Method, device, system and medium for classifying green-blue-gray infrastructure

1 . A method for classifying a green-blue-gray infrastructure, the method comprising:

obtaining a multispectral photo corresponding to a target area;

obtaining an image set of a target and a color orthophoto based on the multispectral photo;

obtaining a sample file based on the color orthophoto, wherein the sample file comprises a training sample and a validation sample; and

obtaining a classification result of the green-blue-gray infrastructure corresponding to the target area based on the image set of the target and the sample file;

overlapping a mesh image with the color orthophoto to obtain a color mesh orthophoto;

identifying a mesh-point attribute corresponding to each mesh point in the color mesh orthophoto;

obtaining a training sample based on the mesh-point attribute; and

deflecting the mesh image of the color mesh orthophoto to obtain the validation sample.

2 . The method for classifying the green-blue-gray infrastructure according to claim 1 , wherein obtaining the image set of the target and the color orthophoto based on the multispectral photo further comprises:

performing a two-dimensional rebuilding operation on the multispectral photo; and

obtaining the image set of the target and the color orthophoto based on the multispectral photo after the two-dimensional rebuilding operation and a preset resolution.

3 . The method of claim 2 , wherein the preset resolution is between 6 cm and 10 cm.

4 . The method of claim 3 , wherein the preset resolution is substantially 6 cm.

5 . The method for classifying the green-blue-gray infrastructure according to claim 1 , wherein obtaining the sample file based on the color orthophoto further comprises:

generating the mesh image based on a preset mesh spacing.

6 . The method for classifying the green-blue-gray infrastructure according to claim 5 , wherein obtaining the classification result of the green-blue-gray infrastructure corresponding to the target area based on the image set of the target and the sample file further comprises:

performing a partition on the image set of the target based on spectral similarity to obtain a shape object;

calculating and obtaining a pre-classification result of the green-blue-gray infrastructure based on the training sample of the sample file and a light-waveband image value of the corresponding shape object;

performing an accuracy evaluation on the pre-classification result of the green-blue-gray infrastructure based on the validation sample of the sample file to obtain an accuracy-evaluation result; and

determining the classification result of the green-blue-gray infrastructure based on the accuracy-evaluation result.

7 . The method for classifying the green-blue-gray infrastructure according to claim 6 , wherein determining the classification result of the green-blue-gray infrastructure based on the accuracy-evaluation result further comprises:

comparing the accuracy-evaluation result with a preset accuracy to obtain a comparison result;

if the comparison result is that the accuracy-evaluation result is not less than the preset accuracy, determining the pre-classification result of the green-blue-gray infrastructure as the classification result of the green-blue-gray infrastructure; and

if the comparison result is that the accuracy-evaluation result is less than the preset accuracy, presetting and processing the mesh image and the image set of the target to obtain an optimal mesh image and an optimal image set of the target, and overlapping the mesh image with the color orthophoto to obtain the color mesh orthophoto.

8 . The method for classifying the green-blue-gray infrastructure according to claim 7 , wherein presetting and processing the mesh image and the image set of the target to obtain the optimal mesh image and the optimal image set of the target further comprises:

performing a zoom-out operation on a mesh spacing corresponding to the mesh image to obtain mesh images with a first preset number;

performing a first preset operation on each mesh image respectively to obtain the optimal mesh image;

performing an increase or decrease operation on the number of images in the image set of the target to obtain image sets of the target with a second preset number; and

performing a second preset operation on each image set of the target based on the optimal mesh image to obtain the optimal image set of the target.

9 . The method of claim 7 , wherein the preset accuracy is set to 0.8 based on actual situations.

10 . The method for classifying the green-blue-gray infrastructure according to claim 1 , wherein after obtaining the sample file based on the color orthophoto, and obtaining the classification result of the green-blue-gray infrastructure corresponding to the target area based on the image set of the target and the sample file, the method further comprises:

generating a classification map corresponding to the target area based on the classification result of the green-blue-gray infrastructure; and

providing a rainwater-utilization analysis data corresponding to the target area based on the classification map.

11 . A system for classifying the green-blue-gray infrastructure, wherein the system comprises a memory, a processor, and a program for classifying the green-blue-gray infrastructure stored on the memory and operable on the processor; wherein the steps of the method for classifying the green-blue-gray infrastructure according to claim 1 are implemented when the program is executed by the processor.

12 . A non-transitory computer-readable storage medium, on which a program for classifying a green-blue-gray infrastructure is stored; wherein the steps of the method for classifying the green-blue-gray infrastructure according to claim 1 are implemented when the program is executed by the processor.

13 . The method of claim 1 , wherein deflecting the mesh image of the color mesh orthophoto to obtain the validation sample, deflects an entire mesh in the color mesh orthophoto downward by one preset mesh spacing.

14 . A device for classifying a green-blue-gray infrastructure, the device comprising:

an obtaining module configured to:

obtain a multispectral photo corresponding to a target area, and

obtain an image set of a target, and a color orthophoto based on the multispectral photo; and

a classifying module configured to:

obtain a sample file based on the color orthophoto, wherein the sample file comprises a training sample and a validation sample,

obtain a classification result of the green-blue-gray infrastructure corresponding to the target area based on the image set of the target and the sample file,

overlap a mesh image with the color orthophoto to obtain a color mesh orthophoto,

identify a mesh-point attribute corresponding to each mesh point in the color mesh orthophoto,

obtain a training sample based on the mesh-point attribute, and

deflect the mesh image of the color mesh orthophoto to obtain the validation sample.