Multi-metal deposit remote sensing prospecting method

Abstract

The invention discloses a multi-metal deposit remote sensing prospecting method, and relates to the technical field of remote sensing prospecting. Comprising the steps of determining ore deposit landform features through a preset prospecting theory to serve as prospecting marks, and obtaining remote sensing images of prospecting mark areas; preprocessing the remote sensing image, calculating a feature vector and a feature value based on waveband data, and performing data segmentation according to the feature value and the feature vector obtained through calculation to obtain a segmented image; constructing a difference value extraction network, collecting samples to train the difference value extraction network, and inputting the segmented image into the trained difference value extraction network to obtain a processed image; and judging whether the processed image is an ore deposit image or not. According to the method, important guidance can be provided for subsequent mineral exploration and resource development, and the prospecting efficiency and accuracy are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of remote sensing prospecting, and particularly relates to a multi-metal deposit remote sensing prospecting method. BACKGROUND

[0002] The current development of science and technology has driven the application of mechanical equipment, and the use of machines depends on the consumption of resources. In human production and life, the consumption of traditional natural resources is increasing, among which mineral resources are the representatives, which are closely related to people's life and are the basis of human production and living materials, and play an indispensable important role in the process of sustainable development of national economy. Therefore, developing a multi-metal remote sensing prospecting method is of great significance to maintain the safety of mineral resources. With the development of science and technology, remote sensing technology has gradually become an important means of multi-metal deposit prospecting due to its high efficiency and comprehensiveness.

[0003] Therefore, to provide a multi-metal deposit remote sensing prospecting method to solve the problems existing in the prior art is a problem that needs to be solved by those skilled in the art. SUMMARY

[0004] Therefore, the present application provides a multi-metal deposit remote sensing prospecting method, which can provide important guidance for subsequent mineral exploration and resource development, and improve the efficiency and accuracy of prospecting.

[0005] In order to achieve the above-mentioned purpose, the present application adopts the following technical scheme:

[0006] A multi-metal deposit remote sensing prospecting method comprises the following steps:

[0007] Determine the deposit landform features as the prospecting marks by the preset prospecting theory, and obtain the remote sensing image of the prospecting mark area;

[0008] Pretreat the remote sensing image, calculate the feature vector and the feature value based on the band data, and perform data segmentation according to the calculated feature value and feature vector to obtain the segmented image;

[0009] Construct a difference extraction network, collect samples to train the difference extraction network, input the segmented image into the trained difference extraction network, and obtain the processed image;

[0010] Determine whether the processed image is a deposit image.

[0011] Optionally, the preset prospecting theory comprises:

[0012] Based on the type of multi-metal deposit, the ore-controlling elements of the preset area are analyzed to establish the source-reservoir-cap mode of the preset area;

[0013] The ore prospecting theory is established by combining the source-reservoir-cap rock pattern with geological features of the preset area.

[0014] Optionally, the remote sensing image is acquired to determine ore prospecting marks, and differences in landform material features of the preset area are preset through remote sensing images.

[0015] The differences in boundary textures of landforms on both sides of structures in the preset area are displayed through remote sensing images, and a structure mark is established.

[0016] The differences in material compositions of intrusive rocks and surrounding rocks in the preset area are displayed through remote sensing images, and a lithology mark is established.

[0017] The differences in material compositions of altered mineral components and surrounding rocks in the preset area are displayed through remote sensing images, and an altered mineral mark is established.

[0018] Optionally, the segmented image comprises:

[0019] The remote sensing image is preprocessed in a manner comprising debounding processing and interference removing processing, and a preprocessed remote sensing image is obtained.

[0020] The preprocessed remote sensing image is calculated by using spatial data set analysis, a band covariance matrix is obtained, and then a feature vector and a feature value are obtained.

[0021] The segmented image is obtained by data segmentation according to the calculated feature vector and feature value.

[0022] Optionally, the difference extraction network comprises:

[0023] A feature extraction module is configured to extract multi-scale feature maps of the segmented image and perform normalization output.

