A remote sensing image supervised classification method based on decision tree correction

Abstract

The application discloses a remote sensing image supervised classification method based on decision tree correction, which comprises the following steps: acquiring satellite images to be classified and reference images, the reference images being satellite images of the same sensor covering a research area and having verifiable ground object types; pre-processing all the images and converting image data DN values into ground surface radiation values; cutting the processed images to acquire image data of a required research area; synthesizing new synthetic image data respectively from pre-processed band data of each image and various parameters calculated based on the band data; drawing a region of interest on image data of a reference synthetic image and making a classification decision tree of the image through machine learning; classifying other synthetic image data to be classified based on the classification decision tree and drawing new regions of interest on the synthetic images according to classification results; and reclassifying ground objects of the synthetic image data by using a support vector machine. The application improves classification accuracy and speed.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of remote sensing image ground object classification, and particularly relates to a remote sensing image supervised classification method based on decision tree correction. BACKGROUND

[0002] Since 1972, Landsat satellites have provided relatively accurate data for long-term monitoring and analysis of forest coverage, urban expansion, farmland changes, water resource distribution, etc. Among them, the TM sensor carried by Landsat5 has seven bands (wavelength range from 0.45 to 2.35 μm, resolution from 30 meters to more than 120 meters), which has provided nearly 29 years of earth imaging data since March 1, 1984. Landsat8 has been in operation since its launch on February 11, 2013, providing higher resolution image data. However, there are problems such as spectral differences, radiation problems, and geometric correction errors between different types of Landsat series satellites, so when comparing results, complex correction and processing work is often required. And the acquisition of Landsat images is limited by weather conditions, such as cloud cover, which can cause data loss or poor quality in some areas or time periods, affecting the accuracy of ground object classification. Further, the sensors of Landsat satellites have limitations in spatial and temporal resolution, with lower resolution, which cannot be used for detailed study of ground object classes in space.

[0003] The commonly used classification methods at present include decision tree classification, support vector machine classification, etc. Decision tree displays the classification process through tree structure, can process data of various types and different measurement scales, has the advantages of fast establishment speed and small calculation amount, and is suitable for complex and variable remote sensing image data. However, the classification effect of this method depends largely on the selection of features, and the relationship between classification accuracy and model complexity needs to be balanced in the classification process. Support vector machine classification performs well in handling nonlinear classification problems, and can better classify data with complex features. However, this method needs to calculate the distance between each sample point and the decision boundary in the training process, so the training time is longer, and a large memory space is required when processing large-scale data sets. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides a remote sensing image supervised classification method based on decision tree correction, which identifies different ground object types by using decision trees, provides reference for division of interest areas, and then combines artificial correction and supervised classification to realize classification processing of remote sensing images of different resolutions and different band numbers under long time series with a unified standard, while significantly improving the classification accuracy and the calculation and classification speed.

[0005] The object of the present application is achieved by the technical solutions as follows:

[0006] A remote sensing image supervised classification method based on decision tree correction, comprising the following steps:

[0007] S1: obtaining satellite images to be classified and reference images, preprocessing them, and converting image data DN values to ground surface radiation values; performing image cropping on the processed satellite images to obtain image data of the required research area; the reference images are satellite images obtained by the same sensor covering the research area and having verifiable ground object types;

[0008] S2: combining the band data after S1 preprocessing with the results of various parameters calculated based on the band data to synthesize new synthetic image data, obtaining satellite synthetic images R1, R2,..., R n , to be classified and a reference synthetic image R;

[0009] S3: drawing a region of interest on the image data of the reference synthetic image R with verifiable ground object types, and making a classification decision tree for the image through machine learning;

[0010] S4: classifying other synthetic image data R1, R2,..., R n to be classified based on the classification decision tree obtained in S3, and drawing new regions of interest on these synthetic images based on the classification results;

[0011] S5: reclassifying the ground objects of these synthetic image data based on the newly drawn regions of interest using support vector machine supervised classification.

[0012] Further, in S2, the various parameters calculated based on the band data include normalized vegetation index NDVI, normalized difference vegetation index EVI, normalized difference water body index NDWI, water body chlorophyll index Chl a, and water body suspended sediment index SSC.

