Forest grassland surface fire detection method based on multispectral characteristic response

Abstract

The invention discloses a forest grassland surface fire detection method based on multispectral characteristic response. The method comprises the following steps: 1, acquiring an infrared response spectrum, an ultraviolet response spectrum and an image response spectrum in real time by a sensor; 2, a feature extraction module extracts time domain feature vector data and frequency domain feature vector data in the infrared response spectrum and the ultraviolet response spectrum, and extracts image feature vector data in the image response spectrum; 3, a preprocessing module carries out standardization and normalization processing on the feature vector sets A, B and C to obtain normalized vector sets A ', B' and C '; 4, the first support vector machine model SVM1 classifies the normalized vector sets A'and B 'into interference type data X and suspected surface fire type data Y; and 5, classifying the normalized vector set C'and the suspected surface fire data Y into interference data X and surface fire data Z by the second support vector machine model SVM2. The method has the advantages that efficient and reliable monitoring and recognition of the surface fire are achieved, and the classification and recognition accuracy of the surface fire is effectively improved.

Description

Technical Field

[0001] The invention relates to the technical field of spectral analysis and image processing, and in particular to a forest grassland surface fire detection method based on multi-spectral characteristic response. Background Art

[0002] Medium-wave infrared sensors and ultraviolet sensors are used to obtain infrared and ultraviolet response spectra respectively. These sensors can be installed on towers or monitoring equipment at a certain height from the ground to achieve real-time monitoring of large-scale forest and grassland areas. According to the National Forestry and Grassland Administration's Forest Fire Prevention 2016-2025 Plan, the video surveillance coverage rate of forest areas across the country will reach more than 95% in 2025, and a large part of the monitoring equipment will use infrared and ultraviolet sensors. Traditional fire monitoring systems usually rely on a single spectrum or manual inspections, lack intelligence and comprehensive analysis capabilities, and cannot detect and warn fires in a timely and accurate manner.

[0003] Most existing fire monitoring systems can only perform simple infrared monitoring and lack the ability to comprehensively analyze multi-spectral information. They cannot effectively identify and distinguish fires from other interference factors, such as direct sunlight, reflection, and electromagnetic interference. Although some systems have automatic alarm functions, the false alarm rate is high and accurate fire location and identification cannot be achieved. In addition, traditional monitoring systems can only record passively and cannot actively analyze fire characteristics or predict the development of fires.

[0004] Disadvantages of existing technologies: lack of comprehensive analysis capabilities of multi-spectral information, unable to effectively identify and distinguish fires from other interference factors; unable to achieve accurate fire location and identification, unable to actively analyze fire characteristics or predict fire development. Summary of the invention

[0005] The invention provides a forest grassland surface fire detection method based on multi-spectral characteristic response, which improves the accuracy of surface fire classification and identification.

[0006] To achieve the above object, the present invention provides a forest grassland surface fire detection method based on multi-spectral feature response, the key of which is to include the following steps:

[0007] Step 1: The medium-wave infrared sensor acquires the infrared response spectrum a in real time and passes it to the feature extraction module; the ultraviolet sensor acquires the ultraviolet response spectrum b in real time and passes it to the feature extraction module; the COMS sensor acquires the image response spectrum c in real time and passes it to the feature extraction module;

[0008] Step 2: The feature extraction module extracts the time domain feature vector data and the frequency domain feature vector data from the infrared response spectrum a to form an infrared spectrum feature vector set A, and passes it to the preprocessing module;

[0009] The feature extraction module extracts the time domain feature vector data and the frequency domain feature vector data from the ultraviolet response spectrum b to form an ultraviolet spectrum feature vector set B, and transmits it to the preprocessing module;

[0010] The feature extraction module extracts image feature vector data from the image response spectrum c to form a time series image feature vector set C, and passes it to the preprocessing module;

[0011] Step 3: The preprocessing module standardizes and normalizes the infrared spectrum feature vector set A, the ultraviolet spectrum feature vector set B and the time series image feature vector set C respectively to obtain the infrared spectrum normalized vector set A', the ultraviolet spectrum normalized vector set B' and the time series image normalized vector set C', and then passes the infrared spectrum normalized vector set A' and the ultraviolet spectrum normalized vector set B' to the first support vector machine model SVM1, and passes the time series image normalized vector set C' to the second support vector machine model SVM2;

[0012] Step 4: The first support vector machine model SVM1 classifies the infrared spectrum normalized vector set A' and the ultraviolet spectrum normalized vector set B' into interference data X and suspected surface fire data Y through a polynomial kernel function, and sends the suspected surface fire data Y to the second support vector machine model SVM2;

[0013] Step 5: The second support vector machine model SVM2 classifies the normalized vector set C' of the time series image and the suspected surface fire data Y into interference data X and surface fire data Z through the radial basis kernel function. When the surface fire data Z is not classified, it indicates that no fire has occurred and the process goes to step 1; otherwise, it indicates that a fire has occurred and a fire alarm is issued.

