVENTION

The purpose of the present invention is to solve the problem of lack of a set of monitoring and evaluation methods for river water replenishment effect in the prior art, and proposes an intelligent monitoring and evaluation method for river water replenishment effect based on high score remote sensing.

The technical scheme of the present invention is: an intelligent monitoring and evaluation method for river water replenishment effect based on high score remote sensing, comprising the following steps:

S1. For the water replenishment range of the river to be monitored, collect multi-source high-resolution remote sensing data before and after water replenishment and during the water replenishment process.

S2. Perform spatiotemporal consistency preprocessing on multi-source high-resolution remote sensing data to obtain high-resolution remote sensing images of river areas.

S3. Perform intelligent monitoring of river water replenishment according to high-resolution remote sensing images of river regions, and form a multi-temporal river water surface range data set.

S4. Evaluate the effect of river water replenishment according to the multi-temporal river water surface range data set.

Further, the selection criteria of multi-source high-scoring remote sensing data in step S1 are:

(1) Select a cloud-free or less cloud-covered image, and the river area is cloud-free;

(2) There are no missing, noise and abnormal pixels in the image;

(3) The image has no obvious aerosol coverage;

(4) There is no ice and snow in the image;

(5) There is no rainfall before and after the imaging date.

Further, step S2 includes the following sub-steps:

S21. Collect DEM data and Landsat 8 data of the image coverage area in the multi-source high-resolution remote sensing data.

S22. Use the Landsat 8 data as a reference image to collect high-precision control points.

S23. Combine the DEM data of the image coverage area and the collected high-precision control points, optimize the rational function model coefficients of the image, and use the rational function model to perform geometric correction on the panchromatic images and multispectral images in the high-resolution remote sensing data.

S24, using the Pansharp fusion method to fuse the geometrically corrected panchromatic image and the multispectral image to obtain a high spatial resolution multispectral image.

S25 , using the image histogram matching method to adjust the tone of the multispectral image with high spatial resolution, and constructing a splicing line according to the image feature points to realize regional image mosaicking.

S26. Construct a buffer zone according to the river centerline vector and the river width, and combine the mosaic image and the buffer zone to crop the river range in the image to obtain a high-score remote sensing image of the river area.

Further, the expression of the rational function model in step S23 is:

Among them, (L _n , _Sn ) represents the normalized coordinates of the pixel row and column coordinates (L, S) in the image after translation and scaling, (X _n , Y _n , Z _n ) represents the ground coordinates (X, Y, Z) Normalized coordinates after translation and scaling, P _l (·) represents the numerator polynomial of the row pixel rational function, Q _l (·) represents the denominator polynomial of the row pixel rational function, and P _s (·) represents the column pixel rational function The numerator polynomial of , Q _s (·) represents the denominator polynomial of the column pixel rational function.

Further, step S3 includes the following sub-steps:

S31. According to the high-resolution remote sensing image of the river area, the artificial interest area method is used to mark the high-precision training samples of water body and non-water body respectively on the pixel layer.

S32. Input the high-precision training samples of the water body and the non-water body into the deep neural network model, and perform optimization training on the model parameters, so as to extract the water surface range of the river replenishment.

S33. Dynamically monitor the water surface range of river replenishment, and form a multi-temporal river water surface range data set.

Further, step S32 includes the following sub-steps:

S321. Construct training sample feature vectors according to high-precision samples of water bodies and non-water bodies

表示水体训练样本特征向量，

表示非水体训练样本特征向量，B_n表示第n个特征波段，n＝1,2,...,N，N表示遥感影像特征波段总数。in

represents the feature vector of water body training samples,

Represents the feature vector of non-water training samples, B _n represents the nth feature band, n=1,2,...,N, N represents the total number of remote sensing image feature bands.

输入到深度神经网络模型中，通过多个隐藏层神经元迭代计算净函数G(x)：S322. Set the training sample feature vector

Input into the deep neural network model, and iteratively calculate the net function G(x) through multiple hidden layer neurons:

where ω _ij represents the random initial weight value of the i-th type of neuron in the j-th layer, x _i represents the i-th type of feature, b _j represents the deviation of the j-th layer of neurons, and k represents the total number of features.

