CN113408370B - Forest change remote sensing detection method based on adaptive parameter genetic algorithm - Google Patents
Forest change remote sensing detection method based on adaptive parameter genetic algorithm Download PDFInfo
- Publication number
- CN113408370B CN113408370B CN202110605694.1A CN202110605694A CN113408370B CN 113408370 B CN113408370 B CN 113408370B CN 202110605694 A CN202110605694 A CN 202110605694A CN 113408370 B CN113408370 B CN 113408370B
- Authority
- CN
- China
- Prior art keywords
- pixel
- class
- forest
- individual
- remote sensing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 54
- 230000002068 genetic effect Effects 0.000 title claims abstract description 46
- 238000001514 detection method Methods 0.000 title claims abstract description 38
- 230000008859 change Effects 0.000 title claims abstract description 30
- 230000003044 adaptive effect Effects 0.000 title claims description 12
- 238000000034 method Methods 0.000 claims abstract description 28
- 238000002310 reflectometry Methods 0.000 claims abstract description 11
- 238000001228 spectrum Methods 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 238000001914 filtration Methods 0.000 claims abstract description 5
- 239000011159 matrix material Substances 0.000 claims description 27
- 239000008186 active pharmaceutical agent Substances 0.000 claims description 24
- 238000012937 correction Methods 0.000 claims description 8
- 230000005855 radiation Effects 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 6
- 230000003595 spectral effect Effects 0.000 claims description 6
- 229910018503 SF6 Inorganic materials 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- SFZCNBIFKDRMGX-UHFFFAOYSA-N sulfur hexafluoride Chemical compound FS(F)(F)(F)(F)F SFZCNBIFKDRMGX-UHFFFAOYSA-N 0.000 claims description 3
- 229960000909 sulfur hexafluoride Drugs 0.000 claims description 3
- 238000002329 infrared spectrum Methods 0.000 claims description 2
- 238000005457 optimization Methods 0.000 claims 1
- 238000004088 simulation Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 230000036039 immunity Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 239000002689 soil Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000004451 qualitative analysis Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 230000001932 seasonal effect Effects 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 238000012876 topography Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
Abstract
The invention belongs to the technical field of remote sensing, and particularly discloses a forest change remote sensing detection method based on a self-adaptive parameter genetic algorithm, which comprises the steps of preprocessing forest remote sensing image data; calculating normalized vegetation indexes of the two atmospheric corrected forest remote sensing images by utilizing the reflectivity of the spectrum wave band, and enabling the normalized vegetation indexes to be a vegetation index feature map; carrying out median filtering on the vegetation index feature map, and constructing a vegetation index difference map on the filtered feature map by using a difference method; and (3) carrying out significance detection on the vegetation index difference graph: pre-classifying the saliency difference map by using a fuzzy C-means clustering algorithm, optimizing class labels of pixels in the uncertain class pixel set by using a genetic algorithm based on self-adaptive parameters, and generating a final change map. The method is used for solving the problems of low forest change detection precision and huge search space caused by a genetic algorithm using a single spectrum characteristic.
Description
Technical Field
The invention belongs to the technical field of remote sensing, relates to a forest variation remote sensing detection method, and in particular relates to a forest variation remote sensing detection method based on a self-adaptive parameter genetic algorithm.
Background
The change detection is one of the most important research subjects in remote sensing image processing, namely, two or more remote sensing images are acquired at different time in the same geographical area, qualitative and quantitative analysis is performed, and the earth surface change is measured, so that the method is widely applied to the fields of environmental monitoring, natural disaster evaluation, forest resource monitoring, agricultural investigation and the like.
Forest is the largest ecosystem on human land covering 31% of the surface of the earth. The forest has important influence on ecological environment, biodiversity and climate change, plays an important role in purifying air, regulating climate, conserving water sources, reducing wind and sand hazard and the like, and the quantity and quality of the forest become important material bases of national economy. The change of forest resource conditions has important influence on the development progress of countries and even on the global ecological environment, the biodiversity and the climate change. Therefore, the method can timely and accurately acquire the forest coverage change information, and has important significance for researching environmental change and forest management planning.
