CN-121977510-A - Multi-target collaborative natural resource investigation and monitoring method

CN121977510ACN 121977510 ACN121977510 ACN 121977510ACN-121977510-A

Abstract

The invention discloses a multi-target collaborative natural resource investigation monitoring method, which relates to the technical field of natural resource investigation, and the invention sets a natural resource investigation monitoring area, lays control points in the set natural resource investigation monitoring area, sets unmanned aerial vehicle airlines based on the laid control points, acquires image data in the natural resource investigation monitoring area in real time, further acquires historical image data in the natural resource investigation monitoring area, and the collected historical image data is processed in a data processing mode, the processed historical image data is subjected to feature extraction and classification after the processing is finished, a collaborative recognition model is constructed based on the classified historical image data features and the classified historical image data, finally, the image data in a real-time collected natural resource investigation monitoring area is intelligently recognized based on the constructed collaborative recognition model, the monitoring precision is adjusted based on the intelligent recognition, and the intelligence of natural resource investigation monitoring is improved.

Inventors

XIE TAO
HAN JIANNAN
ZHANG HUI
ZHANG YUNFENG
FU LEI
GUO CHAO

Assignees

自然资源陕西省卫星应用技术中心

Dates

Publication Date: 20260505
Application Date: 20260205

Claims (10)

