A method for identifying geological remote sensing image data features for mineral exploration

Abstract

The present application relates to the field of image detection technology, and specifically to a method for identifying geological remote sensing image data features for prospecting and ore exploration, the method comprising: obtaining grayscale images of each frame image in a high-altitude remote sensing image; obtaining each connected domain in each grayscale image according to the grayscale value of each corner point and its neighboring pixel points in each grayscale image; determining the motion vector of each corner point in each grayscale image; determining the average motion difference value of each connected domain based on the difference in motion vectors of the corner points between each connected domain and all other connected domains in the grayscale image where it is located; determining the stability of each connected domain; determining the grain flow direction value of each window based on the direction angle of the grain in each window, the MLBP value of the central pixel point of the window, and the grayscale value distribution of the pixel point; determining the debris flow confidence of each connected domain, and detecting the debris flow connected domain. The present application aims to help practitioners avoid debris flow prone areas when prospecting for ore by improving the accuracy of detecting debris flow.

Description

Technical Field

[0001] The present application relates to the field of image detection technology, and in particular to a method for identifying features of geological remote sensing image data for prospecting and mineral exploration. Background Art

[0002] At present, mineral exploration has shifted from shallow to deep, from looking for outcrops to looking for minerals in covered areas, and from low mountain areas to medium and high mountains, making it more difficult to find minerals. Mountainous terrain is steep, and there are many deposits of mud, sand, and stones, which are prone to mudslides during heavy rains. In areas prone to mudslides, geological exploration activities require more safety measures and time. Identifying mudslide disasters can help avoid areas prone to mudslides when looking for minerals, and can more efficiently carry out geological exploration activities in low-risk areas, reduce interference with areas prone to mudslides, reduce damage to the environment, and protect the ecological balance.

[0003] When existing remote sensing detection technology is used to detect debris flows in areas with complex terrain, the feature recognition results of debris flow disasters are poor due to the influence of factors such as clouds and vegetation. In addition, existing algorithms usually perform feature detection on one image, but debris flows and similar objects such as rivers have many similar features in the same still image, and existing algorithms are prone to false detection. Summary of the invention

[0004] In order to solve the above technical problems, the present application provides a method for identifying features of geological remote sensing image data for mineral exploration to solve the existing problems.

[0005] The present invention discloses a method for identifying features of geological remote sensing image data for prospecting and mineral exploration using the following technical solutions:

[0006] An embodiment of the present application provides a method for identifying features of geological remote sensing image data for prospecting and mineral exploration, the method comprising the following steps:

[0007] Obtain the grayscale image of each frame in the high-altitude remote sensing image;

[0008] A corner point detection algorithm is used to obtain each corner point in each grayscale image, and each connected domain in each grayscale image is obtained according to the grayscale values ​​of each corner point and its neighboring pixel points in each grayscale image;

[0009] Determine the motion vector of each corner point in each grayscale image according to the position change of each corner point in each grayscale image in the adjacent grayscale image; determine the average motion difference value of each connected domain based on the difference in motion vectors of the corner points between each connected domain and all other connected domains in the grayscale image where it is located;

[0010] Based on the gray value distribution of the pixels in each connected domain and the difference between the average motion difference value between each connected domain and all the connected domains in the gray image where it is located, the stability of each connected domain is determined; a window is constructed with each pixel in each connected domain as the center, and the texture flow value of each window is determined based on the direction angle of the texture in each window, the MLBP value of the pixel at the center of the window, and the gray value distribution of the pixel;

[0011] Based on the motion vectors of all corner points in any connected domain and the stability level, respectively obtaining matching connected domains of any connected domain from the remaining grayscale images;

[0012] Based on the discreteness of all the texture flow direction values ​​in each connected domain, and the distribution of the contours of each connected domain and its matching connected domain, and combined with the stability, determine the debris flow confidence of each connected domain;

[0013] The debris flow connected domain is detected based on the debris flow confidence.

[0014] In one embodiment, the process of obtaining each connected domain in each grayscale image is:

[0015] For each grayscale image, all corner points in the grayscale image are arranged from large to small according to the grayscale value, and the first preset number of corner points are used as initial seed points. Combined with the region growing algorithm, the absolute value of the difference in grayscale value between the growing pixel point and the adjacent pixel point is calculated, and the absolute value of the difference is used as the similarity between the growing pixel point and the adjacent pixel point. When the normalized value of the similarity is greater than the preset similarity threshold, the growth is stopped, and the growth regions obtained by using the region growing algorithm are used as connected domains.

[0016] In one embodiment, the method for determining the motion vector is: taking all corner points in each grayscale image and the adjacent next frame grayscale image as input of the optical flow method, and outputting the motion vector of each corner point in each grayscale image.

