A method of target recognition in aviation airborne navigation images based on harmony search using spatial entropy

Abstract

The present invention discloses an aviation airborne navigation image target recognition method based on spatial entropy harmony search. The method utilizes the position distribution information of the image grayscale space to calculate the image grayscale space distribution entropy value, and maps it into a uniform distribution function. The function is then processed by 2D-DCT, transform system weighting and inverse 2D-DCT linear transformation to achieve simultaneous enhancement of the global and local contrast of the original image. The method is combined with the maximum entropy harmony search algorithm. Although the method is similar to the genetic algorithm, it adopts a multi-point mating mode in the process of searching for new solutions in the field, so that the new solutions are more diverse and the possibility of the algorithm jumping out of the local optimal solution is improved.

Description

Technical Field

[0001] The invention relates to the field of aviation airborne navigation image processing, and in particular to an aviation airborne navigation image target recognition method based on spatial entropy and harmony search. Background Art

[0002] In the field of aviation airborne navigation image processing, many target detection methods focus on image processing at high resolution. They usually use deep learning algorithms such as R-CNN (Region with CNN feature), SSD (Single Shot MultiBox Detector), ANN, and construct multi-scale feature fusion algorithms. Most of them are complex in principle and difficult to implement. Therefore, this paper introduces an automatic classification technology after airborne navigation image processing. By setting image conversion, feature extraction and result comparison, it realizes the automatic conversion recognition and classification functions of airborne navigation image sets. The entire algorithm system is fully automated under the control of deep learning network technology, and the classification effect of deep learning network is improved through human intervention and optimization algorithm upgrade.

[0003] The system includes an image acquisition preprocessing module, an image recognition module, and an image comparison and classification module. Its working principle is as follows: the image acquisition preprocessing module is mainly responsible for the preliminary processing of the image. First, the airborne navigation image set is input into the system, and after grayscale processing, it is convenient for subsequent algorithm recognition and analysis, and then simple classification such as viewpoint change and zoom analysis is performed; then the image recognition module is input, which is responsible for identifying and extracting the features of similar objects in the grayscale image, and using probability theory and statistical methods, approximation theory and other analytical operations for joint processing to obtain the feature recognition of the target to be measured in the image; finally, image comparison and classification are performed, based on the deep neural network learning model, through artificial variable assignment weight update, simulation prediction analysis and other algorithm processing, the target in the image is automatically classified and identified according to the features.

[0004] The above-mentioned automatic classification technology for airborne navigation image processing is based on the control of deep learning network technology, which is fully automated. It does not need to rely on the operator's experience and knowledge and does not need to adjust parameters repeatedly. The processing process is relatively simple and efficient. However, considering that deep neural networks are still used in this technology, the complexity of its algorithm design is still not low, and it does not pay attention to the fact that small targets in airborne navigation images have little information, low contrast, multiple backgrounds, and are easily affected by weather and environment. This makes the algorithm not very adaptable and robust under such complex environmental conditions.

[0005] In summary, a method for target recognition in aviation airborne navigation images based on spatial entropy and harmony search is needed to address the deficiencies in the prior art. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention provides an aviation airborne navigation image target recognition method based on spatial entropy and harmony search, aiming to solve the above problems.

[0007] To achieve the above object, the present invention provides the following technical solution: an aviation airborne navigation image target recognition method based on spatial entropy harmony search, comprising the following steps:

[0008] Step S1: grayscale the original image and convert it into a grayscale image;

[0009] Step S2: Calculate the grayscale space entropy and distribution function of the image according to the two-dimensional space histogram, and perform contrast enhancement on the image;

[0010] Step S3: transformation and local contrast enhancement, weighting the transformation domain coefficients to enhance the contrast of the local image;

[0011] Step S4: Find the best threshold segmentation point, and find the best threshold segmentation point by searching the optimal algorithm according to the characteristics of the recorded navigation image and the specific target features;

[0012] Step S5: Output of target recognition and segmentation results. According to the determined optimal grayscale vector, the best threshold segmentation point is obtained, and threshold segmentation is performed accordingly to output the final target recognition and segmentation result image.

[0013] Optionally, the grayscale conversion of the original image in step S1 is performed in the following manner:

[0014] Determine the pixel value range corresponding to each gray level, convert the original image into an image with K gray levels, and calculate the two-dimensional spatial histogram on the spatial grid of the image to make the original image correspond to the gray image.