1. The multi-target collaborative natural resource investigation and monitoring method is characterized by comprising the following steps of: s1, setting a natural resource investigation monitoring area, and arranging control points in the set natural resource investigation monitoring area; S2, setting an unmanned aerial vehicle route based on the laid control points, and acquiring image data in a natural resource investigation monitoring area in real time based on the set unmanned aerial vehicle route; s3, acquiring historical image data in a natural resource investigation monitoring area, and processing the acquired historical image data in a data processing mode to obtain processed historical image data; S4, extracting features of the processed historical image data and classifying the processed historical image data to obtain classified historical image features and corresponding classified historical image data; s5, constructing a collaborative recognition model based on the classified historical image data characteristics and the classified historical image data; S6, intelligent recognition is carried out on the image data in the real-time acquisition natural resource investigation monitoring area based on the constructed cooperative recognition model, and monitoring precision is adjusted based on the intelligent recognition.
2. The method for monitoring and investigation of natural resources by multi-objective cooperation according to claim 1, wherein the steps of setting a natural resource monitoring area and arranging control points in the set natural resource monitoring area comprise the steps of: S11, constructing a three-dimensional coordinate system based on a set natural resource investigation monitoring area, and setting a center point of the natural resource investigation monitoring area as a coordinate origin; s12, constructing a kilometer grid by taking 1 kilometer as a unit distance based on the constructed three-dimensional coordinate system, and selecting a representative template as a control point in a kilometer grid intersection and a midpoint; When the control points cannot be distributed according to the normal condition in a large-area water area and a mountain in the kilometer grid, the number of the control points is reduced according to the specific condition in the principle of meeting the correction and calculation of the normal emission of the industry; s13, numbering the control points based on the set control points And storing the control point coordinates corresponding to the numbers in a database.
3. The multi-objective collaborative natural resource investigation and monitoring method according to claim 1, wherein the unmanned aerial vehicle route is set based on the control points arranged, and image data in a natural resource investigation and monitoring area is collected in real time based on the unmanned aerial vehicle route, comprising the following steps: s21, setting an unmanned aerial vehicle route based on the number of the layout control points and the flight distance of the unmanned aerial vehicle; Recording the power consumption condition of the unmanned aerial vehicle under the flight distance of 1 km by a control variable method, and simultaneously estimating the flight distance of the unmanned aerial vehicle based on the recorded power consumption condition of the unmanned aerial vehicle; Planning the unmanned aerial vehicle route based on the estimated flight distance of the unmanned aerial vehicle and the distance between the corresponding numbered control points, comprising the following steps: firstly, arranging a single unmanned aerial vehicle for each control point through a saturated transportation algorithm, and recording a corresponding unmanned aerial vehicle route; Performing two-by-two combination on the recorded unmanned aerial vehicle route through a combination standard, and iterating combination operation; The merging standard is that the merged unmanned aerial vehicle route is smaller than the estimated unmanned aerial vehicle flight distance; Outputting an unmanned aerial vehicle route when no combinable unmanned aerial vehicle route exists, and setting the unmanned aerial vehicle route at the moment as the set unmanned aerial vehicle route; s22, acquiring image data in a natural resource investigation monitoring area in real time based on a set unmanned aerial vehicle route.
4. A multi-objective collaborative natural resource survey monitoring method according to claim 3, wherein the real-time acquisition of image data in a natural resource survey monitoring area based on a set unmanned aerial vehicle route comprises the steps of: determining position information in a natural resource investigation monitoring area in an orthographic correction mode; Setting position information in a natural resource investigation monitoring area to be composed of six elements 、 A determination is made that the data set is, The coordinates of the control points S in the monitoring area are investigated for natural resources, Is unmanned plane and coordinates is connected with the included angle of the vertical axis, Is an included angle between the unmanned aerial vehicle and the ground plane, For unmanned aerial vehicle projection an included angle of the longitudinal axis of the ground plane; selecting three control point coordinates 、、 No man-machine is used for collecting image data in a natural resource investigation monitoring area in real time to find a corresponding reference point; Measuring the coordinates of the corresponding reference point in the two-dimensional plane coordinate system of the image 、、 Orthographic correction is carried out through a collineation equation; Setting the image data after orthographic correction as the image data in a real-time acquisition natural resource investigation monitoring area.
5. The multi-objective collaborative natural resource investigation and monitoring method according to claim 1, wherein the steps of collecting the historical image data in the natural resource investigation and monitoring area, processing the collected historical image data by a data processing mode, and obtaining the processed historical image data comprise the following steps: s31, performing image enhancement on the collected historical image data in an image enhancement mode to obtain the image-enhanced historical image data; Calculating pixel probability distribution of the collected historical image data based on gray value distribution in the collected historical image data, wherein the gray value is a value of which the pixel point meets the condition r=g=b; constructing an image cumulative distribution function according to the calculated pixel probability distribution of the historical image data; the formula for constructing the cumulative distribution function of the image is as follows: ; Wherein, the Representing the total number of gray values in the image, Represent the first The gray values mapped by the cumulative distribution function, Representing gray values as Is used for displaying the number of the pixel points, The total number of the pixel points; image enhancement is carried out on the collected historical image data based on the constructed image cumulative distribution function; Multiplying gray value mapped by cumulative distribution function by linear stretching mode -1, Obtaining image-enhanced historical image data; s32, processing the history image data after image enhancement in a multi-frame comparison mode, and determining objects existing in the history image data; Object recognition is carried out on the history image data after image enhancement, and objects in the history image data after image enhancement are preliminarily determined according to differences between object pixel points and background pixel points; After the object in the history image data after the image enhancement is preliminarily determined, recording the coordinate position of the object in the history image data after the image enhancement in a multi-frame comparison mode, if the object coincidence degree in the history image data after the image enhancement of two adjacent frames is higher than 80%, the object in the history image data after the image enhancement is preliminarily determined, otherwise, the object in the history image data after the image enhancement is not determined; S33, summarizing objects in the history image data after the preliminary determination of the image enhancement, and obtaining the processed history image data.
