A method for extracting ground object information based on positive and negative terrain idea

Abstract

This invention belongs to the field of remote sensing image classification, specifically involving a remote sensing method for extracting ground feature information based on the concept of positive and negative terrain. The method selects remote sensing data containing target ground feature information features; introduces the concepts of positive and negative terrain, and uses difference enhancement to construct positive and negative enhanced data of the target ground features; then, it performs secondary enhancement on the remote sensing data containing target ground feature information features, namely, extracting initial target ground feature information based on the initial remote sensing data and clustering techniques, and applying a regional mask of the initial target ground feature information to enhance the data in both directions, setting 0 values ​​in the data to NAN; finally, the data after secondary enhancement is standardized to 0-1, and clustering techniques are applied to extract the final target ground feature information. This invention features a simple technical process, high accuracy, and strong universality.

Description

Technical Field

[0001] This invention belongs to the field of remote sensing image classification, specifically relating to a remote sensing extraction method for ground feature information based on the concept of positive and negative terrain. Background Technology

[0002] The application of remote sensing technology for large-scale, rapid, and effective extraction of ground feature information has become one of the most important spatial information extraction techniques. Especially with the rapid development of satellite technology and the widespread use of unmanned aerial vehicles (UAVs), massive amounts of remote sensing data urgently need processing and extraction. Ground feature information extraction methods based on remote sensing technology are one of the important research directions in the field of remote sensing. Therefore, the research and innovation of ground feature information remote sensing extraction methods have significant scientific importance and research value. With the increasingly broad application fields and the development of interdisciplinary and cross-disciplinary fields, more and more technologies and theories are being introduced, leading to the rapid development and wider application of various ground feature information remote sensing extraction methods.

[0003] Although there are many methods for remote sensing extraction of ground feature information, and a great deal of research results and applications have been achieved, their advantages and disadvantages are also obvious. The common problems include the presence of many subjective human factors, high data requirements, and high limitations in promotion and application. Summary of the Invention

[0004] To overcome the aforementioned problems in existing remote sensing methods for extracting ground feature information, this invention provides a remote sensing method for extracting ground feature information based on the concept of positive and negative terrain, which reduces the amount of data and minimizes human interference, thereby improving accuracy, automation, and universality.

[0005] The specific technical solution is as follows:

[0006] A remote sensing method for extracting ground feature information based on the concept of positive and negative terrain includes the following steps:

[0007] (1) Selection of remote sensing data for target ground feature information

[0008] For different extracted target features, select remote sensing data of target feature information from appropriate dates;

[0009] (2) Obtaining reverse data of remote sensing raw data of target feature information based on the concept of positive and negative terrain;

[0010] The concepts of positive and negative terrain are introduced. The maximum value of the positive terrain data is calculated, and the positive terrain data is subtracted from the maximum value to obtain the negative terrain data of the remote sensing raw data, i.e., the reverse data.

[0011] According to the concept of positive terrain, the positive terrain data here refers to the original remote sensing data of the selected target land cover information features, also known as positive data;

[0012] (3) Obtain positive and negative augmentation data;

[0013] By subtracting negative terrain data from positive terrain data, positive and negative terrain enhancement data are obtained, which can increase the information difference between target features and non-target features.

[0014] (4) Mask secondary enhancement and standardization

[0015] The raw remote sensing data was divided into two categories using clustering technology: initial target features and initial non-target features. The initial target feature regions from the clustering results were used for both forward and reverse data enhancement using masks, and the zero-value data after masking were processed into NaN values. The enhanced data after masking and the NaN values ​​were then normalized to 0-1 values.

[0016] (5) Extraction of target features.

[0017] Based on remote sensing data standardized to 0 to 1, clustering technology is used to divide the standardized remote sensing data into two categories: one is the final target ground features, and the other is the final non-target ground features.

[0018] This invention provides a remote sensing method for extracting ground feature information based on the concept of positive and negative terrain. Targeting different characteristics of different target ground features, it constructs positive and negative enhanced data based on the concept of positive and negative terrain, and applies clustering techniques to extract the target ground features. The principle of this method is as follows: In the selected remote sensing data containing target ground feature information features, the differences between target and non-target ground features are relatively obvious, but direct extraction still results in significant errors. Based on the concept of positive and negative terrain, positive and negative terrain data of the target ground features are constructed using the concepts of positive and negative terrain. Since the feature information of target and non-target ground features in the positive and negative terrain data is opposite, the positive and negative terrain data (positive and negative data) constructed through the difference enhancement method further amplifies the feature differences between target and non-target ground features, thus providing favorable conditions for subsequent extraction of target ground features with higher accuracy. Finally, the final target ground feature information is extracted through data standardization and clustering techniques. Experiments show that this method has the advantages of simple technical process, strong universality, and high accuracy. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method of the present invention;

[0020] Figure 2 The spatial distribution map of winter wheat planting in Xinji City in 2014 is shown as an example. Detailed Implementation

[0021] The present invention will be further described below with reference to the accompanying drawings and embodiments. The embodiments described below are only a part of the present invention and not all of the embodiments. Therefore, the following embodiments are only used to more clearly describe the technical solutions of the present invention and should not be used to limit the scope of protection of the present invention.