[0014] Through the above design, the multi-spectral sensor combined with image processing technology can comprehensively analyze and identify fire characteristics. Among them, the medium-wave infrared response spectrum and ultraviolet response spectrum can provide characteristic information such as flames and smoke, while image processing can capture the dynamic changes of the fire scene. Through the comprehensive analysis of these three spectral information, fires can be accurately distinguished from other interference factors, and accurate fire monitoring and early warning can be achieved.

[0015] The preprocessing module is used to standardize and normalize the feature vector set. The purpose of normalization is to standardize the data of different features so that they have the same dimension and distribution range, thereby eliminating the dimensional differences between different features and facilitating the subsequent training and classification of the support vector machine model SVM.

[0016] Preferably, in step 2, the time domain feature vector data of the infrared response spectrum a includes duration T, zero crossing rate ZCR, energy E, envelope, variance σ 2 , kurtosis K, autocorrelation ACF, cross-correlation CCF and filter complexity;

[0017] Among them, the duration T is the sum of the time that the signal strength exceeds a certain threshold, the zero crossing rate ZCR is the number of times the signal crosses the zero point per unit time, the energy E is the square sum integral of the signal, and the variance σ 2 is the variance of signal strength, kurtosis K is the kurtosis of the signal, autocorrelation ACF is the correlation of the signal itself, cross-correlation CCF is the correlation between two signals, and filter complexity C is the complexity of the signal.

[0018] The calculation expression of the energy E is as follows:

[0019] E=∫|A(t)| 2 dt

[0020] The variance σ 2 The calculation expression is as follows:

[0021]

[0022] The calculation expression of the kurtosis K is as follows:

[0023]

[0024] The calculation expression of the autocorrelation ACF is as follows:

[0025]

[0026] The calculation expression of the cross-correlation CCF is as follows:

[0027]

[0028] Where A(t) represents the amplitude of the infrared response signal over continuous time t, A i represents the amplitude of the infrared response signal at the i-th discrete time point, A i+τ represents the amplitude of the infrared response signal per unit time τ after the i-th discrete time point, μ represents the average value of the infrared response signal, σ represents the standard deviation of the infrared response signal, and τ represents the time lag; A(i) represents the amplitude of the infrared response signal at the i-th discrete time point, B(i+τ) represents the amplitude of the ultraviolet response signal per unit time τ after the i-th discrete time point, and N represents N discrete time points;

[0029] The time-domain feature vector data of the ultraviolet response spectrum b include duration, zero-crossing rate, energy, envelope, variance, kurtosis, autocorrelation, cross-correlation and filter complexity.

[0030] The calculation expression of the time domain characteristic vector data of the ultraviolet response spectrum b is consistent with the calculation formula of the time domain characteristic vector data of the infrared response spectrum a, and only the input is changed to the ultraviolet response signal.

[0031] Preferably, in step 2, the frequency domain feature vector data of the infrared response spectrum a includes a zero frequency component F0, a non-zero frequency component F nz , amplitude D, phase φ, power spectral density PSD;

[0032] Among them, the amplitude D is the modulus of the frequency domain signal, and the phase φ is the angle of the frequency domain signal.

[0033] The feature extraction module converts the time domain signal A(t) into the frequency domain signal A(f) through fast Fourier transform, and the expression is as follows:

[0034]

[0035] The calculation expression of the amplitude D is as follows:

[0036]

[0037] The calculation expression of the phase φ is as follows:

[0038] φ=arg(A(f))

[0039] The calculation expression of the power spectral density PSD is as follows:

[0040]

[0041] Wherein, A(t) represents the time domain signal of infrared response spectrum a, and A(f) represents the frequency domain signal of infrared response spectrum a;

[0042] The frequency domain feature vector data of the ultraviolet response spectrum b includes zero-frequency component, non-zero-frequency component, amplitude, phase, and power spectrum density.