[0024] A frame generation module is configured to generate a proposal frame according to the feature map.

[0025] An alignment module is configured to perform alignment processing on the proposal frame and the feature map.

[0026] A change detection module is configured to perform segmentation processing on the feature output by the alignment module, and output a change region image.

[0027] Optionally, the change detection module further comprises training and optimizing the difference extraction network by using a multi-task feature function, and the training manner comprises:

[0028] Initializing model weights of the difference extraction network.

[0029] Data augmentation techniques are used to increase the number of training sample data of images from different time periods in a preset region; the difference extraction network is trained end-to-end using backpropagation and stochastic gradient descent algorithms, and the model parameters of the difference extraction network are updated.

[0030] As can be seen from the above technical solution, compared with the prior art, the present invention provides a remote sensing prospecting method for polymetallic deposits, which has the following beneficial effects: 1) The present invention can better reflect the dynamic change information of the mining area by comparing the images of different time phases, which helps to discover potential mineralized areas; 2) The network model based on the difference image method of the present invention can significantly improve the accuracy of change detection, and effectively solves the problem of erroneous changes and false changes compared with traditional change detection methods; 3) The present invention can provide important guidance for subsequent mineral exploration and resource development, and improve the efficiency and accuracy of prospecting. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0032] Figure 1 This is a flowchart of a remote sensing prospecting method for polymetallic deposits disclosed in this invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Reference Figure 1 As shown, this invention discloses a remote sensing prospecting method for polymetallic deposits, comprising the following steps:

[0035] The geomorphological features of mineral deposits are determined by a pre-defined mineral exploration theory and used as mineral exploration indicators. Remote sensing images of the mineral exploration indicator areas are then obtained.

[0036] The remote sensing image is preprocessed, feature vectors and feature values ​​are calculated based on band data, and the data is segmented according to the calculated feature values ​​and feature vectors to obtain the segmented image;

[0037] The difference extraction network is constructed, sample training is performed on the difference extraction network, the segmented image is input into the trained difference extraction network, and a processed image is obtained.

[0038] It is judged whether the processed image is a mineral deposit image.

[0039] Further, the preset prospecting theory includes:

[0040] Based on the type of the polymetallic deposit, the ore-controlling elements of the preset area are analyzed to establish a source-reservoir-cap mode of the preset area.

[0041] The source-reservoir-cap mode is combined with the geological features of the preset area to establish a prospecting theory.

[0042] Specifically, the structure is the trace left by the displacement and deformation of the rock or rock mass under stress, which not only provides a channel for the migration of ore-bearing hydrothermal fluid, but also generates high temperature and high pressure in the movement process, causing rock melting and metamorphism, promoting the flow and aggregation of ore-forming materials, and further providing a favorable ore-forming environment for the formation of polymetallic ore. Lithology refers to some properties that reflect the characteristics of rocks, such as color, composition, structure, etc. It not only provides necessary sources for mineralization, but also provides storage space and cap rock, which is closely related to mineralization. Lithology containing ore-forming materials provides a source for the formation of polymetallic ore, flowing lithology provides a medium for the formation of polymetallic ore, and lithology with high porosity provides storage space for the formation of polymetallic ore. Alteration minerals are minerals that change in composition, structure and structure due to changes in temperature and pressure during rock and mineralization. It not only reflects the physical and chemical conditions of mineralization, the properties and evolution process of ore-forming hydrothermal fluid, but also provides information on the migration, enrichment and precipitation of ore-forming elements. Further, based on the prospecting theory, the differences in topographic material characteristics of the preset area are displayed through remote sensing images, and remote sensing prospecting marks are established.