[0013] Further, the specific calculation formulas of the various parameters calculated based on the band data are as follows:

[0014]

[0015]

[0016]

[0017]

[0018]

[0019] SSC0.7874*(Red / Green)-0.1376*(Red / Green)+0.2098

[0020] Among them, Red is the surface reflectivity of the red light band, with a wavelength range of 0.64-0.67μm; Green is the surface reflectivity of the green light band, with a wavelength range of 0.53-0.59μm; Blue is the surface reflectivity of the blue light band, with a wavelength range of 0.45-0.51μm; Nir is the surface reflectivity of the near-infrared band, with a wavelength range of 0.85-0.88μm; Coastal is the surface reflectivity of the coastal band, with a wavelength range of 0.43-0.45μm, and a wavelength range of 0.43-0.45μm; M is a process quantity.

[0021] Furthermore, the S3 specifically includes the following sub-steps:

[0022] S3.1: Obtain the main image features of different land feature categories through visual interpretation, field research, high-resolution image reference, and existing classification results, and determine the classification type of land features based on actual needs;

[0023] S3.2: Select a reference composite image R and, based on the spectral characteristics and visual interpretation results of various typical landforms in the study area, draw regions of interest corresponding to different landform types;

[0024] S3.3: Based on the region of interest drawn in S3.2, perform machine learning on the reference synthetic image R and obtain a decision tree that supports the classification of the image.

[0025] Furthermore, the S4 specifically includes the following sub-steps:

[0026] S4.1: Use the classification decision tree obtained in S3.3 to classify the satellite synthetic images R1, R2, ..., R n Classify each one;

[0027] S4.2: For the satellite synthetic images R1, R2,…, R n For each image in the dataset, the decision tree classification results are compared with the spectral characteristics and texture of the image, the areas where the decision tree misjudged the image are marked, and then the region of interest of each image is remade.

[0028] Furthermore, in S1, the preprocessing of the satellite image to be classified and the reference image includes radiometric calibration, atmospheric correction, and geometric correction.

[0029] Compared with the existing technology, the beneficial effects of the present invention are as follows:

[0030] The present invention classifies objects in remote sensing images using synthetic image data, expanding the data source for classification conditions and increasing the computational speed of the classification process. The region of interest is then corrected based on a decision tree, significantly reducing the errors easily generated when classifying low- and medium-resolution images. It can also reduce misjudgments caused by radiation errors due to factors such as solar altitude and terrain, and reduce errors caused by subjective visual interpretation by classifiers, thereby increasing accuracy as much as possible. Furthermore, because support vector machines require a large amount of memory and computational resources and are difficult to modify samples, pre-classification using a decision tree provides a reference, and supervised classification of other images based on this decision tree can save computational speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 The present invention is a flowchart of a remote sensing image supervised classification method based on decision tree correction.

[0032] Figure 2 This is the land classification result map of Hangzhou Bay after reasonable and practical processing by the present invention. DETAILED DESCRIPTION

[0033] The present invention will be described in detail below with reference to the accompanying drawings and preferred embodiments, and the purpose and effects of the present invention will become more apparent. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0034] like Figure 1 As shown, the remote sensing image supervised classification method based on decision tree correction of the present invention includes the following steps:

[0035] S1: Obtain the satellite image to be classified and the reference image, preprocess them, and convert the DN value of the image data into the surface radiation value; crop the processed satellite image to obtain the image data of the required study area.

[0036] One or more satellite image data to be classified are represented as r1, r2,…,r n , the reference image is a satellite image obtained by the same sensor that covers the study area and the type of the ground feature can be verified, denoted as r.

[0037] In this embodiment, the study area is the Hangzhou Bay coastal area, involving the land and water bodies of Hangzhou Bay. The research objective is to observe the classification of land use types in the Hangzhou Bay area. The research data is Landsat satellite image data from 2015. The satellite images that can be verified for the types of land features covered by the same sensor in the study area are Landsat satellite image data from 2021.

[0038] The preprocessing operations specifically include radiometric calibration, atmospheric correction, geometric correction, etc. All image preprocessing operations are performed based on ENVI software, and the algorithm used in the atmospheric correction process is the FLAASH atmospheric correction algorithm.