[0043] The calculation expression of the frequency domain characteristic vector data of the ultraviolet response spectrum b is consistent with the calculation expression of the frequency domain characteristic vector data of the infrared response spectrum a, and only the input is changed to the ultraviolet response signal.

[0044] Preferably, in step 2, the image feature vector data includes a color histogram, a texture binary pattern map, a highlight contraction and swing time series, a highlight target area and a highlight boundary curvature.

[0045] Among them, the color histogram is the histogram distribution of each color channel of the image; the texture binary pattern map is a local binary pattern used to describe the image texture; the highlight shrinkage and swing time series is the time series change of the highlight area, the highlight target area is the area of ​​the highlight area in the image, and the highlight boundary curvature is the curvature of the boundary of the highlight area.

[0046] The present invention significantly reduces the probability of misclassification by utilizing the dynamic characteristics of the highlight area of ​​the image time series, and is more reliable and effective than using only the dual-spectrum algorithm.

[0047] Preferably, the spectrum range of the infrared response spectrum is 3.0um-5.0um, the spectrum range of the ultraviolet response spectrum is 180nm-300nm, and the spectrum range of the image response spectrum is 350nm-1000nm.

[0048] Preferably, in step 4, the expression of the polynomial kernel function is as follows:

[0049] K(A′ j ,B′ j )=(γ <A′ j ,B′ j >+r) d

[0050] Among them, K(A′ j ,B′ j ) represents the polynomial kernel function, A′ j represents the jth infrared spectrum normalized vector sample point, B′ j represents the jth UV spectrum normalized vector sample point, <A′ j ,B′ j > represents the sample point A′ j and B′ j The inner product of , that is, the dot product; γ represents the kernel function parameter that controls the influence of the inner product, r represents the offset or constant kernel function parameter; d represents the order of the polynomial, which determines the degree of nonlinearity of the polynomial kernel.

[0051] The role of the polynomial kernel function is to map the input data to a higher-dimensional space so that the data can be linearly separable in this high-dimensional space, thereby achieving more complex classification tasks.

[0052] Preferably, in step 5, the expression of the radial basis kernel function is as follows:

[0053] G(C′ j ,Y j ) = exp(-γ∥C′ j -Y j ∥ 2 )

[0054] Among them, G(C′ j ,Y j ) represents the radial basis kernel function, C′ j represents the normalized vector sample point of the jth time series image, Y j represents the jth suspected surface fire data sample point, ∥C′ j -Y j ∥ 2 Represents C′ j and Y j The square of the Euclidean distance between them, and exp() represents the natural exponential function.

[0055] The characteristic of the radial basis kernel function is that when the distance between two sample points is farther, the value of the kernel function will be closer to zero; when the distance between two sample points is closer, the value of the kernel function will be closer to one. The advantage of the radial basis kernel function is that it can map the original input data to an infinite-dimensional feature space, so that it can handle highly nonlinear data distribution, is suitable for many complex classification and regression problems, and can improve classification and recognition accuracy.

[0056] Preferably, in step 4 and step 5, the input expression of the first support vector machine model SVM1 or the second support vector machine model SVM2 is as follows:

[0057] I={(x1,y1),(x2,y2),…,(x n ,y n )}

[0058] The output expression of the first support vector machine model SVM1 is as follows:

[0059] O={z1,z2,…,z n}

[0060] The decision function of the first support vector machine model SVM1 is expressed as follows:

[0061]

[0062] Among them, I represents the input, and the input in SVM1 is (A ′ ,B′), the input of SVM2 is (C ′ ,Y); n represents the number of samples, O represents the output, O∈{X,Y} in SVM1, and O∈{X,Z} in SVM2, where X represents the interference class, Y represents the suspected surface fire class, and Z represents the surface fire class; α and e are model parameters, α represents the Lagrange multiplier, which represents the contribution of the sample point to the decision surface, and b is the bias term; sign is a sign function used to determine the positive and negative signs of the classification results. If f(x, y)>0, then sign(f(x, y))=1, and the input sample is classified as the positive class.

[0063] If f(x, y)<0, then sign(f(x, y))=-1, and the input sample is classified as the negative class.