6. The multi-objective collaborative natural resource investigation and monitoring method according to claim 1, wherein the feature extraction and classification of the processed historical image data to obtain the classified historical image features and the corresponding classified historical image data comprises the following steps: s41, extracting features of the processed historical image data to obtain historical image data extraction features; S411, dividing the processed historical image data into 16X 16 image data blocks in an equal proportion division mode, and inputting the divided image data blocks into a convolutional neural network; the convolutional neural network comprises an input layer, a convolutional layer, a pooling layer and a full-connection layer, wherein the convolutional layer and the pooling layer are sequentially stacked; s412, after the input layer of the convolutional neural network receives the divided image data blocks, the received image data blocks are transmitted to the convolutional layer, and the convolutional layer extracts local features in the image data blocks through convolutional operation; S413, after the convolution layer extracts local features in the image data block, the pooling layer processes the extracted local features in the image data block, and the pooling layer downsamples the features to reduce the data dimension; feature extraction is carried out on the divided image data blocks through a convolution layer and a pooling layer which are continuously stacked through a convolution neural network until feature convergence and convolution stop are carried out, and the obtained historical image data features are summarized and input into a full-connection layer; S414, integrating the obtained historical image data features by the full connection layer, outputting the historical image data features, and enabling the historical image data features to correspond to the processed historical image data one by one; setting the concrete expression form of the historical image data as a historical image data feature array; S42, classifying the processed historical image data based on the historical image data characteristics to obtain classified historical image characteristics and corresponding classified historical image data.
7. The multi-objective collaborative natural resource survey monitoring method of claim 6, wherein classifying the processed historical image data based on the historical image data features to obtain classified historical image features and corresponding classified historical image data comprises the steps of: Setting the characteristics of the historical image data as labels corresponding to the processed historical image data; The label of the history image data after summarizing is used for constructing a sample set, and N groups of labels are randomly selected from the sample set to serve as an initial classification center; After the initial classification centers are selected, calculating Euclidean distances between labels of the historical image data processed in the sample set and each initial classification center; Dividing labels of the historical image data processed in the sample set into corresponding initial classification centers based on the calculated Euclidean distance, and setting one classification center to correspond to one classification; after the primary division is finished, updating the classification center of each classification through a Euclidean distance algorithm; after the classification centers are updated, the Euclidean distance between the labels of the processed historical image data and each classification center is recalculated; repeating the steps until each classification center converges, and outputting the corresponding classification center and the label of the processed historical image data in the corresponding classification; And summarizing and outputting the labels of the history image data processed in the corresponding classification center to obtain the classified history image characteristics and the history image data processed in the corresponding classification.
8. The multi-objective collaborative natural resource research monitoring method of claim 1, wherein constructing a collaborative recognition model based on the categorized historical image data features and the categorized historical image data comprises the steps of: s51, extracting object contours in the classified historical image data in a contour extraction mode; Selecting historical image data corresponding to the classification center as a collaborative recognition standard image; Establishing a one-dimensional array, recording gray values of 8 neighborhoods around each pixel point in the collaborative identification standard image, and setting the pixel point to be in the interior of the outline and deleting the pixel point when the gray values of 8 neighborhoods around the pixel point are the same as the gray value of the central point; when the gray values of 8 neighborhoods around the pixel point are different from the gray value of the central point, setting the pixel point at the contour edge, and reserving the pixel point; traversing and summarizing each pixel point in the collaborative recognition standard image to obtain the object outline in the classified historical image data; S52, constructing a collaborative recognition model based on the object outline in the classified historical image data; setting a contour comparison model and setting upper and lower limits of a contour comparison result; The contour contrast model is as follows: ; Wherein, the Is of the outline And (3) with The area of the overlapping portion is defined by, And (3) with Areas corresponding to the contours respectively; representing the degree of contour coincidence; Setting an upper limit of a contour coincidence degree threshold and a lower limit of the contour coincidence degree threshold; when the contour coincidence degree of the two groups of objects is smaller than the lower limit of the contour coincidence degree threshold, the matching fails, and the current object is set as a new object; When the contour coincidence degree of the two groups of objects is larger than or equal to the upper limit of the contour coincidence degree threshold, the matching is successful, and the current object and the corresponding classified object are the same object; The collaborative recognition model is obtained by combining a contour comparison model with upper and lower limits of a contour comparison result.
9. The multi-objective collaborative natural resource investigation monitoring method according to claim 1, wherein the intelligent recognition of the image data in the real-time acquisition natural resource investigation monitoring area based on the constructed collaborative recognition model and the adjustment of the monitoring accuracy based on the intelligent recognition comprises the following steps: Collecting image data in a natural resource investigation monitoring area in real time, processing the image data through steps S3 and S4, intelligently identifying the image data in the natural resource investigation monitoring area collected in real time through a collaborative identification model after the processing is completed, and calculating the intelligent identification success rate; The recognition success rate calculation formula is as follows: ; Wherein, the Indicating the number of successful recognitions, Indicating the number of failed recognitions, The accuracy is achieved; Setting accuracy as investigation monitoring standard and setting accuracy threshold, when intelligent recognition accuracy is smaller than the set accuracy threshold, increasing unmanned aerial vehicle acquisition accuracy and frequency, otherwise, keeping unchanged.
10. A multi-target collaborative natural resource investigation monitoring method is characterized by further comprising a route planning module, a data acquisition module, a database, a data processing module, a data classification module, an intelligent identification module and an identification verification module; The route planning module is used for planning the route of the unmanned aerial vehicle; the data acquisition module is used for acquiring image data in a natural resource investigation monitoring area in real time; The database is used for storing image data in a real-time acquisition natural resource investigation monitoring area; The data processing module is used for processing the image data in the database to obtain processed image data; The data classification module is used for carrying out feature extraction and classification according to the processed image data to obtain classified image data; The intelligent recognition module is used for constructing a collaborative recognition model based on the classified image data characteristics and the classified image data; the identification verification module is used for verifying the identification result of the collaborative identification model.