[0039] S2: Synthesize the new synthetic image data by combining the band data preprocessed in S1 with the various parameter results calculated based on the band data. S2 specifically includes the following sub-steps:

[0040] S2.1: Calculate the various index parameter results of each image for auxiliary classification using the surface radiation value of each band processed in S1;

[0041] In this embodiment, since the study area includes land and water, auxiliary classifications such as the normalized vegetation index NDVI, the normalized difference vegetation index EVI, the normalized difference water index NDWI, the water chlorophyll index Chl a, and the water sediment suspended solids index SSC can be calculated. The algorithm formula used is as follows:

[0042] NDVI:

[0043]

[0044] EVI:

[0045]

[0046] NDWI:

[0047]

[0048] Chla:

[0049]

[0050]

[0051] SSC:

[0052] SSC0.7874*(Red / Green)20.1376*(Red / Green)+0.2098

[0053] Among them, Red is the surface reflectivity of the red light band, with a wavelength range of 0.64-0.67μm; Green is the surface reflectivity of the green light band, with a wavelength range of 0.53-0.59μm; Blue is the surface reflectivity of the blue light band, with a wavelength range of 0.45-0.51μm; Nir is the surface reflectivity of the near-infrared band, with a wavelength range of 0.85-0.88μm; Coastal is the surface reflectivity of the coastal band, with a wavelength range of 0.43-0.45μm, and a wavelength range of 0.43-0.45μm; M is a process quantity.

[0054] S2.2: Synthesize the processed band data in S1 and the results of each index parameter in S2.1 into new multi-band images, i.e., satellite composite images R1, R2, …, R n , and the reference composite image R.

[0055] In this embodiment, all the pre-processed band data of the images to be classified and the results of the calculated NDVI, Chl a, and SSC data are synthesized into new satellite composite image data.

[0056] S3: Draw the region of interest of the image data of the reference composite image R with verifiable ground object types, and make a classification decision tree for the image through machine learning; S3 specifically includes the following sub-steps:

[0057] S3.1: Obtain the main image features of different ground object categories through visual interpretation, field investigation, high-resolution image reference, and existing classification results, and determine the classified ground object types according to actual needs;

[0058] In this embodiment, the ground object types of the research area to be classified are divided into eight categories, i.e., ocean, fish pond, paddy field, marsh, river, unused land, dry (grass) land, and building land, based on research needs.

[0059] S3.2: Select the reference composite image R, and draw the region of interest corresponding to different ground object types according to the spectral features of the typical ground objects in the research area and the visual interpretation results;

[0060] S3.3: Based on the region of interest drawn in S3.2, machine learning is performed on the reference composite image R to obtain a classification decision tree supporting the classification of the image.

[0061] S4: Classify the other composite image data to be classified based on the classification decision tree obtained in S3, and draw new regions of interest for these composite images with reference to the classification results.

[0062] S4.1: Classify the satellite composite images R1, R2, …, R n to be classified using the classification decision tree obtained in S3.3;

[0063] S4.2: For each of the satellite composite images R1, R2, …, R n to be classified, compare the classification results of the decision tree with the spectral features of the image and the texture of the image, mark the misjudgment area of the decision tree, and then redraw the region of interest of each image.

[0064] In this embodiment, the terrain types of the river and fish pond are easily misjudged, and the misclassified areas are marked by manual re-drawing of the regions of interest based on the decision tree classification results and image texture information, so as to correct the classification results.

[0065] S5: Based on the newly drawn regions of interest, the terrain of the synthetic image data is re-classified by supervised classification.

[0066] Based on the newly drawn regions of interest in S4.2, the terrain types of the satellite synthetic images R1, R2,..., R n are newly classified by using the support vector machine method, so as to obtain the final classification results, as shown in Figure 2 .

[0067] In order to prove the effect of the method of the present application, the satellite synthetic images R1, R2,..., R n are newly classified by using the support vector machine alone and the decision tree alone, and the final classification results are shown in Table 1.

[0068] Table 1 Comparison of classification accuracy results of Hangzhou Bay land

[0069]

[0070] As can be seen from Table 1, in this embodiment, when only the support vector machine is used for classification, the kappa index can only reach about 0.95, and when only the decision tree is used, the kappa index can only reach about 0.85, and after the support vector machine classification is corrected by the decision tree, the accuracy can generally reach more than 0.97, greatly improving the accuracy of the classification.