S323. Through repeated iterations, a classification discriminant function P(x) is constructed to determine whether each pixel is a water body or a non-water body type:

Among them, P(x)=1 indicates that the pixel category is a water body, that is, the water surface range of river replenishment, and P(x)=0 indicates that the pixel category is a non-water body, that is, a non-water surface range.

Further, step S4 includes the following sub-steps:

S41. Calculate the change value of river water surface area according to the multi-temporal river water surface range data set

Among them, S _T1 represents the river water surface area before water replenishment, and S _T2 represents the river water surface area after water replenishment.

S42. Calculate the change value of river water surface and water width according to the multi-temporal river water surface range data set

Among them, D _T1 represents the water surface water width of the river before water replenishment, and D _T2 represents the water surface water width of the river after water replenishment.

S43. Calculate the change value of the length of the dry river according to the multi-temporal river water surface range data set

where L _T1 is the length of the dry river before water replenishment, and L _T2 is the length of the dry river after water replenishment.

大于0、河流水面水宽变化值

大于0且干涸河流长度变化值

小于0，则河流补水取得成效，否则河流补水尚未取得成效。S44. If the change value of the river water surface area

Greater than 0, the change value of river water surface water width

Greater than 0 and the value of the change in the length of the dry river

If it is less than 0, the river water replenishment is effective, otherwise the river water replenishment has not been effective.

The beneficial effects of the present invention are as follows: the present invention utilizes multi-source high-resolution satellites for collaborative observation, adopts a preprocessing process to obtain a high-resolution remote sensing image set of a river region that is consistent in time and space, and then constructs high-precision training sample points and a deep neural network model for intelligent extraction for a long time. According to the water surface range of river water replenishment, a quantitative evaluation index for river water replenishment effect is proposed. The present invention is an intelligent monitoring and evaluation method for river water replenishment effect with strong applicability, which can be applied to monitoring and evaluation of river water replenishment effect in different river basin scales, and can also be extended to business applications such as intensive supervision of rivers and lakes.

Description of drawings

FIG. 1 shows a flowchart of an intelligent monitoring and evaluation method for river water replenishment effect based on high-resolution remote sensing provided by an embodiment of the present invention.

FIG. 2 is a flowchart of a method for intelligent monitoring and evaluation of river water replenishment effects based on high-resolution remote sensing provided by an embodiment of the present invention.

FIG. 3 is a schematic diagram showing a result of monitoring the water surface range before river water replenishment provided by an embodiment of the present invention.

FIG. 4 is a schematic diagram showing a result of monitoring the water surface range of a river after water replenishment provided by an embodiment of the present invention.

Detailed ways

Exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be understood that the embodiments shown and described in the accompanying drawings are exemplary only, and are intended to illustrate the principles and spirit of the present invention, and not to limit the scope of the present invention.

The embodiment of the present invention provides an intelligent monitoring and evaluation method for river water replenishment effect based on high-resolution remote sensing, as shown in FIG. 1 to FIG. 2 , including the following steps S1 to S4:

S1. For the water replenishment range of the river to be monitored, collect multi-source high-resolution remote sensing data before and after water replenishment and during the water replenishment process.

In the embodiment of the present invention, the multi-source high-resolution remote sensing data are high-resolution optical satellite remote sensing data such as Gaofen-1, Gaofen-2, and Gaofen-6 that are queried and downloaded from the China Resource Satellite Center, and the downloaded data is 1A product , has been processed by systems such as data analysis, uniform radiation correction, denoising, transfer function compensation CCD splicing and band registration.

In the embodiment of the present invention, the selection criteria of the multi-source high-scoring remote sensing data in step S1 are:

(1) Select a cloud-free or less cloud-covered image, and the river area is cloud-free;

(2) There are no missing, noise and abnormal pixels in the image;

(3) The image has no obvious aerosol coverage;

(4) There is no ice and snow in the image;

(5) There is no rainfall before and after the imaging date.

S2. Perform spatiotemporal consistency preprocessing on multi-source high-resolution remote sensing data to obtain high-resolution remote sensing images of river areas.

Since the high-resolution remote sensing data collected in step S1 has not been geometrically finely calibrated, it needs to be preprocessed for spatial and temporal consistency to ensure the spatial accuracy of multi-temporal high-resolution remote sensing images.