The traditional forest change detection method is mainly artificial visual interpretation, special interpretation personnel are required to compare and interpret the change areas of the images in different periods, and it is difficult to rapidly and accurately judge a large-scale forest area. Compared with the traditional method, the remote sensing technology has the advantages of large detection range, short data acquisition period, less limitation by ground conditions and the like, and is widely applied to the forest change detection field in recent years. The remote sensing images of different plants in different seasons are greatly different, and the seasonal difference of vegetation can be overcome by utilizing spectral characteristic parameters such as normalized vegetation index NDVI, specific vegetation index RVI and the like to detect forest variation. However, as the remote sensing images are subjected to noise interference of different degrees in the process of acquisition and transmission, the classification of the remote sensing images by using the spectral features only can lead to lower classification accuracy.
Disclosure of Invention
Aiming at the problems existing in the prior art, the invention aims to provide a forest variation remote sensing detection method based on a self-adaptive parameter genetic algorithm, which is used for solving the problems of low forest variation detection precision and huge search space caused by using a genetic algorithm with single spectral characteristics.
In order to achieve the above purpose, the present invention is realized by the following technical scheme.
A forest change remote sensing detection method based on a self-adaptive parameter genetic algorithm comprises the following steps:
step 1, obtaining two forest remote sensing images I at different moments 1 And I 2 Preprocessing each forest remote sensing image to obtain two preprocessed forest remote sensing images I 1 "and I 2 ”；
Step 2, respectively calculating two preprocessed forest remote sensing images I by utilizing reflectivity of spectrum wave bands 1 ”、I 2 "normalized vegetation index matrix NDVI 1 、NDVI 2 And make it be correspondent vegetation index characteristic map SF 1 、SF 2 ；
Step 3, for vegetation index feature map SF 1 And SF (sulfur hexafluoride) 2 Respectively performing median filtering to obtainDenoised feature map SF 1 '、SF 2 'A'; using a difference method to denoise the denoised characteristic map SF 1 '、SF 2 ' construct a vegetation index difference map DI;
step 4, performing significance detection on the vegetation index difference map DI to obtain a significance difference map DS;
step 5, pre-classifying the saliency difference map DS by using a fuzzy C-means clustering algorithm, and dividing pixels in the DS into a changed class, an unchanged class and an uncertain class, wherein a changed class pixel set, an unchanged class pixel set and an uncertain class pixel set are correspondingly obtained;
step 6, optimizing the class labels of the pixels in the uncertain class pixel set by using a genetic algorithm based on the self-adaptive parameters to obtain optimized class labels corresponding to the uncertain classes; and generating a final change chart according to the optimized category label.
Compared with the prior art, the invention has the beneficial effects that:
(1) The filtered vegetation index difference map DI is subjected to significance detection, the approximate position of a change region is obtained, and the fuzzy C-means clustering algorithm is used for pre-classification in the significance region, so that the search space of evolutionary computation is greatly reduced;
(2) The method selects the genetic algorithm to optimize the pixel label of the forest remote sensing image, and performs multi-point search on the whole, so that the coverage area is large, and the forest remote sensing image is not easy to fall into local optimum;
(3) The genetic algorithm used by the invention not only considers the spectrum information of the image, but also introduces a space neighborhood factor, thereby solving the problem of poor noise immunity of the traditional genetic algorithm on forest variation detection tasks; meanwhile, the adaptive variation genetic operator enables the variation probability of the pixels to be adaptively changed, and the algorithm convergence speed is increased.
Drawings
The invention will now be described in further detail with reference to the drawings and to specific examples.
FIG. 1 is a flow chart of an implementation of the present invention;
FIG. 2 is a forest remote sensing image of a simulation experiment and the processing results of different algorithms; wherein (a) and (b) are forest remote sensing images at different moments respectively; (c) The change map obtained by the original genetic algorithm, and (d) the change map obtained by the method.
Detailed Description
Embodiments of the present invention will be described in detail below with reference to examples, but it will be understood by those skilled in the art that the following examples are only for illustrating the present invention and should not be construed as limiting the scope of the present invention.