[0064] The beneficial effects of the present invention are as follows: through the combination of multi-spectral response characteristics and a two-level support vector machine model, efficient and reliable monitoring and identification of surface fires are achieved, and the accuracy of surface fire classification and identification is effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 It is a schematic diagram of the process of the present invention;

[0066] Figure 2 It is a structural schematic diagram of the present invention. DETAILED DESCRIPTION

[0067] The present invention is further described in detail below in conjunction with the accompanying drawings and specific examples. The following examples or drawings are used to illustrate the present invention, but are not used to limit the scope of the present invention.

[0068] like Figure 1 , Figure 2 As shown: A forest grassland surface fire detection method based on multi-spectral feature response, comprising the following steps:

[0069] Step 1: The medium-wave infrared sensor acquires the infrared response spectrum a in real time and passes it to the feature extraction module; the ultraviolet sensor acquires the ultraviolet response spectrum b in real time and passes it to the feature extraction module; the COMS sensor acquires the image response spectrum c in real time and passes it to the feature extraction module;

[0070] Step 2: The feature extraction module extracts the time domain feature vector data and the frequency domain feature vector data from the infrared response spectrum a to form an infrared spectrum feature vector set A, and passes it to the preprocessing module;

[0071] The feature extraction module extracts the time domain feature vector data and the frequency domain feature vector data from the ultraviolet response spectrum b to form an ultraviolet spectrum feature vector set B, and transmits it to the preprocessing module;

[0072] The feature extraction module extracts image feature vector data from the image response spectrum c to form a time series image feature vector set C, and passes it to the preprocessing module;

[0073] Step 3: The preprocessing module standardizes and normalizes the infrared spectrum feature vector set A, the ultraviolet spectrum feature vector set B and the time series image feature vector set C respectively to obtain the infrared spectrum normalized vector set A', the ultraviolet spectrum normalized vector set B' and the time series image normalized vector set C', and then passes the infrared spectrum normalized vector set A' and the ultraviolet spectrum normalized vector set B' to the first support vector machine model SVM1, and passes the time series image normalized vector set C' to the second support vector machine model SVM2;

[0074] Step 4: The first support vector machine model SVM1 classifies the infrared spectrum normalized vector set A' and the ultraviolet spectrum normalized vector set B' into interference data X and suspected surface fire data Y through a polynomial kernel function, and sends the suspected surface fire data Y to the second support vector machine model SVM2;

[0075] Step 5: The second support vector machine model SVM2 classifies the normalized vector set C' of the time series image and the suspected surface fire data Y into interference data X and surface fire data Z through the radial basis kernel function. When the surface fire data Z is not classified, it indicates that no fire has occurred and the process goes to step 1; otherwise, it indicates that a fire has occurred and a fire alarm is issued.

[0076] In step 2, the time domain feature vector data of the infrared response spectrum a includes duration T, zero crossing rate ZCR, energy E, envelope, variance σ 2 , kurtosis K, autocorrelation ACF, cross-correlation CCF and filter complexity;

[0077] The calculation expression of the energy E is as follows:

[0078] E=∫|A(t)| 2 dt

[0079] The variance σ 2 The calculation expression is as follows:

[0080]

[0081] The calculation expression of the kurtosis K is as follows:

[0082]

[0083] The calculation expression of the autocorrelation ACF is as follows:

[0084]

[0085] The calculation expression of the cross-correlation CCF is as follows:

[0086]

[0087] Where A(t) represents the amplitude of the infrared response signal over continuous time t, A i represents the amplitude of the infrared response signal at the i-th discrete time point, A i+τ represents the amplitude of the infrared response signal per unit time τ after the i-th discrete time point, μ represents the average value of the infrared response signal, σ represents the standard deviation of the infrared response signal, and τ represents the time lag; A(i) represents the amplitude of the infrared response signal at the i-th discrete time point, B(i+τ) represents the amplitude of the ultraviolet response signal per unit time τ after the i-th discrete time point, and N represents N discrete time points;

[0088] The time-domain feature vector data of the ultraviolet response spectrum b include duration, zero-crossing rate, energy, envelope, variance, kurtosis, autocorrelation, cross-correlation and filter complexity.