Step S2 includes the following sub-steps S21 to S26:

S21. Collect DEM (Digital Elevation Model) data and Landsat8 data of the image coverage area in the multi-source high-resolution remote sensing data.

In the embodiment of the present invention, DEM data is used for subsequent correction of positioning errors caused by terrain, and Landsat 8 data selects Landsat 8 panchromatic 15m spatial resolution images for subsequent collection of high-precision control points as reference images.

S22. Use the Landsat 8 data as a reference image to collect high-precision control points.

In the embodiment of the present invention, the selected positions of control points are generally located at road intersections, and are evenly distributed on the image. The number of image control points collected per scene in plain areas is 15-20, and the number of image control points collected per scene in mountainous areas is 25-30 , and the error of the control point in the X and Y directions is not higher than 1 pixel.

S23. Combine the DEM data of the image coverage area and the collected high-precision control points, optimize the rational function model coefficients of the image, and use the rational function model to perform geometric correction on the panchromatic images and multispectral images in the high-resolution remote sensing data.

The rational function model is a general geometric correction model for high-resolution images. The model expresses the pixel coordinates as the ratio of rational polynomials with the corresponding ground point spatial coordinates as independent variables. By introducing high-precision control points, the rational function is optimized. coefficient to achieve high-precision image geometric correction. The expression of the rational function model is:

Among them, (L _n , _Sn ) represents the normalized coordinates of the pixel row and column coordinates (L, S) in the image after translation and scaling, (X _n , Y _n , Z _n ) represents the ground coordinates (X, Y, Z) Normalized coordinates after translation and scaling, P _l (·) represents the numerator polynomial of the row pixel rational function, Q _l (·) represents the denominator polynomial of the row pixel rational function, and P _s (·) represents the column pixel rational function The numerator polynomial of , Q _s (·) represents the denominator polynomial of the column pixel rational function.

S24, using the Pansharp fusion method to fuse the geometrically corrected panchromatic image and the multispectral image to obtain a high spatial resolution multispectral image.

After geometric correction, the panchromatic image of high-resolution remote sensing data has higher spatial resolution, while the multispectral image has rich spectral features. The Pansharp fusion method can be used to obtain multispectral fusion images with high spatial resolution. It has better preservation effect in image information, details and spectrum.

S25. Due to inconsistent image imaging conditions, the fused multispectral image will have a large color difference. Therefore, it is necessary to uniformize the image so that the overall color tone can be kept consistent. In the embodiment of the present invention, the image histogram matching method is used to adjust the tone of the multi-spectral image with high spatial resolution, and a splicing line is constructed according to the image feature points to realize regional image mosaic.

S26. Construct a buffer zone according to the river centerline vector and the river width, and combine the mosaic image and the buffer zone to crop the river range in the image to obtain a high-score remote sensing image of the river area.

S3. Perform intelligent monitoring of river water replenishment according to high-resolution remote sensing images of river regions, and form a multi-temporal river water surface range data set.

Step S3 includes the following sub-steps S31 to S33:

S31. According to the high-resolution remote sensing image of the river area, the artificial interest area method is used to mark the high-precision training samples of water body and non-water body respectively on the pixel layer.

In the embodiment of the present invention, the high-precision water body sample includes all types of surface water bodies, including lakes, rivers, pools, sediment water bodies, and the like. Non-water high-precision samples include farmland, forests, cities, bare land, shrubs, hill shadows, cloud shadows, and other objects.

S32. Input the high-precision training samples of the water body and the non-water body into the deep neural network model, and perform optimization training on the model parameters, so as to extract the water surface range of the river replenishment.

A deep neural network algorithm is a neural network model containing multiple hidden layers, and all neurons are fully connected, which is a type of machine learning supervised classification algorithm.

Step S32 includes the following sub-steps S321-S323:

S321. Construct training sample feature vectors according to high-precision samples of water bodies and non-water bodies

表示水体训练样本特征向量，

表示非水体训练样本特征向量，B_n表示第n个特征波段，n＝1,2,...,N，N表示遥感影像特征波段总数。in

represents the feature vector of water body training samples,

Represents the feature vector of non-water training samples, B _n represents the nth feature band, n=1,2,...,N, N represents the total number of remote sensing image feature bands.