Referring to fig. 1, the remote sensing detection method for forest variation based on the adaptive parameter genetic algorithm provided by the invention comprises the following steps:
step 1, obtaining two forest remote sensing images I at different moments 1 And I 2 Preprocessing each forest remote sensing image to obtain two preprocessed forest remote sensing images I 1 "and I 2 "; the image size is MxNxB; m, N respectively represent I 1 And I 2 B represents the number of spectrum bands of the image, wherein M is more than or equal to 20, N is more than or equal to 20, and B is more than or equal to 4;
step 1.1, reading two sentry 2A satellite forest remote sensing images at different moments, wherein m=350, n=200 and b=4;
step 1.2, because the remote sensing image data is easy to generate systematic and random radiation distortion or distortion in the process of acquisition and transmission, the distortion of the remote sensing image is caused, and interpretation of the remote sensing image are affected. In order to eliminate or correct image distortion caused by radiation errors, the invention uses a forest remote sensing image I 1 And I 2 Inputting into remote sensing image processing software ENVI 5.2, using radiometric calibration kit pair I of ENVI 1 And I 2 Performing radiation correction to obtain a forest remote sensing image I after the radiation correction 1 '、I 2 ‘；
Because the atmospheric and illumination factors influence the reflectivity of the ground object, in order to obtain the true reflectivity of the ground object, the invention uses the FLASH Atmospheric Correction tool kit in ENVI 5.2 to correct the radiation of the forest remote sensing image I 1 '、I 2 ' atmospheric correction is performed to obtain an atmospheric correctedForest remote sensing image I 1 ”、I 2 And the "", is the preprocessed image.
Step 2, respectively calculating two preprocessed forest remote sensing images I by utilizing reflectivity of spectrum wave bands 1 ”、I 2 "normalized vegetation index matrix NDVI 1 、NDVI 2 And make it be correspondent vegetation index characteristic map SF 1 、SF 2 ；
The vegetation index can combine the visible light and the near infrared wave band of the satellite according to the spectral characteristics of the plants, so that the effective measurement of the vegetation condition of the earth surface is realized, the vegetation and the soil are effectively distinguished, and the method is widely applied to the field of forest variation detection. The normalized vegetation index NDVI is an optimal indicator factor of vegetation coverage, is sensitive to the change of soil background, eliminates the influence of topography and shadow to a great extent, and has a good vegetation extraction effect. The invention calculates the normalized vegetation index matrix NDVI of the forest remote sensing image after the atmospheric correction by utilizing the reflectivity of the remote sensing image spectral band, and the calculation formula is as follows:
wherein ρ is NIR And ρ RED Respectively representing the reflectivity of an infrared spectrum band and the reflectivity of a red spectrum band of the forest remote sensing image, so that each preprocessed forest remote sensing image I 1 ”、I 2 "obtained NDVI matrix is vegetation index feature map SF of corresponding forest image 1 、SF 2 The image size is mxn, specifically 350×200.
Step 3, for vegetation index feature map SF 1 And SF (sulfur hexafluoride) 2 Respectively performing median filtering to correspondingly obtain denoised characteristic images SF 1 '、SF 2 'A'; using a difference method to denoise the denoised characteristic map SF 1 '、SF 2 ' construct a vegetation index difference map DI;
3.1, using a median filtering algorithm to obtain two vegetation index feature graphs SF 1 、SF 2 Denoising to obtain a denoised vegetation index feature map SF 1 '、SF 2 ’；
3.2, using a difference method to denoise the denoised characteristic map SF 1 '、SF 2 ' construction of vegetation index difference map DI, i.e. two vegetation index feature maps SF 1 、SF 2 And (3) making differences between pixel values at corresponding positions, taking the differences as the pixel values at the positions corresponding to DI (x, y), and traversing all pixel points to obtain a difference map DI.
DI(x,y)＝SF 1 '(x,y)-SF 2 '(x,y)。
Step 4, performing significance detection on the vegetation index difference map DI to obtain a significance difference map DS;
4.1, in order to reduce the search space of classification calculation, the invention performs a bottom-up image saliency detection on the vegetation index difference map DI based on global contrast. The method calculates the global contrast of each pixel on the whole image, takes the sum of Euclidean distances of the pixel and all other pixels in the image in color as the saliency characteristic of the pixel, and the calculation formula is as follows:
wherein Sam (x, y) is the saliency feature of pixel (x, y), and (x, y) + (i, j); sam (x, y) is then normalized to [0,255]. Since the vegetation index difference map DI is a single-channel image, the saliency feature value is the gray value of the pixel.