[0089] In step 2, the frequency domain feature vector data of the infrared response spectrum a includes a zero frequency component F0, a non-zero frequency component F nz , amplitude D, phase φ, power spectral density PSD;

[0090] The feature extraction module converts the time domain signal A(t) into the frequency domain signal A(f) through fast Fourier transform, and the expression is as follows:

[0091]

[0092] The calculation expression of the amplitude D is as follows:

[0093] D=|A(f)|

[0094] The calculation expression of the phase φ is as follows:

[0095] φ=arg(A(f))

[0096] The calculation expression of the power spectral density PSD is as follows:

[0097]

[0098] Wherein, A(t) represents the time domain signal of infrared response spectrum a, and A(f) represents the frequency domain signal of infrared response spectrum a;

[0099] The frequency domain feature vector data of the ultraviolet response spectrum b includes zero-frequency component, non-zero-frequency component, amplitude, phase, and power spectrum density.

[0100] In step 2, the image feature vector data includes a color histogram, a texture binary pattern map, a highlight shrinkage swing time series, a highlight target area, and a highlight boundary curvature.

[0101] The spectrum range of the infrared response spectrum is 3.0um-5.0um, the spectrum range of the ultraviolet response spectrum is 180nm-300nm, and the spectrum range of the image response spectrum is 350nm-1000nm.

[0102] In step 4, the expression of the polynomial kernel function is as follows:

[0103] K(A′ j ,B′ j )=(γ <A′ j ,B′ j >+r) d

[0104] Among them, K(A′ j ,B′ j ) represents the polynomial kernel function, A′ j represents the jth infrared spectrum normalized vector sample point, B′ j represents the jth UV spectrum normalized vector sample point, <A′ j ,B′ j > represents the sample point A′ j and B′ j The inner product of , that is, the dot product; γ represents the kernel function parameter that controls the influence of the inner product, r represents the offset or constant kernel function parameter; d represents the order of the polynomial, which determines the degree of nonlinearity of the polynomial kernel.

[0105] In step 5, the expression of the radial basis kernel function is as follows:

[0106] G(C′ j ,Y j ) = exp(-γ∥C′ j -Y j ∥ 2 )

[0107] Among them, G(C′ j ,Y j ) represents the radial basis kernel function, C′ j represents the normalized vector sample point of the jth time series image, Y j represents the jth suspected surface fire data sample point, ∥C′ j -Y j ∥ 2 Represents C′ j and Y j The square of the Euclidean distance between them, and exp() represents the natural exponential function.

[0108] In step 4 or step 5, the input expression of the first support vector machine model SVM1 or the second support vector machine model SVM2 is as follows:

[0109] I={(x1,y1),(x2,y2),…,(x n ,y n )}

[0110] The output expression of the first support vector machine model SVM1 is as follows:

[0111] O={z1,z2,…,z n}

[0112] The decision function of the first support vector machine model SVM1 is expressed as follows:

[0113]

[0114] Among them, I represents the input, and the input in SVM1 is (A ′ ,B′), the input of SVM2 is (C ′ ,Y); n represents the number of samples, O represents the output, O∈{X,Y} in SVM1, and O∈{X,Z} in SVM2, where X represents the interference class, Y represents the suspected surface fire class, and Z represents the surface fire class; α and e are model parameters, α represents the Lagrange multiplier, which represents the contribution of the sample point to the decision surface, and b is the bias term; sign is a sign function used to determine the positive and negative signs of the classification results. If f(x, y)>0, then sign(f(x, y))=1, and the input sample is classified as the positive class.

[0115] If f(x, y)<0, then sign(f(x, y))=-1, and the input sample is classified as the negative class.

[0116] The first support vector machine model SVM1 and the second support vector machine model SVM2 are trained by the following steps:

[0117] Step 1: The infrared sensor obtains the infrared response spectrum a and passes it to the feature extraction module; the ultraviolet sensor obtains the ultraviolet response spectrum b and passes it to the feature extraction module; the COMS sensor obtains the image response spectrum c and passes it to the feature extraction module;

[0118] Step 2: The feature extraction module extracts the time domain feature vector data and the frequency domain feature vector data in the infrared response spectrum a to form an infrared spectrum feature vector set A, and transmits it to the preprocessing module;

[0119] The feature extraction module extracts the time domain feature vector data and the frequency domain feature vector data from the ultraviolet response spectrum b to form an ultraviolet spectrum feature vector set B, and transmits it to the preprocessing module;

[0120] The feature extraction module extracts image feature vector data from the image response spectrum c to form a time series image feature vector set C, and passes it to the preprocessing module;