[0017] In one embodiment, the process of determining the average motion difference value is:

[0018] The mean of the motion vectors of all corner points of each connected domain is recorded as the motion mean;

[0019] Any connected domain in any grayscale image is denoted as i, and the difference in the motion mean between the i-th connected domain and the other connected domains in the grayscale image where it is located is denoted as the motion difference; the mean of all the motion differences of the i-th connected domain is taken as the average motion difference value of the i-th connected domain.

[0020] In one embodiment, the process of determining the stability is as follows:

[0021] A grayscale run matrix is ​​constructed according to the grayscale values ​​of all pixels in the i-th connected domain, and the gradient value of each edge pixel in the i-th connected domain is calculated using the edge detection operator, and the average value of the gradient values ​​of all edge pixels in the i-th connected domain is calculated;

[0022] Calculate the average of the average motion difference values ​​of all connected domains in the grayscale image where the i-th connected domain is located, recorded as the motion difference mean; calculate the difference between the average motion difference value of the i-th connected domain and the motion difference mean;

[0023] The stability of the i-th connected domain is positively correlated with the maximum run length of the grayscale run matrix and the average value, and negatively correlated with the difference and the number of corner points of the i-th connected domain.

[0024] In one embodiment, the process of determining the texture flow direction value is as follows:

[0025] The MLBP algorithm is used to calculate the MLBP value of the central pixel of each window, and the average of the grayscale values ​​of all pixels in each window is recorded as the grayscale mean; the second-order moment technology is used to calculate the texture direction angle in each window according to the area of ​​each window and the coordinates of the central pixel of each window, and the texture direction angle is converted into a radian value;

[0026] The texture flow direction value is positively correlated with the MLBP value and the grayscale mean value, and negatively correlated with the radian value.

[0027] In one embodiment, the process of determining the matching connected domain is:

[0028] Calculating the sum of the motion mean and the stability of each connected domain; taking the connected domains in the remaining frame images with the smallest difference in the sum with the i-th connected domain as the initial matching connected domains of the i-th connected domain;

[0029] The pixels on the outermost four sides of each frame of grayscale image are recorded as grayscale image boundary pixels, and any frame of grayscale image is recorded as a. Starting from the a-th frame of grayscale image, all the pixels of the matching connected domain of the i-th connected domain of the a-th frame of grayscale image in each grayscale image are traversed forward and backward frame by frame. If the grayscale image boundary pixels appear in the pixels, the traversal is stopped, and the initial matching connected domain of the i-th connected domain in each frame of grayscale image between the forward traversal stop and the backward traversal stop is used as the matching connected domain of the i-th connected domain.

[0030] In one embodiment, the process of determining the debris flow confidence is as follows:

[0031] The grayscale images of the i-th connected domain and its matching connected domains are recorded as the complete contour image of the i-th connected domain;

[0032] The i-th connected domain in the complete contour image of the i-th connected domain in each frame or the matched connected domain of the i-th connected domain is recorded as the i-th connected domain, and the difference between the Hu moments of the i-th connected domain in any two adjacent frames of the complete contour image of the i-th connected domain is recorded as the contour difference;

[0033] The debris flow confidence of the i-th connected domain is positively correlated with the discrete degree of the texture flow direction values ​​of all windows in the i-th connected domain, the discrete degree of the contour difference of the i-th connected domain, and the mean of the contour difference of the i-th connected domain; it is negatively correlated with the stability of the i-th connected domain.

[0034] In one embodiment, the expression of the debris flow confidence is:

[0035] ; In the formula, represents the debris flow confidence of the i-th connected domain; is the discrete degree of the texture flow direction values ​​of all windows in the i-th connected domain; is the discrete degree of the contour difference of the i-th connected domain; is the mean value of the contour difference of the i-th connected domain; is the stability of the i-th connected domain; ρ is a preset value greater than zero.

[0036] In one embodiment, the method for detecting debris flow connected domains is as follows: for each frame image, a connected domain whose normalized debris flow confidence value is greater than a preset debris flow confidence threshold is determined as a debris flow connected domain; otherwise, it is determined as a non-debris flow connected domain.

