CN105426881A - Mountain background thermal field model constrained underground heat source daytime remote sensing detection locating method - Google Patents

Abstract

The invention discloses a daytime remote sensing detection and positioning method for underground heat sources constrained by a mountain background thermal field model, which belongs to the intersecting field of remote sensing technology, physical geography and pattern recognition. The field model, constrained by the thermal field model, performs background filtering on the original remote sensing infrared image to clearly reveal the location of the underground target. The invention includes the steps of establishing a mountain background thermal field model, the step of mapping from 8 bits to 16 bits of the mountain background thermal field, using the mapped mountain background thermal field model to perform background filtering, and the underground object space constraint mean value clustering detection and positioning step. The present invention uses the background thermal field of the mountain to build a model, uses the background of the real mountain to perform gray scale mapping on the model, ensures that the established thermal field model is close to the real thermal field of the mountain background, and finally performs background filtering processing on the real remote sensing infrared image to determine the underground target s position.

Description

A daytime remote sensing detection and location method for underground heat sources constrained by a mountain background thermal field model

technical field

The invention belongs to the intersection field of remote sensing technology, physical geography and pattern recognition, and specifically relates to a daytime remote sensing detection and positioning method for underground heat sources constrained by a mountain background thermal field model.

Background technique

The development of human beings is inseparable from various natural resources. In order to obtain abundant underground resources, mineral resources, and groundwater resources, it is necessary to build a large number of underground facilities. Therefore, the detection technology of underground objects such as underground buildings and underground facilities is becoming increasingly important. All objects with a temperature higher than absolute zero can produce thermal radiation. Underground targets and surrounding areas have different thermodynamic characteristics. Under the action of the external environment, the existence of underground targets will affect the internal heat conduction process of the surrounding area, resulting in Time-varying temperature differences appear between the place and the surrounding background. Statistically speaking, the temperature field of the underground target is higher or lower than the temperature field of the mountain background. Therefore, these subsurface targets can be considered as subsurface heat sources distinct from background heat sources. Compared with other detection methods, infrared technology detection has certain advantages. Non-remote sensing detection methods cannot achieve large-scale simultaneous observation, and are difficult to use in harsh environments, and the speed of obtaining information is slow, time-consuming and laborious. The existing conventional remote sensing detection methods are mainly aimed at the conditional objects on the surface or water surface, and the acquisition of information is limited by the natural environment and background environment, and it is impossible to detect deep underground heat source targets. At present, electromagnetic induction technology can only detect metal targets in the shallow subsurface, and is also vulnerable to metal fragments scattered underground. Therefore, infrared technology detection has become an effective means of underground target detection.

At present, there are certain studies on the detection of underground targets at home and abroad. In China, it mainly focuses on the detection of shallow targets and the detection of targets under multi-temporal images, and most of these underground targets are large-scale underground heat sources. There are no relevant reports on the detection of deep (more than 10m from the surface) underground heat source targets in many countries, especially small-scale underground heat source targets. There are studies abroad on the use of airborne mid-wave and long-wave infrared scanning sensors to detect underground targets, but there is no relevant report on the use of remote sensing images for underground target detection. For the detection of planar strip targets, the existing pattern recognition methods do not use the background filtering method. Moreover, what it detects is only the suspected target area, and does not accurately locate the location of the underground target. Not only that, the false alarm rate of the detected suspected target area is not ideal, the false alarm rate is high, and the positioning accuracy is not high.

Contents of the invention

The invention proposes a daytime detection and positioning method for underground distributed heat sources in the deep layer (distance from the surface is greater than 10m) under the constraint of the mountain background thermal field model, and solves the existing problem of only detecting and positioning shallow underground heat sources , use simulation software to simulate the thermal field of the mountain, analyze and obtain the thermal field model of the mountain body background, use the real infrared image to map and unify the mountain thermal field model, and use the mapped model to perform background filtering on the real infrared image processing to reduce the influence of the background thermal field of the mountain body on the detection, and finally accurately locate the location of the underground target.

The invention provides a daytime detection and positioning method for underground distributed heat sources based on the constraints of a mountain background thermal field model. When establishing the mountain background thermal field model, appropriate simplification is made according to the actual mountain thermal field to obtain a simplified model. The specific steps are as follows:

(1) The establishment of the mountain background thermal field model includes the following sub-steps:

(1.1) Steps for building a mountain model

(1.1.1) The three-dimensional digital elevation model of the mountain is obtained through remote sensing mapping, and the altitude data information of the real mountain is obtained.