[0121] Step 3: The preprocessing module standardizes and normalizes the infrared spectrum feature vector set A, the ultraviolet spectrum feature vector set B and the time series image feature vector set C respectively to obtain the infrared spectrum normalized vector set A', the ultraviolet spectrum normalized vector set B' and the time series image normalized vector set C', and then passes the infrared spectrum normalized vector set A' and the ultraviolet spectrum normalized vector set B' to the first support vector machine model SVM1, and passes the time series image normalized vector set C' to the second support vector machine model SVM2;

[0122] Step 4: The first support vector machine model SVM1 determines parameters and penalty parameters through a polynomial kernel function, and establishes interference factors for the infrared spectrum normalized vector set A' and the ultraviolet spectrum normalized vector set B', generating dual-spectrum sample sets A" and B" without interference factors and with interference factors, and uses the dual-spectrum sample sets A" and B" to train the first support vector machine model SVM1, and determines the optimal model parameters through cross-validation, thereby optimizing the first support vector machine model, and then classifies the dual-spectrum signals into interference class X and suspected surface fire class Y, and passes the suspected surface fire class Y to the second support vector machine model SVM2;

[0123] Step 5: The second support vector machine model SVM2 determines the parameters and penalty parameters through the radial basis kernel function, establishes the feature vector set C' of fire images, non-fire images, and interference images, trains the second support vector machine model SVM2, selects the optimal parameters through cross-validation, and then optimizes the second support vector machine model SVM2, and classifies the suspected surface fire signals into interference X and surface fire Z.

[0124] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A forest-grassland surface fire detection method based on multispectral feature response, characterized in that: The following steps are involved: Step 1: The medium-wave infrared sensor acquires the infrared response spectrum a in real time and passes it to the feature extraction module; The ultraviolet sensor obtains the ultraviolet response spectrum b in real time and transmits it to the feature extraction module; The COMS sensor acquires the image response spectrum c in real time and passes it to the feature extraction module; Step 2: The feature extraction module extracts the time domain feature vector data and the frequency domain feature vector data from the infrared response spectrum a to form an infrared spectrum feature vector set A, and passes it to the preprocessing module; The feature extraction module extracts the time domain feature vector data and the frequency domain feature vector data from the ultraviolet response spectrum b to form an ultraviolet spectrum feature vector set B, and transmits it to the preprocessing module; The feature extraction module extracts image feature vector data from the image response spectrum c to form a time series image feature vector set C, and passes it to the preprocessing module; Step 3: The preprocessing module standardizes and normalizes the infrared spectrum feature vector set A, the ultraviolet spectrum feature vector set B and the time series image feature vector set C respectively to obtain the infrared spectrum normalized vector set A', the ultraviolet spectrum normalized vector set B' and the time series image normalized vector set C', and then passes the infrared spectrum normalized vector set A' and the ultraviolet spectrum normalized vector set B' to the first support vector machine model SVM1, and passes the time series image normalized vector set C' to the second support vector machine model SVM2; Step 4: The first support vector machine model SVM1 classifies the infrared spectrum normalized vector set A' and the ultraviolet spectrum normalized vector set B' into interference data X and suspected surface fire data Y through a polynomial kernel function, and sends the suspected surface fire data Y to the second support vector machine model SVM2; Step 5: The second support vector machine model SVM2 classifies the normalized vector set C' of the time series image and the suspected surface fire data Y into interference data X and surface fire data Z through the radial basis kernel function. When the surface fire data Z is not classified, it indicates that no fire has occurred and the process goes to step 1; otherwise, it indicates that a fire has occurred and a fire alarm is issued.

2. The forest-grassland surface fire detection method based on multispectral characteristic response according to claim 1 is characterized in that: In step 2, the time domain feature vector data of the infrared response spectrum a includes duration T, zero crossing rate ZCR, energy E, envelope, variance σ 2 , kurtosis K, autocorrelation ACF, cross-correlation CCF and filter complexity; The calculation expression of the energy E is as follows: E=∫|A(t)| 2 dt The variance σ 2 The calculation expression is as follows: The calculation expression of the kurtosis K is as follows: The calculation expression of the autocorrelation ACF is as follows: The calculation expression of the cross-correlation CCF is as follows: Where A(t) represents the amplitude of the infrared response signal over continuous time t, A i represents the amplitude of the infrared response signal at the i-th discrete time point, A i+τ represents the amplitude of the infrared response signal per unit time τ after the i-th discrete time point, μ represents the average value of the infrared response signal, σ represents the standard deviation of the infrared response signal, and τ represents the time lag; A(i) represents the amplitude of the infrared response signal at the i-th discrete time point, B(i+τ) represents the amplitude of the ultraviolet response signal per unit time τ after the i-th discrete time point, and N represents N discrete time points; The time-domain feature vector data of the ultraviolet response spectrum b include duration, zero-crossing rate, energy, envelope, variance, kurtosis, autocorrelation, cross-correlation and filter complexity.