[0022] Adopting such Figure 1 This paper presents a remote sensing method for extracting ground feature information based on the concept of positive and negative terrain. The following example uses this method to extract the winter wheat planting area and spatial distribution in Xinji City, Hebei Province in 2014, and verifies its accuracy. The remote sensing data can be medium- or high-resolution remote sensing images, such as SPOT, GF1, and TM / ETM images. This extraction uses GF1 remote sensing image data with a spatial resolution of 16 meters. The specific extraction process is as follows:

[0023] 1. Selection of remote sensing data for target ground features—winter wheat

[0024] Through literature review and on-site observation, it was found that in late March and early April, other green vegetation in Xinji City had not yet turned green. Wild grasses, trees, and other green disturbance features were mostly still withered and yellow, while the target winter wheat was in the jointing stage after its greening process. Therefore, based on data availability, this embodiment selected Gaofen-1 multispectral data from April 3, 2014. Since the Normalized Difference Vegetation Index (NDVI) is the most commonly used indicator in remote sensing for expressing green vegetation information, this embodiment selected NDVI as the indicator for extracting winter wheat planting information. The NDVI of the GF1 data from Xinji City on April 3, 2014, was calculated according to the NDVI formula, thus obtaining the Gaofen-1 NDVI remote sensing data for Xinji City.

[0025] 2. Forward and reverse data augmentation

[0026] Based on the concepts of positive and negative terrain, the difference enhancement method is applied to subtract the negative terrain data from the positive terrain data of NDVI to construct NDVI positive and negative terrain enhancement data, i.e., NDVI positive and negative enhancement data.

[0027] 3. Mask data

[0028] Based on the raw NDVI remote sensing data, clustering techniques were applied to divide Xinji City into two categories: one containing initial winter wheat planting information.

[0029] Another category is initial non-winter wheat information. Clustering was used to extract forward and reverse NDVI enhancement data from the initial winter wheat planting areas, and the 0 values ​​in the data were set to NAN to obtain secondary enhanced NDVI data.

[0030] 4. Extraction and accuracy verification of winter wheat planting information.

[0031] The enhanced NDVI data was standardized to remote sensing data of 0-1, and clustering technology was applied to divide Xinji City into two categories to extract the final winter wheat planting information.

[0032] Accuracy Verification: Supervised classification based on ground sample points is a highly accurate remote sensing classification method, second only to visual interpretation results. In this embodiment, the accuracy of the supervised classification results based on the maximum likelihood method is compared and verified: that is, using the Gaofen-1 multispectral data from the same day (April 3, 2014), the maximum likelihood method is applied to supervised classification to extract winter wheat planting information in Xinji City, and the confusion matrix method and supervised classification results are used to verify the accuracy of the classification results of the remote sensing extraction method based on the idea of ​​positive and negative terrain. The accuracy verification results are shown in Table 1.

[0033] Table 1. Statistics of confusion matrix information extracted from winter wheat in Xinji City using the model of this invention and the maximum likelihood method supervised classification.

[0034]

[0035] The above embodiments are only used to illustrate the present invention and are not intended to limit the present invention. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, all equivalent technical solutions also fall within the scope of the present invention, and the patent protection scope of the present invention should be defined by the claims.

Claims

1. A remote sensing method for extracting ground feature information based on the concept of positive and negative terrain, characterized in that, Includes the following steps: (1) Selection of remote sensing data for target ground feature information For different extracted target features, select remote sensing data of target feature information from appropriate dates; (2) Obtaining reverse data of remote sensing raw data of target feature information based on the concept of positive and negative terrain; Positive terrain data refers to the raw remote sensing data containing information and features of selected target land features, also known as forward data. The maximum value of the positive terrain data is calculated, and the positive terrain data is subtracted from the maximum value to obtain the inverse terrain data of the original remote sensing data, i.e., the reverse data. (3) Obtaining forward and reverse augmentation data By subtracting negative terrain data from positive terrain data, positive and negative terrain enhancement data are obtained, which can increase the information difference between target features and non-target features. (4) Mask secondary enhancement and standardization The raw remote sensing data is divided into two categories using clustering technology: initial target features and initial non-target features. The initial target feature regions in the clustering results are used to enhance the data in both forward and reverse directions using masks, and the 0-value data after masking is processed into NAN. Then, the data after masking enhancement and the 0-value data processed into NAN are normalized to 0~1 data. (5) Extraction of target features.

2. The remote sensing extraction method for ground feature information based on the concept of positive and negative terrain, as described in claim 1, is characterized in that... Step (5) specifically includes the following process: Based on the remote sensing data standardized to 0~1, the standardized remote sensing data is divided into two categories through clustering technology: one category is the final target ground features, and the other category is the final non-target ground features.

Images