(1.1.2) The construction of ANSYS geometric mountain model is composed of points, lines, surfaces and volumes, and points (coordinates) are the basis for constructing geometric models. Therefore, according to the above-mentioned altitude information, the creation of the geometric mountain is to generate a closed curve from the key points, and the closed curve generates a plane, and then the geometric mountain is surrounded by the closed surface. The geometric mountain constitutes the entire mountain model.

(1.2) Finite element mesh division steps of mountain model

Anasy finite element meshing is a crucial step for numerical simulation analysis. Since the model of the mountain is not regular, free meshing can automatically generate triangular or tetrahedral meshes on the surface freely. Automatically generate tetrahedral grids on the body, and artificially control the intelligent size.

(1.3) Mountain model boundary condition setting and solution steps

Set the load boundary conditions for the above-mentioned mountain after grid division, use the basic heat transfer physical basis of mountain heat conduction and mountain-air heat convection, and set the heat transfer rate according to the heat conductivity K of the mountain and the heat convection rate Φ of the mountain-air Parameters, the temperature field distribution of the mountain is obtained through Ansys solution calculation, and finally the temperature field distribution of the mountain is mapped to the grayscale map and the temperature resolution is adjusted, and finally the mountain thermal field model is obtained.

(2) 8-bit to 16-bit mapping of the mountain background thermal field, including the following sub-steps:

Since the established image of the mountain background infrared thermal field model is 8-bit, while the real mountain background thermal field image is 16-bit, it is necessary to perform an 8-bit to 16-bit comparison between the mountain background thermal field model obtained above and the real mountain background thermal field. Mapping processing. In the case that the thermal field of the model is roughly the same as the distribution of the real thermal field, it is ensured that the thermal field of the model is closer to the thermal field of the real mountain.

(2.1) Steps to determine the gray value range of the real mountain background

The thermal field of the real mountain background is affected by many external factors, such as houses and roads. Because these external factors account for a small proportion in the overall background, they can be regarded as interference. Therefore, regional constraint processing is required, and threshold processing is performed through histogram statistics to eliminate the interference of external factors. The specific process is as follows:

(2.1.1) Let the variable r represent the pixel gray level in the image. In the discrete situation, use r _k to represent the discrete gray level, and use P(r _k ) to represent the probability density function. The following formula holds:

$\begin{array}{cc}{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}{\mathrm{<span\; class="notranslate">\; no</span>}}& <span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; )</span>\end{array}$

k=0,1,2...l-1

In the formula, n _k is the number of pixels that appear r _k gray in the image, n is the total number of pixels in the image, is the frequency in probability theory, and l is the total number of gray levels.

It is known that the influence of external factors accounts for P% of the area in the entire image, and the following formula is given:

P _r (r _k )≥P

Accumulate the gray histogram sequentially. If the accumulated value is greater than or equal to the proportion of the target object, stop accumulating and record the value of r _k as the guiding value of the background.

(2.2) Mapping correction steps of mountain thermal field model

The gray value of the background of the mountain thermal field model is corrected by linear mapping according to the value obtained above and the minimum gray value of the image. The specific formula is as follows:

$\mathrm{<span\; class="notranslate">\; o</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}}{{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; h</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; o</span>}}_{\mathrm{<span\; class="notranslate">\; h</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; o</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; o</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}$

Among them, I is the gray value of the thermal field model, I _l is the lowest brightness gray value of the thermal field model, I _h is the highest brightness gray value of the thermal field model, O _l is the lowest brightness gray value of the real mountain infrared image value, _Oh is the r _k obtained above, and O is the mountain background model after mapping correction.

(3) Use the mapped mountain background thermal field model to perform background filtering steps

Due to the angle of remote sensor observation and the influence of the sun's irradiation angle, the mountain can be divided into sunny and shady sides. The part of the mountain that is directly irradiated by the sun is called the sunny side, and the part that is not directly irradiated by the sun is called the shady side. The mountain background and the surrounding environment have thermal radiation, and there are differences in thermal radiation between direct and indirect exposures. Not only that, the thermal field of the mountain background during the day is different from that at night, and the present invention is aimed at the daytime detection and positioning of underground heat sources constrained by the thermal field model of the mountain background. First find the pixel point at the boundary between the shaded and sunny sides, use the least squares fitting method to fit the dividing line between the shaded and sunny sides, then perform grayscale compensation on the shaded side, and finally perform background filtering.