[0037] This application has at least the following beneficial effects:

[0038] Based on the dynamic characteristics of debris flow disasters, this application uses the optical flow method to calculate the motion vectors of the corner points in the connected domain in the image, determine the average motion difference value of the connected domain, distinguish the connected domain with fluidity from the rest of the connected domains, and analyze the possibility that the connected domain is a debris flow connected domain and a water flow connected domain;

[0039] Furthermore, the stability of the circulation domain is determined by combining the characteristics of the debris flow and river boundary contours to analyze the significance of the connected domain boundary contour; by analyzing the difference between the texture information characteristics of the debris flow connected domain and the river connected domain, the texture flow direction value of the window in the connected domain is determined to characterize the possibility that the connected domain is a debris flow connected domain;

[0040] Furthermore, the debris flow confidence of the connected domain is determined based on the texture flow value of the window in the connected domain, the stability of the circulation domain, and the degree of change of the connected domain contour in adjacent frame images, which fully considers the characteristics of the circulation domain in a single frame image and the difference in the connected domain contour in adjacent frame images, solves the problem of possible misdetection caused by only performing feature recognition on a single image in previous image detection, and improves the accuracy of the analysis of the connected domain as a debris flow connected domain. This application improves the accuracy of feature recognition of remote sensing images for geological exploration and avoids the occurrence of misdetection. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0042] Figure 1 A flowchart of the steps of a method for identifying features of geological remote sensing image data for prospecting and mineral exploration provided in this application;

[0043] Figure 2 Schematic diagram of the process of determining the degree of stability;

[0044] Figure 3 Schematic diagram of the process of determining the confidence level of debris flow;

[0045] Figure 4 Schematic diagram of the process of obtaining debris flow confidence. DETAILED DESCRIPTION

[0046] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following is a detailed description of the method for identifying geological remote sensing image data features for prospecting and mineral exploration proposed in the present application, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0047] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0048] The following is a detailed description of a method for identifying features of geological remote sensing image data for prospecting and mineral exploration provided by the present application in conjunction with the accompanying drawings.

[0049] An embodiment of the present application provides a method for identifying features of geological remote sensing image data for prospecting and ore exploration. Specifically, the following method for identifying features of geological remote sensing image data for prospecting and ore exploration is provided. Figure 1 , the method comprises the following steps:

[0050] Step S1, obtaining the grayscale image of each frame in the high-altitude remote sensing image.

[0051] With the development of technology and the improvement of data acquisition equipment, the acquisition methods of remote sensing images have become more diversified. Among them, the use of drones to acquire remote sensing images can achieve more refined monitoring of surface information. Use drone remote sensing technology to obtain high-altitude remote sensing images within a monitoring period T, and perform frame de-framing on the high-altitude remote sensing images to obtain continuous M frames of images, with the time interval between two adjacent frames of images being t. Taking the a-th frame image as an example, image processing and feature recognition operations are performed on the a-th frame image to determine whether there is a debris flow area in the image. Since noise will inevitably appear during image shooting and transmission, it is necessary to perform noise reduction on the a-th frame image, and after noise reduction is completed, the image is converted into a grayscale image, which is recorded as grayscale image A. Among them, frame de-framing is a well-known technology and will not be repeated in this application.

[0052] In this embodiment, the monitoring period T is 1 hour, and the time interval t is 2 seconds. The values ​​of the monitoring period T and the time interval t are preset manually and can be set by the implementer. This application does not impose any special restrictions.

[0053] In this embodiment, a median filtering method is used to perform noise reduction processing on the a-th frame image. As other implementation methods, on the basis of being able to perform noise reduction processing on the a-th frame image, the implementer may use other existing technologies to perform noise reduction processing on the a-th frame image, such as a mean filtering method, a Gaussian filtering method, etc., and this application does not impose any special restrictions.

[0054] Step S2, obtaining each connected domain in each frame of grayscale image, determining the average motion difference value and stability of each connected domain, determining the texture flow direction value of each window, determining the contour difference of each connected domain, and obtaining the debris flow confidence of each connected domain.

[0055] Step S2.1, using a corner detection algorithm to obtain each corner point in each grayscale image, and obtaining each connected domain in each grayscale image according to the grayscale values ​​of each corner point and its neighboring pixel points in each grayscale image.

[0056] The Harris corner detection algorithm is used to obtain the corner points in the grayscale image A, and the above corner points are sorted from large to small according to the grayscale value. The first L corner points are used as the initial seed points. During the growth process, for each growing pixel point and its adjacent pixel point, the absolute value of the difference in the grayscale value between the two pixels is used as the similarity value. When the normalized value of the similarity value is greater than the preset similarity threshold, the growth is stopped, and the growth regions obtained by the region growing algorithm are used as the connected domains in the grayscale image A. Both the Harris corner detection algorithm and the region growing algorithm are well-known technologies and will not be described in detail in this embodiment.

[0057] In this embodiment, L is 10, and the similarity threshold is 0.2. The values ​​of L and the similarity threshold are preset manually and can be set by the implementer. This application does not impose any special restrictions.