4.2, generating a binarized saliency difference map S by using an OTSU (maximum inter-class variance method) image segmentation algorithm on the saliency map Sam, wherein the pixel value of a non-salient region is 0, and the pixel value of the salient region is 1; then generating a saliency map DS by using the binarized saliency map S and a vegetation index difference map DI:
DS(x,y)＝S(x,y)·DI(x,y)。
step 5, pre-classifying the saliency difference map DS by using a fuzzy C-means clustering algorithm, and dividing pixels in the DS into a changed class, an unchanged class and an uncertain class, wherein a changed class pixel set, an unchanged class pixel set and an uncertain class pixel set are correspondingly obtained;
5.1, setting an objective function of a fuzzy C-means clustering algorithm as follows:
wherein c is the number of categories, n is the total number of pixels of the saliency map DS, u ij Representative pixel x j Membership belonging to the ith cluster, with a value of [0,1]And (2) andm represents a fuzzy weight coefficient, d ij ＝||x j -v i I, denote pixel x j And cluster center v i Is a euclidean distance of (c).
In a specific operation, the invention sets c=3, m=2, and the threshold epsilon=10 -5 Maximum number of iterations t=10 3 The method comprises the steps of carrying out a first treatment on the surface of the Initializing the iteration times t=1, and randomly initializing three clustering centers v 1 、v 2 、v 3 ；
5.2, calculating a new cluster center and a membership function:
if the current iteration algebra t=1, the membership degree of each pixel is directly updated according to the pixel value of the clustering center; if the current iteration algebra t is more than or equal to 2, updating the clustering center by the membership matrix obtained by the previous generation, and updating the membership matrix according to the clustering center. The specific formula is as follows:
wherein k=1, 2,3;represents the ith cluster center in the t-th iteration, < >>Representing pixel x in the t-th iteration j Membership belonging to the ith cluster;
5.3 ifAnd T is less than or equal to T, let t=t+1, return to step 5.2; otherwise, the iteration is terminated, and the class label is distributed according to the final membership of each pixel: for pixel x j If arg k {max{u kj }}＝arg k {max{v k }, and max { u } kj The pixel belongs to a variation class if the pixel is more than or equal to 0.9; arg (arg) k {max{v k -representing class k, arg with largest cluster center pixel value k {max{u kj Pixel x is represented by } } j A category corresponding to the maximum membership of the group (a); if arg k {max{u kj }}＝arg k {min{v k }, and max { u } kj The pixel belongs to unchanged class if the number of the pixels is more than or equal to 0.9; in other cases, the pixel belongs to an uncertainty class. The pixels of the already determined classes, i.e. the changed class and the unchanged class, are no longer calculated in the algorithm after step 6.
Step 6, optimizing the class labels of the pixels in the uncertain class pixel set by using a genetic algorithm based on the self-adaptive parameters to obtain optimized class labels corresponding to the uncertain classes; and generating a final change chart according to the optimized category label.
6.1, initializing a population:
setting the size of a population as num=40, namely, 40 individuals coded into a binary two-dimensional matrix in the population, wherein the size of an individual matrix Ind is the same as the size of a saliency difference map DS, namely, 350×200; initializing all individual matrixes Ind according to the class label obtained in the step 5, namely modifying the pixel value of the position corresponding to the changed class pixel set into 1, modifying the pixel value of the position corresponding to the unchanged class pixel set into 0, and randomly encoding the pixel value of the position corresponding to the uncertain class pixel set into 0 or 1; establishing a position index of the uncertain pixel in the individual matrix, setting iteration times k' =1 and a threshold valueε'＝10 -2 Maximum number of iterations k=10 5 ；
6.2, calculating individual fitness:
and determining the probability of selecting the individual to execute genetic operation according to the fitness value of the individual and the objective function. Unlike traditional genetic algorithm, the improved genetic algorithm used in the invention introduces spatial information OF pixels, designs an objective function OF with a neighborhood factor, and has the following calculation formula:
wherein r is a category identifier, r=0, 1, r=0 corresponds to an unchanged category, and r=1 corresponds to a changed category; c r All pixels representing the corresponding class in the saliency region, n r And m r Representing the statistical number and average gray value of the corresponding class of pixels, N q Representing the center pixel x j 3 x 3 neighborhood N of (2) j Where q=1, 2,3., 9,l q Is N q And center pixel x j L is the Euclidean distance of pixel x j Distance weights from its surrounding pixels; DS (DS) j Representing the center pixel x in the saliency difference map DS j And the pixel corresponding to the position.