(3.1) Extraction steps of the dividing line between the negative side and the positive side of the original infrared image

(3.1.1) Utilize the difference between the gray level difference between the shaded side and the sunny side to determine the pixel point at the boundary. According to the information of the image data, it can be determined that the boundary line is east-west, so it is only necessary to compare the gray value of the upper and lower pixel points. If The gray difference between the upper and lower adjacent pixels is greater than K, the formula is as follows:

G(x,y)>G(x,y-1)+K

G(x,y)>G(x,y-2)+K

G(x,y)>G(x,y-3)+K

G(x,y)>G(x,y-1)+K

If the above four inequalities are true, consider (x, y) as the pixel points near the dividing line, and traverse the whole image to get all the pixel points near the dividing line.

(3.1.2) Next, carry out the cubic polynomial least squares fitting linear processing on the pixel points near all the boundary lines obtained above, and the specific process is as follows:

in is the cubic polynomial fitted by the least squares, and err is the error objective function. By minimizing err to achieve the optimal cubic polynomial fitting, the final dividing line is obtained.

(3.2) Steps to eliminate direct sunlight energy on the sunny side

After obtaining the shaded area and sunny area of the infrared image according to the dividing line, the following mapping strategy is used to eliminate the gray value of the sunny area of the infrared image:

${{\mathrm{<span\; class="notranslate">\; f</span>}}^{<span\; class="notranslate">\; \&prime;</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; \&lsqb;</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; +</span>\frac{{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}}{{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}}<span\; class="notranslate">\; *</span>{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; \&rsqb;</span>$

D＝F _n ' _oshadow (i,j)-F _noshadow (i,j)

In the formula, F' _noshadow (i, j) is the gray value of the sunny area of the infrared image that is not directly irradiated by the sun, F _noshadow (i, j) is the gray value of the area of the infrared image that is directly irradiated by the sun, and m _shadow and σ _shadow are the infrared The mean and variance of the gray value of the sunny area of the image, m _noshadow and σ _noshadow are the mean and variance of the gray value of the shaded area of the infrared image adjacent to the non-infrared image, A is the compensation intensity coefficient, and D is the energy gray value of the sun directly irradiating the sun .

After getting the gray value of direct sunlight from the above, iterate over all detected shades

In the surface area, the gray value compensation D of each point in the infrared image is subtracted from the sunny area of the infrared image to obtain the gray image after eliminating the influence of sunlight. The influence of solar radiation was not considered before, and the component of solar radiation was filtered out, and the energy irradiated by the sun on the sun's surface was filtered out. The thermal radiation component caused by solar radiation was absorbed by the sun's surface. Under the condition of thermal balance, the direct sunlight was filtered out. weight.

(3.3) Background filtering steps of real infrared images of mountains

The mountain background thermal field model obtained by simulation is used as the background of the real mountain infrared image. Since the target heat radiation in the mountain conforms to the mathematical model of heat conduction, it is assumed that at a time t ₀ the target position (x ₀ , y ₀ , z ₀ ) The thermal radiation surface is BT(x,y,z,t)

BT(x,y,z,t)

=B(x,y,z,t)+T(x,y,z,t)*R _b (x,y,z,t)

+A(x,y,z,t)+δ(x,y,z,t)

At this moment, the target background body radiation field BT(x, y, z, t), the multi-media body radiation field B(x, y, z, t), the target body radiation field distorted by the multi-media body T(x, y,z,t)*R _b (x,y,z,t), the amount of radiation δ(x,y,z,t) lost by the contact surface between water body/ground body and air and the influence of solar radiation A(x,y , z, t) co-generated.

The influence of the target background body radiation field is mainly produced by B(x,y,z,t) and A(x,y,z,t), so the following formula is given:

T(x,y,z,t)=BT(x,y,z,t)-k*B(x,y,z,t)-A(x,y,z,t)

Where T(x,y,z,t) is the approximate thermal radiation field of the target, BT(x,y,z,t) is the radiation field of the target background body, and B(x,y,z,t) is the multi-media body Background radiation field, A(x, y, z, t) is the energy field affected by solar radiation, and k is the adjustable coefficient of the background radiation field. Finally, the background filtered image of the real mountain infrared image is obtained.

(4) Detection and positioning steps of underground target space constrained mean value clustering

From the image after background filtering, select s image blocks to be recognized, and the template size is 3*3, which are respectively b ₁ , b ₂ , b ₃ ,...,b _s , including the target image area below and the following No target image area, avoid the influence of houses and roads when selecting.