[0058] Step S2.2, determining the motion vector of each corner point in each grayscale image according to the position change of each corner point in each grayscale image in the adjacent grayscale image; determining the average motion difference value of each connected domain based on the difference in motion vectors of the corner points between each connected domain and all other connected domains in the grayscale image where it is located.

[0059] Take the grayscale image A and all the corner points in the next grayscale image B as input, and use the KLT optical flow method to calculate the motion vector of each corner point, which is recorded as . The motion vector obtained by the KLT optical flow method is usually expressed as a two-dimensional vector, in which u and v represent the displacement of the corner point in the horizontal direction and the vertical direction respectively. The KLT optical flow method is a well-known technology and will not be described in detail in this application. Since fixed objects such as mountains, land and trees are stationary, the motion vectors of the corner points in the connected domains of these fixed objects are highly consistent, and the connected domains of the above-mentioned fixed objects are recorded as solid connected domains. However, the water flow formed after the rapid melting of rain and snow is one of the important causes of debris flow disasters. The water flow washes away the mud and stones and gradually forms debris flow disasters. Therefore, special attention should be paid to the connected domains of non-fixed objects to detect debris flow disasters. Because the internal water flow in the debris flow connected domain and the river connected domain has an obvious sense of flow, the motion vector will be greatly different compared with the connected domain of fixed objects.

[0060] Based on the above analysis, the average motion difference value of each connected domain is determined based on the difference in motion vectors of the corner points between each connected domain and all other connected domains in the grayscale image where it is located. The expression is:

[0061] ; In the formula, is the average motion difference value of the i-th connected domain, representing the average motion difference between the i-th connected domain and the rest of the connected domains; N is the number of connected domains in the grayscale image A; is the mean of the motion vectors of all corner points in the i-th connected domain; is the mean of the motion vectors of all corner points in the rth connected domain in the grayscale image A.

[0062] It should be noted that: since there are many types of solids such as mountains, rocks, and trees in remote sensing images, there will be more solid connected domains obtained after performing the region growing algorithm. And because the motion vector of the corner point of the solid connected domain is very small, if the i-th connected domain is a solid connected domain, the smaller the difference between the mean of the i-th connected domain and the mean of the other connected domains, The smaller the value of The smaller the value of is; on the contrary, if the i-th connected domain is a non-solid connected domain, then The larger the value of .

[0063] Step S2.3, determining the stability of each connected domain based on the gray value distribution of the pixels in each connected domain and the difference between the average motion difference value between each connected domain and all connected domains in the gray image where the connected domain is located.

[0064] Debris flow is a high-speed flowing material composed of a mixture of water, silt, rock fragments and other materials. The flow path of a debris flow will be strongly affected by the terrain, and may form obvious ditches in the valley, but the boundaries of the ditches may become blurred due to the erosion of the debris flow and the accumulation of silt and stones. In addition, due to the fast flow rate of the debris flow, the silt and debris carried may be redistributed during the flow, causing the boundaries of the debris flow to constantly change, resulting in a relatively blurred outline. Generally speaking, the boundary outline of a river is clearer than that of a debris flow, because the river is a natural waterway formed under the action of long-term water flow, and the outline is constrained by the river bank, and the river bank reflects the stable flow path of the river to a certain extent. The flow rate of the river is relatively slow, and the silt and sediment carried are deposited on the riverbed and riverbank, forming a relatively stable and clear river outline. A grayscale run matrix is ​​constructed based on the grayscale values ​​of all pixels in the i-th connected domain, where the run direction of this embodiment is The implementation of the run direction can be set by the implementer according to the actual situation, and this application does not impose any special restrictions on this.

[0065] Based on the above analysis, the stability of each connected domain is determined based on the gray value distribution of pixels in each connected domain and the difference in the average motion difference between each connected domain and all connected domains in the gray image where it is located. The expression is:

[0066] ; In the formula, is the stability of the i-th connected domain; is the maximum run length of the grayscale run matrix of the i-th connected domain; is the mean of the gradient values ​​of all edge pixels in the i-th connected domain; is the average motion difference value of the i-th connected domain; is the mean of the average motion difference values ​​of all connected domains in the grayscale image A where the i-th connected domain is located; h is the number of corner points of the i-th connected domain; τ is a preset value greater than zero in order to avoid the denominator being 0. The value of τ is preset manually and can be set by the implementer. In this embodiment, the value of τ is 0.001.

[0067] In this embodiment, the Sobel operator is used to calculate the gradient value of the edge pixel point. As other implementation methods, on the basis of being able to calculate the gradient value of the edge pixel point, the implementer may use other existing technologies to calculate the gradient value of the edge pixel point, such as the Roberts operator, the Prewitt operator, etc., and this application does not impose any special restrictions.