After obtaining the objective function of each individual, taking the reciprocal of the objective function to obtain the Fitness Fitness of each individual:
6.3, performing elite selection and crossover operation on the population:
if the current iteration algebra k' =1, carrying out genetic operation on all individuals; if the current iteration algebra k '. Gtoreq.2, the current iteration individual is a child individual, and the k' -1 generation individual is a parent individualThen, according to the fitness values of all individuals of the offspring and the father, selecting Num individuals from high to low to execute genetic operation, namely selecting 40 individuals, selecting the individual with the highest fitness value as the best individual of the current iteration algebra, storing the best fitness value in a matrix Maxfit, and eliminating unselected individuals in evolution. Then, the individuals selected to perform the genetic operation are subjected to the crossover operation, which is performed by the following steps: firstly, grouping all individuals performing genetic operation in pairs to form a plurality of groups of individuals to be crossed; initializing a matrix A with the same size as the individual between (0, 1) at random, traversing the elements of the matrix A, and when the value of an element is smaller than the preset crossover probability p c =0.8, then the position of the element is marked; then exchanging the elements with the same positions as the marks in each group of individuals to be crossed, wherein the other positions are unchanged, so as to form a crossed population;
6.4, calculating the variation probability of each individual in the crossed population pixel by pixel:
for each individual in the crossed population, traversing the individual matrix for the target pixel x j 3×3 neighborhood L of (2) j Each neighborhood pixel L in (1) j (s), s=1, 2, 3..9, calculating the membership u of which belongs to class r sr ：
Wherein r=0, 1: m is m r Representing the average gray values of the pixels of each class obtained in the step 6.2);
according to membership u sr For neighborhood pixel L j (s) assign class labels if u s0 ≥u s1 C is s =0, otherwise, c s =1; thus, the target pixel x j The probability of variation of (2) is:
wherein d s Is a neighborhood pixel L j (s) to the center pixel point x j Spatial distance of B j Is x j Binary encoded values of (a);
if the obtained variation probability p (x j ) If the standard variation probability p=0.2 is larger than the preset standard variation probability p=0.2, the coding of the position is changed, namely 0 is changed to 1 or 1 is changed to 0, otherwise, the coding value is unchanged; traversing each individual matrix to obtain a mutated population, and taking the mutated population as a offspring population of the previous generation; calculating the fitness of each individual in the child population, and selecting the optimal fitness with the largest fitness as the child;
6.5, if the difference of the optimal fitness obtained by the algebra of two adjacent iterations is larger than the set threshold value, i.e. |MaxFit k'+1 -Maxfit k' The I is not less than epsilon ', K' is not more than K, K '=k' +1 is returned to the step 6.3, otherwise, the individual with the highest adaptability of the iteration is output;
and 6.6, supplementing the class labels of the pixels corresponding to the changed class and the unchanged class obtained in the step 5 on the best output individual, and outputting a final binary change graph with the labels.
Simulation experiment
The correctness and validity of the invention are further illustrated by the simulation data processing results.
The simulation content: the two multispectral forest images shown in fig. 2 (a) and 2 (b) are subjected to change detection by using an original genetic algorithm and the algorithm of the invention, and the results are shown in fig. 2 (c) and 2 (d), wherein fig. 2 (c) is a change chart obtained by the original genetic algorithm, and fig. 2 (d) is a change chart obtained by the algorithm of the invention.
Finally, the average classification accuracy was counted, and the results are shown in table 1.
Table 1 detection accuracy of original genetic algorithm and algorithm of the present invention
| Algorithm | Detection accuracy |
| Original genetic algorithm | 69.5％ |
| The algorithm of the invention | 82.2％ |
As can be seen from fig. 2 (c) and 2 (d), the method of the invention can realize accurate ground object classification on the remote sensing image, can limit elimination of interference factors in forest ground objects, has stronger robustness, and realizes accurate detection. And as can be seen from table 1, the invention has higher detection accuracy.