[0043] To predict the type of mineral, the location of ore body and the degree of mineralization and enrichment using the information obtained by analyzing the above ore-controlling elements. It is necessary to summarize and summarize the above various ore-controlling elements, and establish a preliminary source-reservoir-cap mode (the source-reservoir-cap mode describes the process of mineral deposit formation, preservation and enrichment. This mode considers biological factors, reservoir capacity and cap rock in the earth's crust, and has guiding significance for the formation of mineral deposits in a specific area). When the source-reservoir-cap mode is combined with the geological features of a specific area, the geological background, structural characteristics, lithology distribution, mineralization and other factors of the area can be further refined and improved. Through the study of the formation mechanism of the mineral deposit in the specific area, mineral exploration, resource evaluation and mineral deposit development can be better guided.

[0044] Further, the remote sensing image is acquired to determine the ore prospecting mark, the differences in the topographic material characteristics of a preset area are shown through the remote sensing image, the differences in the boundary texture of the topography on both sides of the structure in the preset area are shown through the remote sensing image, and a structure mark is established;

[0045] The differences in the material composition of the intrusive rock and the surrounding rock in the preset area are shown through the remote sensing image, and a lithology mark is established;

[0046] The differences in the material composition of the altered mineral components and the surrounding rock in the preset area are shown through the remote sensing image, and an altered mineral mark is established.

[0047] Specifically, the differences in the topographic landscape, material composition and attributes of the mining area are different from those of the general area. They will cause the spectrum, hue and texture of the mining area on the remote sensing image to be abnormal, thereby becoming a direct remote sensing ore prospecting mark. For example, the topography, hue and texture anomalies on both sides of the structure can be shown to establish a structure mark; the differences in the material composition of the intrusive rock and the surrounding rock can be shown to cause the hue, texture and ring structure anomalies of the intrusive rock and the surrounding rock on the image, thereby establishing a lithology mark; the differences in the material composition of the altered mineral components and the surrounding rock in the altered zone can be shown to cause the hue and spectrum anomalies of the altered mineral components and the surrounding rock on the remote sensing image, thereby establishing an altered mineral mark. Further, the differences in the trace elements, water content and water content between the toxic vegetation and the normal vegetation in the mining area can be shown to cause the hue and spectrum variations on the image, which can also be used as a remote sensing ore prospecting mark.

[0048] Further, the remote sensing image includes the remote sensing images of the same wave band before and after the same phase of the same area.

[0049] Further, obtaining the segmented image includes:

[0050] The remote sensing image is preprocessed in a manner including debounding processing and interference removal processing to obtain a preprocessed remote sensing image;

[0051] The preprocessed remote sensing image is calculated by spatial data set analysis to obtain a wave band covariance matrix, and then the characteristic vector and the characteristic value are obtained;

[0052] The segmented image is obtained by data segmentation according to the calculated characteristic vector and characteristic value.

[0053] Further, the difference extraction network includes:

[0054] The feature extraction module is configured to extract multi-scale feature maps of the segmented image and perform normalization output; the box generation module is configured to generate a proposal box according to the feature map, and generate the proposal box on the feature map output by the feature extraction network using a sliding window, and use two convolution kernels on the sliding window to respectively predict the bounding box and the category of the target to be detected in the original image corresponding to the window, and output the center coordinates, width and height of the proposal region; the alignment module is configured to align the proposal box and the feature map, and serve as an input of a classification, bounding box regression and segmentation branch in the change area detection module;

[0055] The change detection module is configured to perform classification, bounding box regression and pixel-level segmentation processing on the features output by the alignment layer, and optimize the convolutional neural network model using a multi-task feature function, and then perform classification, bounding box regression and pixel-level segmentation tasks, and use image segmentation to achieve accurate change area positioning, and output a change area image.

[0056] Further, the change detection module further includes training and optimizing the difference extraction network using a multi-task feature function, and the training method includes:

[0057] initializing the model weight of the difference extraction network;

[0058] using a data augmentation technique to expand the number of training sample data of different time images in a preset area, and after translating the original image in the training sample data by 0-30% in the horizontal direction and the vertical direction, flipping, randomly rotating by 0-45°, and then translating and flipping again;

[0059] using a back propagation algorithm and a stochastic gradient descent algorithm to perform end-to-end training on the difference extraction network, and updating the model parameters of the difference extraction network.