In the embodiment of the present invention, each image input band is used as a feature. For example, Gaofen-1 and Gaofen-2 have four feature bands. When the Gaofen-1 or Gaofen-2 Gaofen satellite remote sensing data is selected, N = 4, and Gaofen-6 has eight characteristic bands, when the Gaofen-6 satellite remote sensing data is selected, N=8.

输入到深度神经网络模型中，通过多个隐藏层神经元迭代计算净函数G(x)：S322. Set the training sample feature vector

Input into the deep neural network model, and iteratively calculate the net function G(x) through multiple hidden layer neurons:

where ω _ij represents the random initial weight value of the i-th type of neuron in the j-th layer, x _i represents the i-th type of feature, b _j represents the deviation of the j-th layer of neurons, and k represents the total number of features.

S323. Through repeated iterations, a classification discriminant function P(x) is constructed to determine whether each pixel is a water body or a non-water body type:

Among them, P(x)=1 indicates that the pixel category is a water body, that is, the water surface range of river replenishment, and P(x)=0 indicates that the pixel category is a non-water body, that is, a non-water surface range.

S33. Dynamically monitor the water surface range of river replenishment, and form a multi-temporal river water surface range data set.

In the embodiment of the present invention, the monitoring result of the water surface range before the river replenishment is shown in FIG. 3 , and the monitoring result of the water surface range after the river replenishment is shown in FIG. 4 . According to FIG. 3 and FIG. Monitoring results.

S4. Evaluate the effect of river water replenishment according to the multi-temporal river water surface range data set.

Step S4 includes the following sub-steps S41 to S44:

S41. Calculate the change value of river water surface area according to the multi-temporal river water surface range data set

Among them, S _T1 represents the river water surface area before water replenishment, and S _T2 represents the river water surface area after water replenishment. Both S _T1 and S _T2 can be obtained from the multi-temporal river water surface range data set.

S42. Calculate the change value of river water surface and water width according to the multi-temporal river water surface range data set

Among them, D _T1 represents the water surface water width of the river before water replenishment, and D _T2 represents the water surface water width of the river after water replenishment. Both D _T1 and D _T2 can be obtained from the multi-temporal river water surface range data set.

S43. Calculate the change value of the length of the dry river according to the multi-temporal river water surface range data set

Among them, L _T1 represents the length of the dry river before water replenishment, and L _T2 represents the length of the dry river after water replenishment. Both L _T1 and L _T2 can be obtained from the multi-temporal river water surface range data set.

大于0(即河流水面面积增大)、河流水面水宽变化值

大于0(即河流水面增宽)且干涸河流长度变化值

小于0(即干涸河流长度减少)，则河流补水取得成效，否则河流补水尚未取得成效。S44. If the change value of the river water surface area

Greater than 0 (that is, the river water surface area increases), the change value of the river water surface water width

Greater than 0 (that is, the river surface widens) and the change in the length of the dry river

If it is less than 0 (that is, the length of the dry river is reduced), the river water replenishment has been effective, otherwise the river water replenishment has not been effective.

Those of ordinary skill in the art will appreciate that the embodiments described herein are intended to assist readers in understanding the principles of the present invention, and it should be understood that the scope of protection of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations without departing from the essence of the present invention according to the technical teaching disclosed in the present invention, and these modifications and combinations still fall within the protection scope of the present invention.

Claims (7)