The road segment b ₁ , b ₂ , b ₃ ,..., b _s are divided into two types, the area with target image and the area without target image, by means of space-constrained mean clustering algorithm. The specific implementation process of the spatially constrained mean clustering algorithm is as follows:

Step1: For all sample points b _i , calculate the distance ratio

${\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; the\; s</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}\frac{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}{\underset{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; the\; s</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}$

Select the point b _i with the smallest V _i as the first centroid, and set q=1;

Step2: For p=1,2, assign b _i , i=1,2,...,s to the nearest class, and update the class center i=1,2, N _i is the number of samples of the i-th class;

Step3: Set q=q+1, if q>2, the algorithm stops;

Step4: Select the best initial center point of the next class as For the smallest point b _i , turn to Step2.

The result of clustering is obtained through the above formula, the category with large gray value is regarded as a category of suspected underground targets, and the category with small gray value is regarded as a category of non-suspected underground targets. Finally, the position of the underground target is obtained through the spatial constraint clustering algorithm.

Description of drawings

Fig. 1 is a schematic flow chart of the present invention;

Figure 2(a) is a schematic diagram of mountain elevation information;

Figure 2(b) is a schematic diagram of mountain contours;

Figure 2(c) is a schematic diagram of the mountain before grid division;

Figure 2(d) is a schematic diagram of the mountain after grid division;

Fig. 3 is the schematic diagram of the thermal radiation model of the mountain;

Figure 4 is a real infrared image of the mountain;

Fig. 5 is the image of the background thermal field model after mapping;

Fig. 6 is a schematic diagram of pixels near the dividing line of the Yin-Yang surface of the mountain;

Figure 7 is a schematic diagram of the dividing line between the Yin and Yang sides of the mountain;

Figure 8 is the infrared image of the mountain body after eliminating the influence of direct sunlight;

Fig. 9 is the infrared image after background filtering based on the thermal field model;

Fig. 10 is a schematic diagram of underground target detection results.

detailed description

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not constitute a conflict with each other.

The process flow of the present invention is shown in Figure 1, wherein the specific implementation method includes the following steps. The present invention includes the steps of establishing a mountain background thermal field model, the step of mapping from 8 bits to 16 bits of the mountain background thermal field, the step of background filtering using the mapped mountain background thermal field model, and the steps of detecting and locating the underground object space-constrained mean value clustering:

(1) The establishment of the mountain background thermal field model includes the following sub-steps:

(1.1) Steps for building a mountain model

(1.1.1) The three-dimensional digital elevation model of the mountain is obtained through remote sensing mapping, and the altitude data information of the real mountain is obtained, as shown in Figure 2(a).

(1.1.2) The construction of ANSYS geometric mountain model is composed of points, lines, surfaces and volumes, and points (coordinates) are the basis for constructing geometric models. Therefore, according to the above-mentioned altitude information, the creation of the geometric mountain is to generate a closed curve from the key points, and the closed curve generates a plane, and then the geometric mountain is surrounded by the closed surface. The geometric mountain constitutes the entire mountain model, as shown in Figure 2(b) and Figure 2(c).

(1.2) Finite element mesh division steps of mountain model

Anasy finite element meshing is a crucial step for numerical simulation analysis. Since the model of the mountain is not regular, free meshing can automatically generate triangular or tetrahedral meshes on the surface freely. Automatically generate tetrahedral grids on the body, and artificially control the intelligent size. In this example, the quadratic tetrahedron unit (unit 92) is used to ensure the calculation accuracy, as shown in Figure 2(d).

(1.3) Mountain model boundary condition setting and solution steps

Set the load boundary conditions for the above-mentioned mountain after grid division, use the basic heat transfer physical basis of mountain heat conduction and mountain-air heat convection, and set the heat transfer rate according to the heat conductivity K of the mountain and the heat convection rate Φ of the mountain-air Parameters, the temperature field distribution of the mountain is obtained through Ansys solution calculation, and finally the temperature field distribution of the mountain is mapped to the grayscale map and the temperature resolution is adjusted, and finally the mountain thermal radiation model is obtained. In this example, K=3.49Kg/m3, Φ=3W/(m^2.C), and the solution result is shown in Figure 3.