[0060] Further, whether the image is a deposit image can be determined by setting a threshold, comparing a standard image, and comprehensively judging the features in the image in combination with the experience and knowledge of geologists. If the image contains features that meet the remote sensing prospecting criteria for deposits, such as specific structural anomalies, lithological anomalies, or alteration mineral anomalies, etc., it can be preliminarily determined that the image is a deposit image.

[0061] Based on the judgment result of the deposit image, the type of the mineral, the location of the ore body, and the degree of mineralization enrichment can be further predicted. For example, the type of the mineral can be inferred from the spectral characteristics of the alteration minerals in the image; the approximate location of the ore body can be determined by analyzing the distribution characteristics of the structure and lithology; and the degree of mineralization enrichment can be evaluated in combination with the range and intensity of the abnormal area in the image. These prediction results can provide important guidance for subsequent mineral exploration and resource development, and improve the efficiency and accuracy of prospecting.

[0062] The foregoing description of the disclosed embodiments enables a person skilled in the art to make or use the application. Modifications of these embodiments will occur to persons of skill in the art, and that the appended claims are intended to cover all such modifications that do not depart from the true spirit and scope of the application. Therefore, the application is not limited to the embodiments shown but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A remote sensing method for polymetallic deposit prospecting, comprising the following steps: determining the landform features of the deposit according to a preset prospecting theory as a prospecting marker, and obtaining a remote sensing image of the prospecting marker area; preprocessing the remote sensing image, calculating the eigenvectors and eigenvalues based on the band data, and performing data segmentation according to the calculated eigenvectors and eigenvalues to obtain a segmented image; constructing a difference extraction network, collecting samples to train the difference extraction network, inputting the segmented image into the trained difference extraction network, and obtaining a processed image; judging whether the processed image is a deposit image.

2. The method according to claim 1, characterized in that, The preset prospecting theory comprises: based on the type of the polymetallic deposit, analyzing the ore-controlling elements of a preset area, and establishing a source-reservoir-cap mode of the preset area; combining the source-reservoir-cap mode with the geological features of the preset area to establish a prospecting theory.

3. The method according to claim 1, characterized in that, The remote sensing image is obtained to determine the prospecting marker, the differences in the landform material features of the preset area are shown through the remote sensing image, the differences in the boundary texture of the landform on both sides of the structure in the preset area are shown through the remote sensing image, and a structural marker is established; the differences in the material composition of the intrusive rock and the surrounding rock in the preset area are shown through the remote sensing image, and a lithological marker is established; the differences in the altered mineral composition and the surrounding rock material composition in the preset area are shown through the remote sensing image, and an altered mineral marker is established.

4. The method according to claim 1, characterized in that, The segmented image comprises: the remote sensing image is preprocessed by a method comprising debounding and interference removal to obtain a preprocessed remote sensing image; the preprocessed remote sensing image is calculated by spatial data set analysis to obtain a band covariance matrix, and then the eigenvectors and eigenvalues are obtained; the data is segmented according to the calculated eigenvectors and eigenvalues to obtain a segmented image.

5. The method according to claim 1, characterized in that, The difference extraction network comprises: a feature extraction module for extracting multi-scale feature maps of the segmented image and performing normalization output; a box generation module for generating proposal boxes according to the feature maps; an alignment module for aligning the proposal boxes and the feature maps; a change detection module for segmenting the features output by the alignment module to output a change region image.

6. The method according to claim 1, characterized in that, The change detection module further comprises training and optimizing the difference extraction network using a multi-task feature function, and the training method comprises: initializing the model weight of the difference extraction network; using data augmentation technology to expand the number of training sample data of different time images of the preset area; using a back propagation algorithm and a stochastic gradient descent algorithm to train the difference extraction network end to end, and updating the model parameters of the difference extraction network.

Images