According to the invention, the NDVI spectrum characteristic image of the forest remote sensing image is used as a change detection image to be separated, and denoising and significance detection pretreatment are carried out on the NDVI spectrum characteristic image. In order to further reduce the search space of the change detection algorithm, fuzzy C-means clustering is used, class labels are distributed to pixels with membership degrees larger than a threshold value according to gray values, and genetic algorithm classification based on self-adaptive parameters is carried out on the pixels with membership degrees smaller than the threshold value.
The method is different from the traditional genetic algorithm, introduces the spatial neighborhood information of the image in the objective function and the variation probability, adaptively modifies the genetic operator parameters, avoids the problems of poor noise immunity and huge search space of the traditional genetic algorithm classification model, effectively balances noise suppression and detail reservation, ensures that evolution is rapidly converged towards the optimal solution direction, greatly improves the detection precision of the algorithm, further obtains the variation information of forest areas more accurately, and enhances the detection capability of forest land variation.
While the invention has been described in detail in this specification with reference to the general description and the specific embodiments thereof, it will be apparent to one skilled in the art that modifications and improvements can be made thereto. Accordingly, such modifications or improvements may be made without departing from the spirit of the invention and are intended to be within the scope of the invention as claimed.
Claims (8)
1. The forest change remote sensing detection method based on the adaptive parameter genetic algorithm is characterized by comprising the following steps of:
step 1, obtaining two forest remote sensing images I at different moments 1 And I 2 Preprocessing each forest remote sensing image to obtain two preprocessed forest remote sensing images I 1 "and I 2 "; wherein I is 1 And I 2 The sizes are the same, M is multiplied by N is multiplied by B, M, N respectively represents the number of rows and columns of the image, B represents the number of spectral bands of the image, M is more than or equal to 20, N is more than or equal to 20, and B is more than or equal to 4;
step 2, respectively calculating two preprocessed forest remote sensing images I by utilizing reflectivity of spectrum wave bands 1 ”、I 2 "normalized vegetation index matrix NDVI 1 、NDVI 2 And make it be correspondent vegetation index characteristic map SF 1 、SF 2 ；
Step 3, for vegetation index feature map SF 1 And SF (sulfur hexafluoride) 2 Respectively performing median filtering to correspondingly obtain denoised characteristic images SF 1 '、SF 2 'A'; using a difference method to denoise the denoised characteristic map SF 1 '、SF 2 ' construct a vegetation index difference map DI;
step 4, performing significance detection on the vegetation index difference map DI to obtain a significance difference map DS;
step 5, pre-classifying the saliency difference map DS by using a fuzzy C-means clustering algorithm, and dividing pixels in the DS into a changed class, an unchanged class and an uncertain class, wherein a changed class pixel set, an unchanged class pixel set and an uncertain class pixel set are correspondingly obtained;
step 6, optimizing the class labels of the pixels in the uncertain class pixel set by using a genetic algorithm based on the self-adaptive parameters to obtain optimized class labels corresponding to the uncertain classes;
the genetic algorithm based on the adaptive parameters is used for optimizing the class labels of the pixels in the uncertain class pixel set, and specifically comprises the following steps:
6.1, initializing a population: setting the size of a population as Num, namely, the population has Num individuals coded into a binary two-dimensional matrix, and the size of an individual matrix Ind is the same as the size of a saliency difference map DS; initializing all individual matrixes Ind in the population according to the class label obtained in the step 5, namely modifying the pixel value of the position corresponding to the changed class pixel set into 1, modifying the pixel value of the position corresponding to the unchanged class pixel set into 0, and randomly encoding the pixel value of the position corresponding to the uncertain class pixel set into 0 or 1; establishing a position index of the uncertain class pixels in the individual matrix, setting iteration times K '=1, a threshold epsilon' and a maximum iteration time K;