[0015] Optionally, the two-dimensional spatial histogram on the spatial grid of the calculated image is:

[0016] Assuming the original image size is H*W, x(i,j) is the value of each pixel in image X, x(i,j) [0, z+], and As a set of K gray levels of image X, , calculate the gray level on the spatial grid of X The two-dimensional space histogram of is:

[0017] ,

[0018] Where m, n z+,h k (m,n) [0,z+] is the region of the image located in the spatial grid The number of times grayscale Xx appears,

[0019] The total number of grids on the two-dimensional histogram is M*N, using the number K of different gray levels and the ratio , ,

[0020] in It means taking the nearest integer to form grayscale The two-dimensional space histogram of , while preserving the aspect ratio of the original image on the spatial grid.

[0021] Optionally, the entropy measure for calculating the gray level in step S2 is:

[0022] ,

[0023] Evaluating discrete functions Measuring Grayscale Relative to the rest of the grayscale , and normalize it:

[0024] ,

[0025] ,

[0026] Where K is the number of gray levels.

[0027] Optionally, the distribution function F is calculated in step S2 k for:

[0028] ,

[0029] ,

[0030] In the formula , , , .

[0031] Optionally, the transformation and local contrast enhancement in step S3 are performed in the following manner:

[0032] Set the original image and the transformed image of the same size to X and D respectively.

[0033] ,

[0034] ,

[0035] ,

[0036] , ,

[0037] wherein the values of k and l respectively represent frequency components,

[0038] The transform domain coefficient d(k, l) is corrected to obtain a higher frequency transform coefficient ,

[0039] ,

[0040] ,

[0041] In the formula, is the weighted frequency allocation coefficient after being increased,

[0042] Perform inverse transformation:

[0043] ,

[0044] ,

[0045] An output image with enhanced global or local contrast is obtained.

[0046] Optionally, in step S4, the best threshold segmentation point is found in the following manner:

[0047] Step S41: According to the characteristics of the airborne navigation image and the characteristics of specific targets, a set of initial feature thresholds is preset as the gray vector X = (x1, x2,..., xn) of the initial threshold segmentation point, where 1 < n < 128, and the threshold segmentation point is normally distributed;

[0048] Step S42: Under each initial threshold segmentation point, calculate the sum of the entropy values H(n) of the two categories of the target and the background, generate an initial harmony memory library HM, select one or more harmonies from the harmony library, and perform pitch fine-tuning to generate new harmonies;

[0049] Step S43: Calculate the entropy values of the two categories A and B at each gray level corresponding to the new harmony vector, take the maximum entropy value as the solution under this vector, and if this value is greater than the value in HM, update the harmony memory library HM;

[0050] Step S44: When the number of times of updating the gray vector reaches the preset number of iterations, or when the sum of the entropy values of the two categories A and B obtained by solving under the new gray vector reaches the maximum, determine the best threshold segmentation point.

[0051] Optionally, in step S43, the entropy values of the two categories A and B at each gray level corresponding to the new harmony vector are:

[0052] It is assumed that the image grayscale is divided into two categories, A and B, from Th, where A and B are composed of all pixels in the image with grayscale levels in [0, Th] and [Th+1, n] respectively.

[0053] , ,

[0054] Where: is the sum of gray level probabilities from 1 to Th, is the sum of gray level probabilities from Th+1 to n, P i is the probability distribution of gray levels in the image.

[0055] Optionally, the sum of the entropy values ​​of categories A and B in step S44 is:

[0056] ,

[0057] The harmony memory banks are:

[0058] .

[0059] Optionally, the threshold image segmentation in step S5 is:

[0060] Binarize the image, select a threshold to convert it into black and white, and judge the grayscale value in the image and the threshold value. If the grayscale value is greater than the threshold, the grayscale value of the pixel is set to 255, otherwise it is set to 0.