[0068] It should be noted that if the i-th connected domain is a debris flow connected domain, since the debris flow contains a large amount of sediment and gravel, the change of the gray value of the pixels inside the debris flow connected domain is more complex and frequent, and the conversion between different gray values ​​is more common. The value of is small; if the i-th connected domain is a river connected domain, since the river is clear, the grayscale values ​​of the pixels inside the river connected domain are relatively consistent. Even if reflection occurs, it is a large-area reflection. There are many pixels with the same grayscale level in the river connected domain, so The value of is larger; The larger the value is, the smaller the possibility that it is a debris flow connected domain. is the mean of the gradient values ​​of all edge pixels of the ith connected domain, reflecting the overall prominence of the connected domain. Ordinary river areas have significant boundary contours because of their slow flow and the constraints of the river bank. However, debris flow areas have fast flow rates, which will wash away vegetation and expose the land under the vegetation. Since the color of the land is similar to that of the debris flow, and due to the erosion and accumulation of sediment, the boundary contour of the debris flow will be relatively blurred. Therefore, when The larger the value of is, the more significant the boundary contour of the i-th connected domain is, and the less likely it is that the connected domain is a debris flow connected domain.

[0069] If the i-th connected domain is a debris flow connected domain, due to the faster flow velocity of the debris flow, the greater the difference in the average motion difference between this connected domain and the other connected domains, the greater the difference in the average motion difference between this connected domain and the other connected domains. The value of is large. If the i-th connected domain is a river connected domain, since the river flow rate is relatively slow, The value of is small. Since the river flow is slow and the waves are small, the river surface is smooth overall, and only some wave areas are areas with drastic changes in characteristics. Therefore, the number of corner points in the river connection domain is small, and the value of h is small; since the debris flow has a fast flow rate and there are many floating objects such as stones and vegetation, the number of corner points in the debris flow flow domain is large, and the value of h is large. At this time, if The larger the value of is, the less likely the i-th connected domain is to be a debris flow connected domain, and the more likely it is to be a river connected domain. The schematic diagram of the stability determination process is as follows: Figure 2 shown.

[0070] Step S2.4, construct a window with each pixel in each connected domain as the center, and determine the texture flow value of each window based on the direction angle of the texture in each window, the MLBP value of the pixel at the center of the window, and the gray value distribution of the pixel.

[0071] Since debris flow carries a large amount of vegetation, rocks and other objects, there will be many objects with inconsistent colors and shapes in the debris flow. These objects float on the water surface, causing a large number of ripples on the water surface, and also causing the color of the water surface to have certain differences. And because the characteristics of many objects are different, and the debris flow carries a large amount of sediment during the flow process, the image features of the debris flow area in the image are inconsistent in grayscale value and chaotic texture; because the debris flow has fluidity and flows in one direction as a whole, the directional features in the connected domain of the debris flow have greater consistency. With each pixel point in the i-th connected domain as the center, an N*N window is constructed respectively. All the pixels in each window are used as input, and the MLBP algorithm is used to calculate the MLBP value of the central pixel point of each window, and the MLBP value is used to describe the texture information in each window. At the same time, the second-order moment technology is used to calculate the texture direction angle in each window according to the area of ​​each window and the coordinates of the central pixel point of each window. Among them, the MLBP algorithm and the second-order moment technology are both known technologies, and this application will not repeat them.

[0072] In this embodiment, the value of N is 5. The value of N is preset manually and can be set by the implementer. This application does not impose any special restrictions.

[0073] Based on the above analysis, the texture flow value of each window is determined based on the direction angle of the texture in each window, the MLBP value of the central pixel of the window, and the gray value distribution of the pixel. The expression is:

[0074] ; In the formula, is the texture flow value of the window centered at pixel k; is the MLBP value of pixel k; is the mean of the grayscale values ​​of all pixels in the window centered on pixel k; σ is a preset constant greater than zero, the purpose of which is to avoid the denominator being 0. The value of σ can be set by the implementer. In this embodiment, the value of σ is 0.001; is the texture direction angle in the window centered at pixel k; It is to convert the texture direction angle in the window centered on pixel point k into radian values, which can be calculated with other coefficients.

[0075] Furthermore, the discrete degree of the texture flow direction values ​​of all windows in the i-th connected domain is calculated. Since the flow direction inside the debris flow connected domain is roughly the same, the value difference of the texture direction angle in all windows is not large. Since the texture information and grayscale distribution between different windows in the debris flow connected domain are quite different, the texture flow direction values ​​of different windows are quite different. The greater the discrete degree of the texture flow direction value, the greater the degree of chaos of the texture and grayscale value inside the connected domain, and the more it conforms to the characteristics of the debris flow connected domain.