Description

Multi-target collaborative natural resource investigation and monitoring method Technical Field The invention relates to the technical field of natural resource investigation, in particular to a multi-target collaborative natural resource investigation monitoring method. Background Natural resource monitoring is a long-term and continuous national strategy, and is an important foundation for supporting ecological civilization construction and realizing sustainable development. However, the traditional natural resource investigation mainly comprises manual in-situ investigation, so that the economic cost is high, the investigation period is long, and the investigation accuracy cannot meet the actual requirements. The prior art such as bulletin number is: S1, collect and process through the base map data of remote sensing, make full use of the existing result data, collect and process the work accurately and efficiently to the base map data information of remote sensing, the beneficial effects of the invention are that the invention is through the base map data of remote sensing to collect and process setting up, help the building information to collect accurately and efficiently, the invention is through drawing the outline of the building bottom surface of the building, adopt the semi-automatic fast extraction mode that the automatic extraction of the computer combines with artificial visual interpretation, fully utilize each automatic extraction software and GIS space to overlap and analyze the method that combines together in every link, collect the building vector data of rapid and accurate, the invention is through the result arrangement and analysis, through introducing, developing various quality inspection, arrangement, statistical analysis tool and management platform, guarantee the result quality. Aiming at the scheme, the current natural resource investigation is also a semi-automatic investigation mode combined with manual visual interpretation, the investigation period is long, and the economic cost is high. Disclosure of Invention The invention aims to provide a multi-target collaborative natural resource investigation and monitoring method which solves the problems existing in the background technology. In order to solve the technical problems, the invention provides a multi-objective collaborative natural resource investigation and monitoring method, which specifically comprises the following steps: s1, setting a natural resource investigation monitoring area, and arranging control points in the set natural resource investigation monitoring area; S2, setting an unmanned aerial vehicle route based on the laid control points, and acquiring image data in a natural resource investigation monitoring area in real time based on the set unmanned aerial vehicle route; s3, acquiring historical image data in a natural resource investigation monitoring area, and processing the acquired historical image data in a data processing mode to obtain processed historical image data; S4, extracting features of the processed historical image data and classifying the processed historical image data to obtain classified historical image features and corresponding classified historical image data; s5, constructing a collaborative recognition model based on the classified historical image data characteristics and the classified historical image data; S6, intelligent recognition is carried out on the image data in the real-time acquisition natural resource investigation monitoring area based on the constructed cooperative recognition model, and monitoring precision is adjusted based on the intelligent recognition. Preferably, the setting a natural resource investigation monitoring area and arranging control points in the set natural resource investigation monitoring area includes the following steps: S11, constructing a three-dimensional coordinate system based on a set natural resource investigation monitoring area, and setting a center point of the natural resource investigation monitoring area as a coordinate origin; s12, constructing a kilometer grid by taking 1 kilometer as a unit distance based on the constructed three-dimensional coordinate system, and selecting a representative template as a control point in a kilometer grid intersection and a midpoint; When the control points cannot be distributed according to the normal condition in a large-area water area and a mountain in the kilometer grid, the number of the control points is reduced according to the specific condition in the principle of meeting the correction and calculation of the normal emission of the industry; s13, numbering the control points based on the set control points And storing the control point coordinates corresponding to the numbers in a database. Preferably, the unmanned aerial vehicle route is set based on the laid control points, and the image data in the natural resource investigation monitoring area is collected in real time based on