[0061] Beneficial effects of the present invention:

[0062] 1. In the present invention, the airborne navigation image is taken from a high altitude, and the contrast of the image is relatively low, making it difficult to distinguish the target from the background, and the target recognition is relatively difficult. The position distribution information of the image grayscale space is used to calculate the image grayscale space distribution entropy value, and it is mapped into a uniform distribution function, and then through 2D-DCT, transformation system weighting and inverse 2D-DCT linear transformation processing, the global and local contrast of the original image is enhanced simultaneously;

[0063] 2. In the present invention, the harmony search algorithm combined with maximum entropy is used. Although it is similar to the genetic algorithm, in the process of searching for new solutions in the field, a multi-point mating mode is adopted to make the new solutions more diverse and increase the possibility of the algorithm jumping out of the local optimal solution;

[0064] 3. In the present invention, the combination and application of the entropy concept makes full use of the grayscale information of the image to achieve simultaneous enhancement of the global and local contrast of the traditional image, and takes the threshold value when the entropy value of the target and background takes the maximum entropy value as the optimal threshold segmentation point to achieve accurate recognition and segmentation of specific targets in the image;

[0065] 4. In the present invention, the maximum entropy harmony search algorithm is combined, and the amount of calculation is significantly reduced. When searching, the grayscale set based on the initial threshold is appropriately selected, which can effectively accelerate the speed of algorithm search convergence and quickly obtain the optimal threshold segmentation point. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 The present invention is a flow chart of a method.

[0067] Figure 2 The present invention is a flow chart of calculating distribution function.

[0068] Figure 3 This is a flow chart of an optimal threshold point of the present invention. DETAILED DESCRIPTION

[0069] In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0070] like Figure 1 , 2 As shown in , a method for target recognition of aviation airborne navigation images based on harmony search of spatial entropy includes the following contents:

[0071] Using the position distribution information of the image grayscale space, the entropy value of the image grayscale space distribution is calculated to obtain a uniform distribution function, which is then mapped into a uniform distribution function to achieve global image contrast enhancement. After 2D-DCT, transform domain coefficient weighting, and 2D-DCT inverse transformation, the local image contrast is further enhanced. After obtaining simultaneous enhancement of the global and local images, the improved harmony search optimal algorithm is used to find the optimal background and image segmentation threshold point to achieve the final target detection and recognition. The algorithm implementation steps are as follows:

[0072] The original image is grayed out and converted into a grayscale image. Assuming the original image size is H*W, x(i,j) is the value of each pixel in image X, x(i,j) [0, z+], and As a set of K gray levels of image X, , calculate the gray level on the spatial grid of X The two-dimensional space histogram of is:

[0073] ,

[0074] Where m, n z+,hk(m,n) [0,z+] is the region of the image located in the spatial grid The number of times grayscale xx appears. The total number of grids on the two-dimensional histogram is M*N, using the number K of different gray levels and the ratio , ,

[0075] in It means taking the nearest integer to form grayscale The two-dimensional space histogram of When , the aspect ratio of the original image is preserved on the spatial grid, thereby preserving the spatial characteristics of the pixels;

[0076] Calculate the grayscale space entropy and distribution function of the image: According to the two-dimensional spatial histogram, calculate the entropy measure of the grayscale:

[0077] ,

[0078] Evaluating discrete functions Measuring Grayscale Relative to the rest of the grayscale , the relative importance of K, and normalized:

[0079] , ,

[0080] Calculate the cumulative distribution function and map it to image Y according to the distribution function:

[0081] , ,

[0082] in, , , .

[0083] After the above global contrast enhancement, in order to enhance the local contrast at the same time, based on the above processing, two-dimensional discrete cosine transform (2D-DCT) is used again, and after transform domain coefficient weighting and 2D-DCT inverse transformation, an output image with global and local enhanced contrast is obtained;

[0084] The steps of 2D-DCT transformation, weighting and inverse transformation of transform domain coefficients are as follows: Assume that the original image and the transformed image of the same size generated by 2D discrete cosine transform are X and D respectively, where:

[0085] ,

[0086] ,

[0087] ,

[0088] , ,

[0089] Among them, the values of k and l respectively represent frequency components. To increase the local contrast, higher frequency transformation coefficients need to be adjusted, that is, the transformation domain coefficient d(k, l) is corrected to obtain higher frequency transformation coefficients , where is the weighted frequency allocation coefficient after adjustment.

[0090] ,

[0091] ,

[0092] Among them, the higher the value of ɑ, the higher the local enhancement degree. The selection of parameters is calculated according to the entropy value estimation formula mentioned above; γ ∈[0, 1], which determines the degree of local contrast enhancement. When the value is 0, only global enhancement occurs. In this algorithm, its value can be selected as 1. After weighting the transformation domain coefficients, the following formula is used for the inverse 2D-DCT transformation to obtain an output image with both global and local enhancement:

[0093] .

[0094] ,

[0095] After the above-mentioned 2D-DCT transformation, transformation domain function weighting, and 2D-DCT inverse transformation processing, an image with both global and local contrast enhancement is obtained.