(2) The 8-bit to 16-bit mapping process of the mountain background thermal field includes the following sub-steps:

Since the established image of the mountain background infrared thermal field model is 8-bit, while the real mountain background thermal field image is 16-bit, it is necessary to perform an 8-bit to 16-bit comparison between the mountain background thermal field model obtained above and the real mountain background thermal field. Mapping processing. In the case that the thermal field of the model is roughly the same as the distribution of the real thermal field, it is ensured that the thermal field of the model is closer to the thermal field of the real mountain.

(2.1) Determination steps of the range of real mountain background gray value

The thermal field of the real mountain background is affected by many external factors, such as houses and roads. Because these external factors account for a small proportion in the overall background, they can be regarded as interference. Therefore, regional constraint processing is required, and threshold processing is performed through histogram statistics to eliminate the interference of external factors. The real mountain infrared image is shown in Figure 4, and the specific process is as follows:

(2.1.1) Let the variable r represent the pixel gray level in the image. In the discrete situation, use r _k to represent the discrete gray level, and use P(r _k ) to represent the probability density function. The following formula holds:

$\begin{array}{cc}{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}{\mathrm{<span\; class="notranslate">\; no</span>}}& <span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; )</span>\end{array}$

k=0,1,2...l-1

In the formula, n _k is the number of pixels that appear r _k gray in the image, n is the total number of pixels in the image, is the frequency in probability theory, and l is the total number of gray levels.

It is known that the influence of external factors accounts for P% of the area in the entire image, and the following formula is given:

P _r (r _k )≥P

Accumulate the gray histogram sequentially. If the accumulated value is greater than or equal to the proportion of the target object, stop accumulating and record the value of r _k as the guiding value of the background. In this example, P=0.5, _rk =31000.

(2.2) Mapping correction steps of mountain thermal field model

The gray value of the mountain thermal field background is corrected by linear mapping according to the value obtained above and the minimum gray value of the image. The specific formula is as follows:

$\mathrm{<span\; class="notranslate">\; o</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}}{{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; h</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; o</span>}}_{\mathrm{<span\; class="notranslate">\; h</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; o</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; o</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}$

Among them, I is the gray value of the thermal field model, I _l is the lowest brightness gray value of the thermal field model, I _h is the highest brightness gray value of the thermal field model, O _l is the lowest brightness gray value of the real mountain infrared image value, _Oh is the r _k obtained above, and O is the mountain background model after mapping correction. The corrected mountain background is shown in Figure 5. In this example, O _l =29852.

(3) Use the mapped mountain background thermal field model to carry out the background filtering step, including the following sub-steps:

Due to the angle of remote sensor observation and the influence of the sun's irradiation angle, the mountain can be divided into sunny and shady sides. The part of the mountain that is directly irradiated by the sun is called the sunny side, and the part that is not directly irradiated by the sun is called the shady side. The mountain background and the surrounding environment have thermal radiation, and there are differences in thermal radiation between direct and indirect exposures. Not only that, the thermal field of the mountain background during the day is different from that at night, and the present invention is aimed at the daytime detection and positioning of underground heat sources constrained by the thermal field model of the mountain background.

First find the pixel point at the boundary between the shaded and sunny sides, use the least squares fitting method to fit the dividing line between the shaded and sunny sides, then perform grayscale compensation on the shaded side, and finally perform background filtering.

(3.1) Extraction steps of the dividing line between the negative side and the positive side of the original infrared image

(3.1.1) Utilize the difference between the gray level difference between the shaded side and the sunny side to determine the pixel point at the boundary. According to the information of the image data, it can be determined that the boundary line is east-west, so it is only necessary to compare the gray value of the upper and lower pixel points. If The gray difference between the upper and lower adjacent pixels is greater than K, the formula is as follows:

G(x,y)>G(x,y-1)+K

G(x,y)>G(x,y-2)+K

G(x,y)>G(x,y-3)+K

G(x,y)>G(x,y-1)+K

If the above four inequalities are true, consider (x, y) as the pixel points near the dividing line, and traverse the whole image to get all the pixel points near the dividing line, as shown in Figure 6. In this example, K=40.

(3.1.2) Next, carry out the cubic polynomial least squares fitting linear processing on the pixel points near all the boundary lines obtained above, and the specific process is as follows:

in is the cubic polynomial fitted by least squares, and err is the error objective function. By minimizing err to achieve the optimal cubic polynomial fitting, the final dividing line is obtained, as shown in Figure 7. In this example, a ₀ =0, a ₁ =0.001, a ₂ =0.4635, a ₃ =42.5124.