[0076] In this embodiment, the degree of discreteness of the grain flow direction value is the standard deviation. As other implementation methods, on the basis of being able to measure the uneven distribution of the grain flow direction value, the implementer may adopt other existing technologies for measurement, such as variance, coefficient of variation, etc., and this application does not impose any special restrictions.

[0077] Step S2.5, based on the motion vectors of all corner points in any connected domain and the stability level, respectively obtain the matching connected domains of any connected domain from the remaining grayscale images.

[0078] The mean of the motion vectors of all corner points of each connected domain is recorded as the motion mean, and the sum of the motion mean and the stability of each connected domain is calculated. Taking the i-th connected domain in the a-th frame grayscale image as an example, the connected domains in the remaining frames of images with the smallest difference in the sum with the i-th connected domain are taken as the initial matching connected domains of the i-th connected domain;

[0079] The pixels on the outermost four sides of each frame of grayscale image are recorded as grayscale image boundary pixels, and any frame of grayscale image is recorded as a. Starting from the a-th frame of grayscale image, all the pixels of the matching connected domain of the i-th connected domain of the a-th frame of grayscale image in each grayscale image are traversed forward and backward frame by frame. If the grayscale image boundary pixels appear in the pixels, the traversal is stopped, and the initial matching connected domain of the i-th connected domain in each frame of grayscale image between the forward traversal stop and the backward traversal stop is used as the matching connected domain of the i-th connected domain.

[0080] Step S2.6, based on the discreteness of all the texture flow direction values ​​in each connected domain, the distribution of the contours of each connected domain and its matching connected domain, and combined with the stability, determine the debris flow confidence of each connected domain.

[0081] The grayscale images of the i-th connected domain and its matching connected domains are recorded as the complete contour image of the i-th connected domain; the i-th connected domain or the matching connected domain of the i-th connected domain in the complete contour image of the i-th connected domain in each frame is recorded as the i-th connected domain. For the complete contour images of the i-th connected domain in any two adjacent frames, the Hu moment function is used to match the contours of the i-th connected domain in the above two complete contour images, and the difference between the Hu moments of the same connected domain in the above two complete contour images is obtained, which is recorded as the contour difference of the i-th connected domain. The discrete degree and mean value of the contour difference of the i-th connected domain between all any two adjacent complete contour images are calculated; if the discrete degree is smaller, it means that the contour difference of the i-th connected domain between any two adjacent frames is more consistent; if the discrete degree is larger, it means that the contour change degree of the i-th connected domain is greater, and it is more consistent with the characteristics of the debris flow connected domain; if the mean value of the contour difference is larger, it means that the contour difference of the i-th connected domain between any two adjacent frames is larger. Among them, the Hu moment function is a well-known technology and will not be described in detail in this application.

[0082] In this embodiment, the degree of discreteness of the contour difference is the standard deviation. As other implementation methods, on the basis of being able to measure the uneven distribution of contour differences, the implementer may use other existing technologies for measurement, such as variance, coefficient of variation, etc., and this application does not impose any special restrictions.

[0083] Based on the above analysis, based on the discrete degree of all the texture flow direction values ​​in each connected domain, and the distribution of the contours of each connected domain and its matching connected domain, and combined with the stability, the debris flow confidence of each connected domain is determined, and the expression is:

[0084] ; In the formula, represents the debris flow confidence of the i-th connected domain; is the discrete degree of the texture flow direction values ​​of all windows in the i-th connected domain; is the discrete degree of the contour difference of the i-th connected domain; is the mean value of the contour difference of the i-th connected domain; is the stability of the ith connected domain; ρ is a preset value greater than zero, the purpose of which is to avoid the denominator being 0. The value of ρ is preset manually and can be set by the implementer. In this embodiment, the value of ρ is 0.001.

[0085] It should be noted that: if the discrete degree of the texture flow value of all windows in the i-th connected domain The larger the value, the greater the degree of disorder of the texture and grayscale value inside the connected domain, and the greater the possibility that the connected domain is a debris flow connected domain; since the flow rate of debris flow is very fast, the boundary contour changes quickly, and the river connected domain has a stable boundary contour, so if the discrete degree of the contour difference of the connected domain is With the mean The larger the value is, the more likely it is that the connected domain is a debris flow connected domain; when the stability of the connected domain is smaller, the more likely it is that the connected domain is a debris flow connected domain; therefore, when the debris flow confidence value of the connected domain is larger, the more likely it is that the connected domain is a debris flow connected domain. The schematic diagram of the debris flow confidence determination process is as follows Figure 3 The schematic diagram of the debris flow confidence acquisition process is shown in Figure 4 shown.