[0096] After obtaining the output contrast-enhanced image, an improved harmony search optimal algorithm will be used to find the optimal threshold segmentation point, so as to accurately segment and identify specific targets in the image;

[0097] First, according to the characteristics of airborne navigation images and the characteristics of specific targets, an initial feature threshold set is preset as the gray vector X = (x1, x2,..., xn) of the initial threshold segmentation point, where 1 < n < 128, and the threshold segmentation point follows a normal distribution; under each initial threshold segmentation point, there is a sum of entropy values H(n) of the two categories of target and background calculated at this threshold point, generating an initial harmony memory bank HM. Among them, the principle and formula for calculating the entropy value are as follows:

[0098] For an image with gray levels of n, p1, p2, ..., pn are the probability distributions of the gray levels in the image. The image gray levels are divided into two categories, A and B, from Th, where A and B are composed of all pixels in the image with gray levels in [0, Th] and [Th+1, n] respectively.

[0099] make represents the sum of gray level probabilities from 1 to Th, represents the sum of gray level probabilities from Th+1 to n, then the probability distribution of each gray level pixel in class A in the entire class A pixel set is: , , ..., ,

[0100] The probability distribution of each grayscale pixel in class B in the entire class B pixel set is: , , ..., The information entropy H(A) and H(B) of categories A and B are calculated as follows:

[0101] , ,

[0102] Then the sum of the information entropy of categories A and B is:

[0103] .

[0104] According to the initial threshold segmentation point set and the corresponding entropy value, HMS harmony vectors are randomly generated and put into the harmony memory HM, and the entropy function f(X) under the corresponding vector is recorded. The generated initial harmony memory HM is in the following form:

[0105] .

[0106] According to the dichotomy and empirical considerations, one or more harmonies are selected from the harmony library, and the pitch is fine-tuned to generate a new harmony. Assume that each variable in the new harmony variable X' (k=1,2,...,m), the probability of HMCR (memory consideration rate, the probability of selecting harmony from the harmony library) is taken from the initial harmony library HM If any value is selected from HM, there is a probability of 1-HCMR to randomly select any value in the entire initial threshold set. When the new variable is taken from HM, in order to increase the diversity of solutions and expand the search range, it is necessary to fine-tune the variable. The probability of fine-tuning is PAR. At this time, the new variable :

[0107] ,

[0108] Where rand is a random number uniformly distributed between [0,1], BW is the bandwidth of pitch fine-tuning (generally 1 in this grayscale image threshold), and the updated harmony vector can be obtained by the above update criteria;

[0109] Calculate the entropy values ​​of the two categories A and B under each grayscale sequence corresponding to the new harmony vector, and take the largest entropy value as the solution of f(X') under this vector. If this value is greater than f(X') in HM, i ), it is replaced, indicating that the threshold segmentation effect under the new vector is better, so the harmony memory HM is updated;

[0110] If the number of grayscale vector updates reaches the preset number of iterations, or the sum of the entropy values ​​of the two categories A and B under the new grayscale vector reaches the maximum (or the entropy increase is no longer obvious), it means that the grayscale value at this time reaches the optimal threshold segmentation point; if the number of iterations is not reached or the entropy value continues to increase, return to step 6 to re-update the grayscale vector, and continue to calculate the entropy value and determine whether the termination condition is met;

[0111] According to the determined optimal grayscale vector, the best threshold segmentation point is obtained and threshold segmentation is performed accordingly, and the final target recognition and segmentation result image is output.

[0112] Threshold image segmentation is a common image processing algorithm. The core is the binary processing of the image. A threshold is selected to convert it into a black and white binary image: if the gray value in the image is greater than a certain threshold, the gray value of the pixel is set to 255, otherwise it is set to 0. It is often used for image segmentation and edge extraction. Assuming that the optimal threshold point is s obtained by the above method, and the image after the original image is grayed is x, the image threshold segmentation can be achieved by using the MATLAB professional drawing function (im2bw(s,x)).

[0113] Alternative 1:

[0114] For low-contrast images, the grayscale distribution is too concentrated in an area with a very small dynamic range, and the grayscale changes between the target and the background are not large, which makes it difficult to segment the target and the background. For this reason, linear or nonlinear transformations are usually used to expand the dynamic range of the grayscale, or histogram equalization is used to make the grayscale distribution fill the entire grayscale level space to improve the image contrast.