(3.2) Steps to eliminate direct sunlight energy on the sunny side

After obtaining the shaded area and sunny area of the infrared image according to the dividing line, the following mapping strategy is used to eliminate the gray value of the sunny area of the infrared image:

${{\mathrm{<span\; class="notranslate">\; f</span>}}^{<span\; class="notranslate">\; \&prime;</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; \&lsqb;</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; +</span>\frac{{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}}{{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}}<span\; class="notranslate">\; *</span>{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; \&rsqb;</span>$

D＝F _n ' _oshadow (i,j)-F _noshadow (i,j)

In the formula, F' _noshadow (i, j) is the gray value of the sunny area of the infrared image that is not directly irradiated by the sun, F _noshadow (i, j) is the gray value of the area of the infrared image that is directly irradiated by the sun, and m _shadow and σ _shadow are the infrared The mean and variance of the gray value of the sunny area of the image, m _noshadow and σ _noshadow are the mean and variance of the gray value of the shaded area of the infrared image adjacent to the non-infrared image, A is the compensation intensity coefficient, and D is the energy gray value of the sun directly irradiating the sun . In this example, m _shadow =6478, m _noshadow =4478, σ _shadow =55.2752, σ _noshadow =61.9714, D=225, A=1.0.

After obtaining the gray value generated by direct sunlight from the above, iterate through all detected solar

In the surface area, the gray value compensation D of each point in the infrared image is subtracted from the sunny area of the infrared image to obtain the gray image after eliminating the influence of sunlight. The influence of solar radiation was not considered before, and the component of solar radiation was filtered out, and the energy irradiated by the sun on the sun's surface was filtered out. The thermal radiation component caused by solar radiation was absorbed by the sun's surface. Under the condition of thermal balance, the direct sunlight was filtered out. , and the results are shown in Figure 8.

(3.3) Background filtering steps of real infrared images of mountains

The mountain background thermal field model obtained by simulation is used as the background of the real mountain infrared image. Since the target heat radiation in the mountain conforms to the mathematical model of heat conduction, it is assumed that at a time t ₀ the target position (x ₀ , y ₀ , z ₀ ) The thermal radiation surface is BT(x,y,z,t)

BT(x,y,z,t)

=B(x,y,z,t)+T(x,y,z,t)*R _b (x,y,z,t)

+A(x,y,z,t)+δ(x,y,z,t)

At this moment, the target background body radiation field BT(x, y, z, t), the multi-media body radiation field B(x, y, z, t), the target body radiation field distorted by the multi-media body T(x, y,z,t)*R _b (x,y,z,t), the amount of radiation δ(x,y,z,t) lost by the contact surface between water body/ground body and air and the influence of solar radiation A(x,y , z, t) co-generated.

The influence of the target background body radiation field is mainly produced by B(x,y,z,t) and A(x,y,z,t), so the following formula is given:

T(x,y,z,t)=BT(x,y,z,t)-k*B(x,y,z,t)-A(x,y,z,t)

Where T(x,y,z,t) is the approximate thermal radiation field of the target, BT(x,y,z,t) is the radiation field of the target background body, and B(x,y,z,t) is the multi-media body The background radiation field, A(x, y, z, t) is the energy field generated by the sun, and k is the adjustable coefficient of the background radiation field. Finally, the filtered image of the real mountain infrared image is obtained, as shown in Figure 9. In this example, k=0.8.

(4) Detection and positioning steps of underground target space constrained mean value clustering

From the image after background filtering, select s image blocks to be recognized, and the template size is 3*3, which are respectively b ₁ , b ₂ , b ₃ ...b _s , including the target image area and the target image area below Image area, avoid the influence of houses and roads when selecting.

The road segment b ₁ , b ₂ , b ₃ ,..., b _s are divided into two types, the area with target image and the area without target image, by means of space-constrained mean clustering algorithm. The specific implementation process of the spatially constrained mean clustering algorithm is as follows:

Step1: For all sample points b _i , calculate the distance ratio

${\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; the\; s</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}\frac{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}{\underset{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; the\; s</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}$

Select the point b _i with the smallest V _i as the first centroid, and set q=1;

Step2: For p=1,2, assign b _i , i=1,2,...,s to the nearest class, and update the class center i=1,2, N _i is the number of samples of the i-th class;

Step3: Set q=q+1, if q>2, the algorithm stops;

Step4: Select the best initial center point of the next class as For the smallest point b _i , turn to Step2.