[0086] Step S3, detecting the debris flow connected domain according to the debris flow confidence of the connected domain, so as to realize the recognition of the geological exploration remote sensing image features.

[0087] According to the method of calculating the debris flow confidence of the i-th connected domain, the debris flow confidence of the remaining connected domains is calculated. The connected domain whose normalized value of the debris flow confidence is greater than the preset debris flow confidence threshold is determined as a debris flow connected domain, otherwise, it is determined as a non-debris flow connected domain. When a debris flow connected domain is detected in the image, it means that a debris flow disaster has occurred at the location where the image was taken. The location where the image was taken is marked as a debris flow prone area to help practitioners avoid debris flow prone areas when prospecting.

[0088] In this embodiment, the debris flow confidence threshold is 0.7. The value of the debris flow confidence threshold is preset manually and can be set by the implementer. This application does not impose any special restrictions.

[0089] In summary, based on the characteristics of debris flow disasters being mobile, the present application uses the optical flow method to calculate the motion vectors of the corner points in the connected domain in the image, determines the average motion difference value of the connected domain, distinguishes the connected domain with fluidity from the rest of the connected domains, and analyzes the possibility that the connected domain is a debris flow connected domain and a water flow connected domain;

[0090] Furthermore, the stability of the circulation domain is determined by combining the characteristics of the debris flow and river boundary contours to analyze the significance of the connected domain boundary contour; by analyzing the difference between the texture information characteristics of the debris flow connected domain and the river connected domain, the texture flow direction value of the window in the connected domain is determined to characterize the possibility that the connected domain is a debris flow connected domain;

[0091] Furthermore, the debris flow confidence of the connected domain is determined based on the texture flow value of the window in the connected domain, the stability of the circulation domain, and the degree of change of the connected domain contour in adjacent frame images, which fully considers the characteristics of the circulation domain in a single frame image and the difference in the connected domain contour in adjacent frame images, solves the problem of possible misdetection caused by only performing feature recognition on a single image in previous image detection, and improves the accuracy of the analysis of the connected domain as a debris flow connected domain. This application improves the accuracy of feature recognition of remote sensing images for geological exploration and avoids the occurrence of misdetection.

[0092] It should be noted that the above sequence of the embodiments of the present application is for description only and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0093] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0094] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Modifications to the technical solutions recorded in the aforementioned embodiments, or equivalent replacement of some of the technical features therein, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for identifying features of geological remote sensing image data for prospecting and prospecting, characterized in that: The method comprises the following steps: Obtain the grayscale image of each frame in the high-altitude remote sensing image; A corner point detection algorithm is used to obtain each corner point in each grayscale image, and each connected domain in each grayscale image is obtained according to the grayscale values ​​of each corner point and its neighboring pixel points in each grayscale image; Determine the motion vector of each corner point in each grayscale image according to the position change of each corner point in each grayscale image in the adjacent grayscale image; determine the average motion difference value of each connected domain based on the difference in motion vectors of the corner points between each connected domain and all other connected domains in the grayscale image where it is located; Based on the gray value distribution of the pixels in each connected domain and the difference between the average motion difference value between each connected domain and all the connected domains in the gray image where it is located, the stability of each connected domain is determined; a window is constructed with each pixel in each connected domain as the center, and the texture flow value of each window is determined based on the direction angle of the texture in each window, the MLBP value of the pixel at the center of the window, and the gray value distribution of the pixel; Based on the motion vectors of all corner points in any connected domain and the stability level, respectively obtaining matching connected domains of any connected domain from the remaining grayscale images; Based on the discreteness of all the texture flow direction values ​​in each connected domain, and the distribution of the contours of each connected domain and its matching connected domain, and combined with the stability, determine the debris flow confidence of each connected domain; Detecting a debris flow connected domain based on the debris flow confidence; The process of determining the stability is: A grayscale run matrix is ​​constructed according to the grayscale values ​​of all pixels in the i-th connected domain, and the gradient value of each edge pixel in the i-th connected domain is calculated using the edge detection operator, and the average value of the gradient values ​​of all edge pixels in the i-th connected domain is calculated; Calculate the average of the average motion difference values ​​of all connected domains in the grayscale image where the i-th connected domain is located, recorded as the motion difference mean; calculate the difference between the average motion difference value of the i-th connected domain and the motion difference mean; The stability of the i-th connected domain is positively correlated with the maximum run length of the grayscale run matrix and the average value, and negatively correlated with the difference and the number of corner points of the i-th connected domain.