[0071] In addition, when only the decision tree classification method is used, the classification process only screens the spectral values and band combination results of single pixels, so the classification speed is fast, but the classification accuracy is low due to the mixed pixel phenomenon in the medium and low resolution images. When only the support vector machine classification method is used, the data is projected to a higher dimension for analysis, the surrounding pixels of the target pixel are extracted to generate new data features, so that the information of the regular-shaped terrain types is extracted, and the classification accuracy is improved, but the calculation speed is slow due to the large amount of analysis. The method of the present application combines the characteristics of the two methods, the reliability of the regions of interest corrected and drawn based on the decision tree is higher, and the speed of re-classification by the support vector machine method can be improved by reducing the number of regions of interest, so that the method of the present application improves the classification accuracy while speeding up the classification.

[0072] Those skilled in the art can understand that the above description is only the preferred examples of the present application and is not used to limit the present application, and although the present application is described in detail with reference to the foregoing examples, those skilled in the art can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacements for part of the technical features. Any modification, equivalent replacement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A supervised classification method for remote sensing images based on decision tree correction, characterized in that: The steps include: S1: Obtain the satellite image to be classified and the reference image, pre-process them, and convert the DN value of the image data into the surface radiation value; crop the processed satellite image to obtain the image data of the required study area; the reference image is a satellite image obtained by the same sensor that covers the study area and the type of the ground feature can be verified; S2: The band data pre-processed by S1 is combined with various parameter results calculated based on the band data to form new synthetic image data, and the satellite synthetic images to be classified are obtained: R1, R2, ..., R n , and a reference composite image R; S3: Draw the region of interest on the image data of the reference synthetic image R whose ground object type can be verified, and create a classification decision tree for the image through machine learning; S3 specifically includes the following sub-steps: S3.1: Obtain image features of different land feature categories through visual interpretation, field research, high-resolution image reference, and existing classification results, and determine the type of land feature to be classified based on actual needs; S3.2: Select a reference composite image R and, based on the spectral characteristics and visual interpretation results of various typical landforms in the study area, draw regions of interest corresponding to different landform types; S3.3: Based on the region of interest drawn in S3.2, perform machine learning on the reference synthetic image R and obtain a decision tree that supports the classification of the image; S4: Other synthetic image data R1, R2, ..., R to be classified based on the classification decision tree obtained in S3 n Perform classification and draw new regions of interest on these synthetic images based on the classification results; S5: Based on the newly drawn regions of interest, the supervised classification of the support vector machine is used to reclassify the objects in these synthetic image data.

2. The remote sensing image supervised classification method based on decision tree correction according to claim 1 is characterized in that: In S2, various parameters calculated based on band data include normalized difference vegetation index NDVI, normalized difference vegetation index EVI, normalized difference water index NDWI, water chlorophyll index Chl a, and water sediment suspended matter index SSC.

3. The remote sensing image supervised classification method based on decision tree correction according to claim 2 is characterized in that: The specific calculation formulas for various parameters calculated based on band data are as follows: ； ； ； ； ； ； Among them, Red is the surface reflectivity of the red light band, with a wavelength range of 0.64-0.67μm; Green is the surface reflectivity of the green light band, with a wavelength range of 0.53-0.59μm; Blue is the surface reflectivity of the blue light band, with a wavelength range of 0.45-0.51μm; Nir is the surface reflectivity of the near-infrared band, with a wavelength range of 0.85-0.88μm; Coastal is the surface reflectivity of the coastal band, with a wavelength range of 0.43-0.45μm, and a wavelength range of 0.43-0.45μm; M is a process quantity.

4. The remote sensing image supervised classification method based on decision tree correction according to claim 1 is characterized in that: The S4 specifically includes the following sub-steps: S4.1: Use the classification decision tree obtained in S3.3 to classify the satellite synthetic images R1, R2, ..., R n Classify each one; S4.2: For the satellite synthetic images R1, R2,…, R n For each image in the dataset, the decision tree classification results are compared with the spectral characteristics and texture of the image, the areas where the decision tree misjudged the image are marked, and then the region of interest of each image is remade.

5. The remote sensing image supervised classification method based on decision tree correction according to claim 1 is characterized in that: In S1, the preprocessing of the satellite image to be classified and the reference image includes radiometric calibration, atmospheric correction, and geometric correction.

Images