[0115] This algorithm is based on the grayscale information of the image. First, after the image is grayed, the grayscale mean is taken for a rectangular area centered on a certain point of the image to replace the original grayscale value, and then the grayscale variance image of the original image is calculated, thereby achieving image contrast enhancement; then grayscale partitioning is performed, and the image is classified according to the grayscale attributes of the pixels, and a group of grayscale values ​​are selected to divide it into different grayscale areas, and the variance in each interval is minimized, thereby dividing the grayscale of the entire image into several areas; finally, the areas with the same grayscale are connected together, and the median filtering method is used to remove noise and then connect the areas with the same grayscale, thereby achieving the segmentation of the target and the background.

[0116] The method is simple in principle, easy to implement and convenient to calculate. It makes full use of the grayscale information of the image and eliminates the influence of low contrast and background noise of the image. It completes the segmentation of the image target only by calculating the mean and variance of the grayscale, performing grayscale variance interval division and median filtering.

[0117] Alternative 2:

[0118] The threshold segmentation method based on maximum entropy is one of the most commonly used algorithms in image segmentation processing. Entropy in information theory is a concept used to measure the amount of information: the more uniform the distribution, the greater the entropy. In image processing, it is to make full use of the grayscale information of the image, gray the image, and then divide all the pixels in the image into two categories, target and background, according to a certain threshold. When both tend to be evenly distributed, the values ​​of the target entropy and background entropy calculated separately also tend to be maximum, and the effect of separating the target and background is the best at this time.

[0119] When this solution is adopted, the original image is directly grayed out, and then the entropy value is calculated according to the preset threshold. All gray values ​​are traversed to calculate the target and background entropy. When the sum of the two reaches the maximum, this threshold is the optimal threshold segmentation point. By performing image segmentation according to the optimal threshold segmentation point, the target can be identified and segmented.

[0120] This method is simple to implement, easy to operate, and has stable performance. It only requires graying the image, calculating the entropy value at each gray level, and finding the gray value that maximizes the entropy value. This eliminates the algorithm step of designing and searching for the optimal threshold, thereby greatly reducing the complexity of the image segmentation algorithm and has been widely used in practice.

[0121] Airborne navigation images are taken from a high altitude, and the contrast of the images is relatively low, making it difficult to distinguish between the target and the background, and target recognition is relatively difficult. The position distribution information of the image grayscale space is used to calculate the image grayscale space distribution entropy value, and it is mapped into a uniform distribution function. Then, it is processed by 2D-DCT, transform system weighting and inverse 2D-DCT linear transformation to achieve simultaneous global and local contrast enhancement of the original image.

[0122] The application of the harmony search algorithm combined with maximum entropy is similar to the genetic algorithm, but in the process of searching for new solutions in the field, a multi-point mating mode is adopted to make the new solutions more diverse and increase the possibility of the algorithm jumping out of the local optimal solution;

[0123] The combination and application of the concept of entropy value makes full use of the grayscale information of the image to achieve simultaneous enhancement of the global and local contrast of the traditional image, and uses the threshold value when the entropy value of the target and background takes the maximum entropy value as the optimal threshold segmentation point to achieve accurate recognition and segmentation of specific targets in the image;

[0124] Combined with the maximum entropy harmony search algorithm, the amount of calculation is significantly reduced. When searching, the grayscale set based on the initial threshold is properly selected, which can effectively speed up the algorithm search convergence and quickly obtain the optimal threshold segmentation point.

[0125] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modification, equivalent substitution or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for target recognition in aviation airborne navigation images based on spatial entropy and harmony search, characterized in that: It includes the following steps: Step S1: Grayscale the original image and convert it into a grayscale image; Step S2: Calculate the grayscale spatial entropy and distribution function of the image according to the two-dimensional spatial histogram, and enhance the contrast of the graph; Step S3: Transform and enhance the local contrast, perform weighted processing on the transform domain coefficients, and enhance the contrast of the local image; Step S4: Find the optimal threshold segmentation point. According to the characteristics of the recorded navigation image and specific target features, search for the optimal threshold segmentation point through the optimal algorithm; Step S5: Output the target recognition and segmentation results. According to the determined optimal grayscale vector, obtain the optimal threshold segmentation point, and perform threshold segmentation based on this, and output the image of the final target recognition and segmentation results.

2. The method for aviation airborne navigation image target recognition based on spatial entropy and harmony search according to claim 1 is characterized in that: In the above-mentioned step S1, the original image is grayscaled in the following way: Determine the pixel value range corresponding to each grayscale level, convert the original image into an image with K grayscale levels, and calculate the two-dimensional spatial histogram on the spatial grid of the image, so that the original image corresponds to the grayscale image.