6.2, calculating individual fitness:
introducing spatial information OF pixels, and designing an objective function OF with a neighborhood factor:
wherein r is a category identifier, r=0, 1, r=0 corresponds to an unchanged category, and r=1 corresponds to a changed category; c r All pixels representing the corresponding class in the saliency region, n r And m r Representing the statistical number and average gray value of the corresponding class of pixels, N q Representing the center pixel x j 3 x 3 neighborhood N of (2) j The q-th pixel, l q Is N q And center pixel x j L is the Euclidean distance of pixel x j Distance weights from its surrounding pixels; DS (DS) j Representing the center pixel x in the saliency difference map DS j Pixels corresponding to the positions;
after obtaining the objective function of each individual, taking the reciprocal of the objective function to obtain the Fitness Fitness of each individual:
6.3, performing elite selection and crossover operation on the population:
if the current iteration algebra k' =1, carrying out genetic operation on all individuals; if the current iteration algebra k '. Gtoreq.2, the current iteration individual is a child individual, and the k' -1 generation individual is a parent individual, then, selecting Num individuals from high to low according to the fitness values of all the child and parent individuals, performing genetic operation, selecting the individual with the highest fitness value as the best individual of the current iteration algebra, and storing the best fitness value in a matrix Maxfit, wherein the unselected individuals are eliminated in evolution; then, intersecting the individuals selected to perform genetic operation to obtain intersecting populations;
6.4, calculating the variation probability of each individual in the crossed population pixel by pixel:
if the obtained variation probability p (x j ) If the code value is larger than the preset standard variation probability p, the code of the position is changed, namely 0 is changed to 1 or 1 is changed to 0, otherwise, the code value is unchanged; traversing each individual matrix to obtain a mutated population, and taking the mutated population as a offspring population of the previous generation; calculating the fitness of each individual in the child population, and selecting the optimal fitness with the largest fitness as the child;
6.5, if the difference of the optimal fitness obtained by the algebra of two adjacent iterations is larger than the set threshold value, i.e. |MaxFit k'+1 -Maxfit k' The I is not less than epsilon ', K' is not more than K, K '=k' +1 is returned to the step 6.3, otherwise, the individual with the highest adaptability of the iteration is output; determining a class label of the uncertain pixel set after optimization according to the individual;
and generating a final change chart according to the optimized category label.
2. The method for remote sensing forest variation detection based on adaptive parameter genetic algorithm as set forth in claim 1, wherein in step 1, the preprocessing is a process of preprocessing the input signal to the input signal 1 And I 2 And respectively carrying out radiation correction, and then carrying out atmosphere correction on the images after the radiation correction.
3. The method for remotely sensing forest variation according to claim 2, wherein in step 2, the calculation formula of the normalized vegetation index matrix is:
wherein ρ is NIR And ρ RED Respectively representing the reflectivity of an infrared spectrum band and the reflectivity of a red spectrum band of the preprocessed forest remote sensing image;
enabling each preprocessed forest remote sensing image I to 1 ”、I 2 "normalized vegetation index matrix NDVI obtained 1 、NDVI 2 Vegetation index feature map SF corresponding to forest image 1 、SF 2 The image size is mxn.
4. The method for remote sensing forest variation detection based on adaptive parameter genetic algorithm according to claim 1, wherein the denoised feature map SF is obtained by using a difference method 1 '、SF 2 The' constructed vegetation index difference map DI is specifically: feature map SF after denoising 1 '、SF 2 The pixel values of the corresponding positions are differenced, and the differenced result is used as the pixel value of the corresponding position of the index difference map DI.