2. A method for identifying features of geological remote sensing image data for prospecting and mineral exploration as claimed in claim 1, characterized in that: The process of obtaining each connected domain in each grayscale image is as follows: For each grayscale image, all corner points in the grayscale image are arranged from large to small according to the grayscale value, and the first preset number of corner points are used as initial seed points. Combined with the region growing algorithm, the absolute value of the difference in grayscale value between the growing pixel point and the adjacent pixel point is calculated, and the absolute value of the difference is used as the similarity between the growing pixel point and the adjacent pixel point. When the normalized value of the similarity is greater than the preset similarity threshold, the growth is stopped, and the growth regions obtained by using the region growing algorithm are used as connected domains.

3. A method for identifying features of geological remote sensing image data for prospecting and mineral exploration as claimed in claim 1, characterized in that: The method for determining the motion vector is: taking all corner points in each grayscale image and the adjacent next frame grayscale image as input of the optical flow method, and outputting the motion vector of each corner point in each grayscale image.

4. A method for identifying features of geological remote sensing image data for prospecting and mineral exploration as claimed in claim 1, characterized in that: The process of determining the average motion difference value is as follows: The mean of the motion vectors of all corner points of each connected domain is recorded as the motion mean; Any connected domain in any grayscale image is denoted as i, and the difference in the motion mean between the i-th connected domain and the other connected domains in the grayscale image where it is located is denoted as the motion difference; the mean of all the motion differences of the i-th connected domain is taken as the average motion difference value of the i-th connected domain.

5. The method for identifying features of geological remote sensing image data for prospecting and mineral exploration according to claim 1, characterized in that: The process of determining the texture flow direction value is as follows: The MLBP algorithm is used to calculate the MLBP value of the central pixel of each window, and the average of the grayscale values ​​of all pixels in each window is recorded as the grayscale mean; the second-order moment technology is used to calculate the texture direction angle in each window according to the area of ​​each window and the coordinates of the central pixel of each window, and the texture direction angle is converted into a radian value; The texture flow direction value is positively correlated with the MLBP value and the grayscale mean value, and negatively correlated with the radian value.

6. A method for identifying features of geological remote sensing image data for prospecting and mineral exploration as claimed in claim 4, characterized in that: The process of determining the matching connected domain is as follows: Calculating the sum of the motion mean and the stability of each connected domain; taking the connected domains in the remaining frame images with the smallest difference in the sum with the i-th connected domain as the initial matching connected domains of the i-th connected domain; The pixels on the outermost four sides of each frame of grayscale image are recorded as grayscale image boundary pixels, and any frame of grayscale image is recorded as a. Starting from the a-th frame of grayscale image, all the pixels of the matching connected domain of the i-th connected domain of the a-th frame of grayscale image in each grayscale image are traversed forward and backward frame by frame. If the grayscale image boundary pixels appear in the pixels, the traversal is stopped, and the initial matching connected domain of the i-th connected domain in each frame of grayscale image between the forward traversal stop and the backward traversal stop is used as the matching connected domain of the i-th connected domain.

7. A method for identifying features of geological remote sensing image data for prospecting and mineral exploration as claimed in claim 4, characterized in that: The determination process of the debris flow confidence is as follows: The grayscale images of the i-th connected domain and its matching connected domains are recorded as the complete contour image of the i-th connected domain; The i-th connected domain in the complete contour image of the i-th connected domain in each frame or the matched connected domain of the i-th connected domain is recorded as the i-th connected domain, and the difference between the Hu moments of the i-th connected domain in any two adjacent frames of the complete contour image of the i-th connected domain is recorded as the contour difference; The debris flow confidence of the i-th connected domain is positively correlated with the discrete degree of the texture flow direction values ​​of all windows in the i-th connected domain, the discrete degree of the contour difference of the i-th connected domain, and the mean of the contour difference of the i-th connected domain; it is negatively correlated with the stability of the i-th connected domain.

8. A method for identifying features of geological remote sensing image data for prospecting and mineral exploration as claimed in claim 7, characterized in that: The expression of debris flow confidence is: ; In the formula, represents the debris flow confidence of the i-th connected domain; is the discrete degree of the texture flow direction values ​​of all windows in the i-th connected domain; is the discrete degree of the contour difference of the i-th connected domain; is the mean value of the contour difference of the i-th connected domain; is the stability of the i-th connected domain; ρ is a value that is preset to be greater than zero.

9. A method for identifying features of geological remote sensing image data for prospecting and mineral exploration as claimed in claim 1, characterized in that: The method for detecting debris flow connected domains is as follows: for each frame image, a connected domain whose normalized debris flow confidence value is greater than a preset debris flow confidence threshold is determined as a debris flow connected domain, otherwise, it is determined as a non-debris flow connected domain.

Images