3. The forest-grassland surface fire detection method based on multispectral characteristic response according to claim 1 is characterized in that: In step 2, the frequency domain feature vector data of the infrared response spectrum a includes a zero frequency component F0, a non-zero frequency component F nz , amplitude D, phase φ, power spectral density PSD; The feature extraction module converts the time domain signal A(t) into the frequency domain signal A(f) through fast Fourier transform, and the expression is as follows: The calculation expression of the amplitude D is as follows: D=|A(f)| The calculation expression of the phase φ is as follows: φ=arg(A(f)) The calculation expression of the power spectral density PSD is as follows: Wherein, A(t) represents the time domain signal of infrared response spectrum a, and A(f) represents the frequency domain signal of infrared response spectrum a; The frequency domain feature vector data of the ultraviolet response spectrum b includes zero-frequency component, non-zero-frequency component, amplitude, phase, and power spectrum density.

4. The forest-grassland surface fire detection method based on multispectral characteristic response according to claim 1 is characterized in that: In step 2, the image feature vector data includes a color histogram, a texture binary pattern map, a highlight contraction and swing time series, a highlight target area, and a highlight boundary curvature.

5. The forest-grassland surface fire detection method based on multispectral characteristic response according to claim 1 is characterized in that: The spectrum range of the infrared response spectrum is 3.0um-5.0um, the spectrum range of the ultraviolet response spectrum is 180nm-300nm, and the spectrum range of the image response spectrum is 350nm-1000nm.

6. The forest-grassland surface fire detection method based on multispectral characteristic response according to claim 1 is characterized in that: In step 4, the expression of the polynomial kernel function is as follows: K(A′ j ,B′ j )=(γ <A′ j ,B′ j >+r) d Among them, K(A′ j ,B′ j ) represents the polynomial kernel function, A′ j represents the jth infrared spectrum normalized vector sample point, B′ j represents the jth UV spectrum normalized vector sample point, <A′ j ,B′ j > represents the sample point A′ j and B′ j The inner product of , that is, the dot product; γ represents the kernel function parameter that controls the influence of the inner product, r represents the offset or constant term kernel function parameter; d represents the order of the polynomial, which determines the degree of nonlinearity of the polynomial kernel.

7. The forest-grassland surface fire detection method based on multispectral characteristic response according to claim 1 is characterized in that: In step 5, the expression of the radial basis kernel function is as follows: G(C′ j ,Y j )=exp(-γ∥C′ j -Y j ∥ 2 ) Among them, G(C′ j ,Y j ) represents the radial basis kernel function, C′ j represents the normalized vector sample point of the jth time series image, Y j represents the jth suspected surface fire data sample point, ∥C′ j -Y j ∥ 2 Represents C′ j and Y j The square of the Euclidean distance between them, and exp() represents the natural exponential function.

8. The forest-grassland surface fire detection method based on multispectral characteristic response according to claim 1, characterized in that: In step 4 or step 5, the input expression of the first support vector machine model SVM1 or the second support vector machine model SVM2 is as follows: I={(x1,y1),(x2,y2),…,(x n ,y n )} The output expression of the first support vector machine model SVM1 is as follows: O={z1,z2,…,z n } The decision function of the first support vector machine model SVM1 is expressed as follows: Where I represents input, the input in SVM1 is (A′, B′), and the input in SVM2 is (C′, Y); n represents the number of samples, O represents output, O∈{X, Y} in SVM1, and O∈{X, Z} in SVM2, where X represents interference class, Y represents suspected surface fire class, and Z represents surface fire class; α and e are model parameters, α represents the Lagrange multiplier, which represents the contribution of the sample point to the decision surface, and b is the bias term; sign is a sign function used to determine the positive or negative sign of the classification result. If f(x, y)>0, then sign(f(x, y))=1, and the input sample is classified as the positive class. If f(x, y)<0, then sign(f(x, y))=-1, and the input sample is classified as the negative class.

Images