In this example, m ₁ =310, m ₂ =400

The result of clustering is obtained through the above formula, the category with large gray value is regarded as a category of suspected underground targets, and the category with small gray value is regarded as a category of non-suspected underground targets. Finally, the location of the underground target is obtained through the spatial constraint clustering algorithm, as shown in Figure 10.

Those skilled in the art can easily understand that the above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention, All should be included within the protection scope of the present invention.

Claims (10)

1. an underground heat source diurnal remote sensing detection and positioning method constrained by a mountain background thermal field model, it is characterized in that, the method comprises the steps: (1) The establishment of the mountain background thermal field model includes the following sub-steps: (1.1) Establishment of mountain model; (1.2) Finite element mesh division of mountain model; (1.3) Setting and solving of mountain model boundary conditions; (2) 8-bit to 16-bit mapping of the mountain background thermal field, including the following sub-steps: (2.1) Determination of the gray value range of the real mountain background; (2.2) Mapping correction of the mountain thermal field model; (3) Use the mapped mountain background thermal field model to perform background filtering, including the following sub-steps: (3.1) Extraction of the dividing line between the negative side and the positive side of the original infrared image; (3.2) Eliminate direct sunlight on the sunny side; (3.3) background filtering processing of real infrared images of mountains; (4) Detection and positioning of underground targets based on spatially constrained mean value clustering. 2. The method according to claim 1, wherein said step (4) specifically comprises the following sub-steps: From the image after background filtering, select s image blocks to be recognized, namely b ₁ , b ₂ , b ₃ ,..., b _s , including the area with the target image below and the area without the target image below. The constrained mean value clustering algorithm divides the road segment b ₁ , b ₂ , b ₃ ,...,b _s into two types: the region with the target image below and the region without the target image below, and the specific implementation process of the spatially constrained mean value clustering algorithm as follows: Step1: For all sample points b _i , calculate the distance ratio ${\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; the\; s</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}\frac{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}{\underset{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; the\; s</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}$ Select the point b _i with the smallest V _i as the first centroid, and set q=1; Step2: For p=1,2, assign b _i , i=1,2,...,s to the nearest class, and update the class center i=1,2, N _i is the number of samples of the i-th class; Step3: Set q=q+1, if q>2, the algorithm stops; Step4: Select the best initial center point of the next class as For the smallest point b _i , turn to Step2. 3. The method according to claim 1 or 2, characterized in that, said step (1.1) specifically comprises the following sub-steps: (1.1.1) Obtain the three-dimensional digital elevation model of the mountain body through remote sensing mapping, and obtain the altitude data information of the real mountain body; (1.1.2) The construction of ANSYS geometric mountain model is composed of points, lines, surfaces and volumes. Points are the basis for constructing geometric models. According to the above-mentioned altitude information, the creation of geometric mountains is to generate closed curves from key points , the plane is generated from the closed curve, and then the geometric mountain is surrounded by the closed surface, and the geometric mountain constitutes the entire mountain model. 4. The method according to claim 1 or 2, characterized in that, said step (2.1) specifically comprises the following sub-steps: (2.1.1) Let the variable r represent the pixel gray level in the image, use r _k to represent the discrete gray level, and use P(r _k ) to represent the probability density function, the following formula holds: ${\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; )</span>$ k=0,1,2...l-1 In the formula, n _k is the number of pixels that appear r _k gray in the image, n is the total number of pixels in the image, is the frequency in probability theory, and l is the total number of gray levels; It is known that the influence of external factors accounts for P% of the area in the entire image, and the following formula is given: P _r (r _k )≥P Accumulate the gray histogram sequentially. If the accumulated value is greater than or equal to the proportion of the target object, stop accumulating and record the value of r _k as the guiding value of the background. 5. the method as claimed in claim 1 or 2, is characterized in that, described step (2.2) is specifically: The gray value of the background of the mountain thermal field model is corrected by linear mapping according to the value obtained above and the minimum gray value of the image; the specific formula is as follows: $\mathrm{<span\; class="notranslate">\; o</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}}{{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; h</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; o</span>}}_{\mathrm{<span\; class="notranslate">\; h</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; o</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; o</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}$ Among them, I is the gray value of the thermal field model, I _l is the lowest brightness gray value of the thermal field model, I _h is the highest brightness gray value of the thermal field model, O _l is the lowest brightness gray value of the real mountain infrared image value, _Oh is the r _k obtained above, and O is the mountain background model after mapping correction. 