1. an intelligent monitoring and evaluation method for river water replenishment effect based on high score remote sensing, is characterized in that, comprises the following steps: S1. For the water replenishment range of the river to be monitored, collect multi-source high-resolution remote sensing data before and after water replenishment and during the water replenishment process; S2. Perform spatiotemporal consistency preprocessing on multi-source high-resolution remote sensing data to obtain high-resolution remote sensing images of river areas; S3. Perform intelligent monitoring of river water replenishment according to high-resolution remote sensing images of river areas, and form a multi-temporal river water surface range data set; S4. Evaluate the effect of river water replenishment according to the multi-temporal river water surface range data set. 2. The intelligent monitoring and evaluation method of river water replenishment effect according to claim 1, is characterized in that, in described step S1, the selection criterion of multi-source high score remote sensing data is: (1) Select a cloud-free or less cloud-covered image, and the river area is cloud-free; (2) There are no missing, noise and abnormal pixels in the image; (3) The image has no obvious aerosol coverage; (4) There is no ice or snow in the image; (5) There is no rainfall before and after the imaging date. 3. The intelligent monitoring and evaluation method of river water replenishment effect according to claim 1, is characterized in that, described step S2 comprises the following sub-steps: S21. Collect DEM data and Landsat 8 data of the image coverage area in the multi-source high-resolution remote sensing data; S22. Use Landsat 8 data as a reference image to collect high-precision control points; S23. Combine the DEM data of the image coverage area and the collected high-precision control points, optimize the rational function model coefficients of the image, and use the rational function model to perform geometric correction on the panchromatic images and multispectral images in the high-resolution remote sensing data; S24, using the Pansharp fusion method to fuse the geometrically corrected panchromatic image and the multispectral image to obtain a high spatial resolution multispectral image; S25, using the image histogram matching method to adjust the tone of the high spatial resolution multispectral image, and constructing a splicing line according to the image feature points to realize regional image mosaicking; S26. Construct a buffer zone according to the river centerline vector and the river width, and combine the mosaic image and the buffer zone to crop the river range in the image to obtain a high-score remote sensing image of the river area. 4. the intelligent monitoring and evaluation method of river water replenishment effect according to claim 3, is characterized in that, the expression of the rational function model in described step S23 is:

Among them, (L _n , _Sn ) represents the normalized coordinates of the pixel row and column coordinates (L, S) in the image after translation and scaling, (X _n , Y _n , Z _n ) represents the ground coordinates (X, Y, Z) Normalized coordinates after translation and scaling, P _l (·) represents the numerator polynomial of the row pixel rational function, Q _l (·) represents the denominator polynomial of the row pixel rational function, and P _s (·) represents the column pixel rational function The numerator polynomial of , Q _s (·) represents the denominator polynomial of the column pixel rational function. 5. The intelligent monitoring and evaluation method of river water replenishment effect according to claim 1, is characterized in that, described step S3 comprises the following sub-steps: S31. According to the high-resolution remote sensing image of the river area, the artificial interest area method is used to mark the high-precision training samples of water body and non-water body respectively on the pixel layer; S32, input the high-precision training samples of water body and non-water body into the deep neural network model, and optimize the training parameters of the model, so as to extract the water surface range of river replenishment; S33. Dynamically monitor the water surface range of river replenishment, and form a multi-temporal river water surface range data set. 6. The intelligent monitoring and evaluation method for river water replenishment effect according to claim 5, wherein the step S32 comprises the following sub-steps: S321. Construct training sample feature vectors according to high-precision samples of water bodies and non-water bodies

in

represents the feature vector of water body training samples,

Represents the non-water body training sample feature vector, B _n represents the nth feature band, n=1,2,...,N, N represents the total number of remote sensing image feature bands; S322. Set the training sample feature vector

Input into the deep neural network model, and iteratively calculate the net function G(x) through multiple hidden layer neurons:

where ω _ij represents the random initial weight value of the i-th type of neuron in the j-th layer, x _i represents the i-th type of feature, b _j represents the deviation of the j-th layer of neurons, and k represents the total number of features; S323. Through repeated iterations, a classification discriminant function P(x) is constructed to determine whether each pixel is a water body or a non-water body type:

Among them, P(x)=1 indicates that the pixel category is a water body, that is, the water surface range of river replenishment, and P(x)=0 indicates that the pixel category is a non-water body, that is, a non-water surface range. 7. The intelligent monitoring and evaluation method for river water replenishment effect according to claim 1, wherein the step S4 comprises the following sub-steps: S41. Calculate the change value of river water surface area according to the multi-temporal river water surface range data set

Among them, S _T1 represents the river water surface area before water replenishment, and S _T2 represents the river water surface area after water replenishment; S42. Calculate the change value of river water surface and water width according to the multi-temporal river water surface range data set

Among them, D _T1 represents the water surface water width of the river before water replenishment, and D _T2 represents the water surface water width of the river after water replenishment; S43. Calculate the change value of the length of the dry river according to the multi-temporal river water surface range data set

where L _T1 represents the length of the dry river before water replenishment, and L _T2 represents the length of the dry river after water replenishment; S44. If the change value of the river water surface area

Greater than 0, the change value of river water surface water width

Greater than 0 and the value of the change in the length of the dry river

If it is less than 0, the river water replenishment is effective, otherwise the river water replenishment has not been effective.

Images