3. The method for aviation airborne navigation image target recognition based on spatial entropy and harmony search according to claim 2 is characterized in that: The calculation of the two-dimensional spatial histogram on the spatial grid of the image is as follows: Assuming the original image size is H*W, x(i,j) is the value of each pixel in image X, x(i,j) [0, z+], and As a set of K gray levels of image X, , calculate the gray level on the spatial grid of X The two-dimensional space histogram of is: ， Where m, n z+,h k (m,n) [0,z+] is the region of the image located in the spatial grid The number of times grayscale Xx appears, The total number of grids on the two-dimensional histogram is M*N, using the number K of different gray levels and the ratio , , in It means taking the nearest integer to form grayscale The two-dimensional space histogram of , while preserving the aspect ratio of the original image on the spatial grid.

4. The method for aviation airborne navigation image target recognition based on spatial entropy and harmony search according to claim 1 is characterized in that: In the above-mentioned step S2, the entropy measure of the grayscale level is: ， Evaluating discrete functions Measuring Grayscale Relative to the rest of the grayscale , and normalize it: ， ， In the formula, K is the number of grayscale levels.

5. The method for aviation airborne navigation image target recognition based on spatial entropy and harmony search according to claim 1 is characterized in that: In step S2, the distribution function F is calculated. k for: ， ， In the formula , , , .

6. The method for aviation airborne navigation image target recognition based on spatial entropy and harmony search according to claim 1 is characterized in that: In the above-mentioned step S3, the transformation and local contrast enhancement are carried out in the following way: Set the original image of the picture and the transformed image of the same size generated by the transformation as X and D respectively, ， ， ， ， ， where: the values of k and l respectively represent the frequency components, Modify the transform domain coefficient d(k,l) to obtain a higher frequency transform coefficient , ， ， In the formula, To increase the weighted frequency allocation coefficient, Perform inverse transformation: ， ， Obtain the output image with enhanced global or local contrast.

7. The method for aviation airborne navigation image target recognition based on spatial entropy and harmony search according to claim 1 is characterized in that: In the above-mentioned step S4, the optimal threshold segmentation point is found in the following way: Step S41: According to the characteristics of the airborne navigation image and specific target features, preset the initial feature threshold set as the grayscale vector X=(x1,x2,...,xn) of the initial threshold segmentation point, where 1 < n < 128, and the threshold segmentation point is a normal distribution; Step S42: Under each initial threshold segmentation point, calculate the sum of the entropy values H(n) of the two categories of target and background, generate the initial harmony memory library HM, select one or more harmonies from the harmony library, and perform pitch fine-tuning to generate new harmonies; Step S43: Calculate the entropy values of the two categories A and B at each grayscale level corresponding to the new harmony vector, take the maximum entropy value as the solution under this vector, and if this value is greater than the value in HM, update the harmony memory library HM; Step S44: When the number of times of updating the grayscale vector reaches the preset number of iterations, or when the sum of the entropy values of the two categories A and B under the new grayscale vector reaches the maximum, determine the optimal threshold segmentation point.

8. The method for aviation airborne navigation image target recognition based on spatial entropy and harmony search according to claim 7 is characterized in that: In the above-mentioned step S43, the entropy values of the two categories A and B at each grayscale level corresponding to the new harmony vector are: Set the image grayscale level to be divided into two categories A and B at Th, where A and B are respectively composed of all pixels with grayscale levels in [0,Th] and [Th + 1,n] in the image, ， ， Where: is the sum of gray level probabilities from 1 to Th, is the sum of gray level probabilities from Th+1 to n, P i is the probability distribution of gray levels in the image.

9. The method for aviation airborne navigation image target recognition based on spatial entropy and harmony search according to claim 8 is characterized in that: In the above-mentioned step S44, the sum of the entropy values of the two categories A and B is: ， The harmony memory library is: 。 10. The method for aviation airborne navigation image target recognition based on spatial entropy and harmony search according to claim 1, characterized in that: In the above-mentioned step S5, the threshold image segmentation is: Perform binary processing on the image, select a threshold to convert it into a black and white image, and judge the size of the grayscale value in the image and the threshold. If the grayscale value is greater than the threshold, set the grayscale value of the pixel to 255, otherwise set it to 0.

Images