6. The method according to claim 1 or 2, characterized in that said step (3.1) specifically comprises the following sub-steps: (3.1.1) Determine the pixel point at the boundary by using the difference between the gray value of the shaded side and the sunny side, and determine that the boundary line is east-west according to the information of the image data, and compare the gray value of the upper and lower pixels. The difference is greater than K, the formula is as follows: G(x,y)>G(x,y-1)+K G(x,y)>G(x,y-2)+K G(x,y)>G(x,y-3)+K G(x,y)>G(x,y-1)+K If the above four inequalities are true, then (x, y) is regarded as the pixel points near the dividing line, and all the pixels near the dividing line are obtained by traversing the whole image; (3.1.2) Carry out cubic polynomial least squares fitting linear processing to the pixel points near all boundary lines obtained above, and the specific process is as follows: in is the cubic polynomial fitted by the least squares, and err is the error objective function. By minimizing err to achieve the optimal cubic polynomial fitting, the final dividing line is obtained. 7. The method according to claim 1 or 2, characterized in that said step (3.2) specifically comprises the following sub-steps: After obtaining the shaded area and sunny area of the infrared image according to the dividing line, the following mapping strategy is used to eliminate the gray value of the sunny area of the infrared image: ${{\mathrm{<span\; class="notranslate">\; f</span>}}^{<span\; class="notranslate">\; \&prime;</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; \&lsqb;</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; +</span>\frac{{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}}{{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}}<span\; class="notranslate">\; *</span>{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; \&rsqb;</span>$ D＝F′ _noshadow (i,j)-F _noshadow (i,j) In the formula, F' _noshadow (i, j) is the gray value of the sunny area of the infrared image that is not directly irradiated by the sun, F _noshadow (i, j) is the gray value of the area of the infrared image that is directly irradiated by the sun, and m _shadow and σ _shadow are the infrared The mean and variance of the gray value of the sunny area of the image, m _noshadow and σ _noshadow are the mean and variance of the gray value of the shaded area of the infrared image adjacent to the non-infrared image, A is the compensation intensity coefficient, and D is the energy gray value of the sun directly irradiating the sun . 8. The method according to claim 1 or 2, characterized in that said step (3.3) specifically comprises the following sub-steps: Use the simulated mountain background thermal field model as the background of the real mountain infrared image, assuming that the thermal radiation surface at the target position (x ₀ , y ₀ , z ₀ ) at a time t ₀ is BT(x,y,z,t ) BT(x,y,z,t) =B(x,y,z,t)+T(x,y,z,t)*R _b (x,y,z,t) +A(x,y,z,t)+δ(x,y,z,t) At this moment, the target background body radiation field BT(x, y, z, t), the multi-media body radiation field B(x, y, z, t), and the target body radiation field distorted by the multi-media body T(x, y, z, t)*R _b (x, y, z, t), the amount of radiation δ(x, y, z, t) lost by the contact surface between water body/ground body and air, and the influence of solar radiation A(x, y , z, t) co-generated; The influence of the target background body radiation field is produced by B(x,y,z,t) and A(x,y,z,t), which has the following formula: T(x,y,z,t)=BT(x,y,z,t)-k*B(x,y,z,t)-A(x,y,z,t) Where T(x,y,z,t) is the approximate thermal radiation field of the target, BT(x,y,z,t) is the radiation field of the target background body, and B(x,y,z,t) is the multi-media body The background radiation field, A(x, y, z, t) is the energy field affected by solar radiation, and k is the adjustable coefficient of the background radiation field; finally, the image of the real mountain infrared image after background filtering is obtained. 9. The method according to claim 1 or 2, characterized in that, the step (1.2) is specifically: Carry out free mesh division, freely and automatically generate triangular or tetrahedral meshes on the surface, automatically generate tetrahedral meshes on the body, and artificially control the intelligent size. 10. The method according to claim 1 or 2, characterized in that, the step (1.3) is specifically: Set the load boundary conditions for the above-mentioned mountain after grid division, use the basic heat transfer physical basis of mountain heat conduction and mountain-air heat convection, and set the heat transfer rate according to the heat conductivity K of the mountain and the heat convection rate Φ of the mountain-air Parameters, the temperature field distribution of the mountain is obtained through Ansys solution calculation, and finally the temperature field distribution of the mountain is mapped to the grayscale image and the temperature resolution is adjusted to finally obtain the mountain thermal field model.