5. The remote sensing method for forest variation based on the adaptive parameter genetic algorithm according to claim 1, wherein the significance detection of the vegetation index difference map DI comprises the following specific steps:
4.1, calculating the global contrast of each pixel (x, y) in the vegetation index difference map DI on the DI, taking the sum of Euclidean distances of the pixel and all other pixels (i, j) in the DI on the color as the saliency characteristic of the pixel, wherein the calculation formula is as follows:
wherein Sam (x, y) is the saliency feature of pixel (x, y), and (x, y) + (i, j); representation euclidean distance; then, sam (x, y) is normalized to [0,255] to obtain a saliency map Sam;
4.2, generating a binarized saliency difference map S by using a maximum inter-class variance method on the saliency map Sam; then generating a saliency map DS by using the binarized saliency map S and a vegetation index difference map DI:
DS(x,y)＝S(x,y)·DI(x,y)。
6. the remote sensing detection method for forest variation based on the adaptive parameter genetic algorithm according to claim 1, wherein the pre-classification of the saliency difference map DS by using the fuzzy C-means clustering algorithm is specifically as follows:
5.1, setting an objective function of a fuzzy C-means clustering algorithm as follows:
wherein c is the number of categories, n is the total number of pixels of the saliency map DS, u ij Representative pixel x j Membership belonging to the ith cluster, with a value of [0,1]And (2) andm represents a fuzzy weight coefficient, d ij ＝||x j -v i I, denote pixel x j And cluster center v i Is the euclidean distance of (2);
c, m, a threshold epsilon and a maximum iteration number T are set; initializing the iteration times t=1, and randomly initializing three clustering centers v 1 、v 2 、v 3 ；
5.2, calculating a new cluster center and a membership function:
if the current iteration algebra t=1, the membership degree of each pixel is directly updated according to the pixel value of the clustering center; if the current iteration algebra t is more than or equal to 2, updating a clustering center by using the membership matrix obtained by the previous generation, and updating the membership matrix according to the clustering center; the specific formula is as follows:
wherein k=1, 2,3;represents the ith cluster center in the t-th iteration, < >>Representing pixel x in the t-th iteration j Membership belonging to the ith cluster;
5.3 ifAnd T is less than or equal to T, let t=t+1, return to step 5.2; otherwise, the iteration is terminated, and the class label is distributed according to the final membership of each pixel: for pixel x j If arg k {max{u kj }}＝arg k {max{v k }, and max { u } kj The pixel belongs to a variation class if the pixel is more than or equal to 0.9; arg (arg) k {max{v k -representing class k, arg with largest cluster center pixel value k {max{u kj Pixel x is represented by } } j A category corresponding to the maximum membership of the group (a); if arg k {max{u kj }}＝arg k {min{v k }, and max { u } kj The pixel belongs to unchanged class if the number of the pixels is more than or equal to 0.9; in other cases, the pixel belongs to an uncertainty class.
7. The adaptive parameter-based of claim 1The forest variation remote sensing detection method of the digital genetic algorithm is characterized in that the intersecting operation is carried out on the individuals selected to execute the genetic operation, and the specific method comprises the following steps: firstly, grouping all individuals performing genetic operation in pairs to form a plurality of groups of individuals to be crossed; initializing a matrix A with the same size as the individual between (0, 1) at random, traversing the elements of the matrix A, and when the value of an element is smaller than the preset crossover probability p c Marking the location of the element; and then exchanging the elements with the same positions as the marks in each group of individuals to be crossed, wherein the other positions are unchanged, so as to form a crossed population.
8. The method for remotely sensing forest variation based on the adaptive parameter genetic algorithm according to claim 1, wherein the calculating the variation probability of each individual in the crossed population pixel by pixel comprises the following specific steps:
for each individual in the crossed population, traversing the individual matrix for the target pixel x j 3×3 neighborhood L of (2) j Each neighborhood pixel L in (1) j (s), s=1, 2, 3..9, calculating the membership u of which belongs to class r sr ：
Wherein r=0, 1: m is m r Representing the average gray values of the pixels of each class obtained in the step 6.2;
according to membership u sr For neighborhood pixel L j (s) assign class labels if u s0 ≥u s1 C is s =0, otherwise, c s =1; thus, the target pixel x j The probability of variation of (2) is:
wherein d s Is a neighborhood pixel L j (s) to the center pixel point x j Spatial distance of B j Is x j Binary encoded values of (a).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110605694.1A CN113408370B (en) | 2021-05-31 | 2021-05-31 | Forest change remote sensing detection method based on adaptive parameter genetic algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110605694.1A CN113408370B (en) | 2021-05-31 | 2021-05-31 | Forest change remote sensing detection method based on adaptive parameter genetic algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113408370A CN113408370A (en) | 2021-09-17 |
| CN113408370B true CN113408370B (en) | 2023-12-19 |
Family
ID=77675513
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110605694.1A Active CN113408370B (en) | 2021-05-31 | 2021-05-31 | Forest change remote sensing detection method based on adaptive parameter genetic algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113408370B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116109176B (en) * | 2022-12-21 | 2024-01-05 | 成都安讯智服科技有限公司 | Alarm abnormity prediction method and system based on collaborative clustering |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101561882A (en) * | 2009-05-22 | 2009-10-21 | 中国科学院地理科学与资源研究所 | Sub-pixel spatial mapping method based on spatial correlation |
| CN103020978A (en) * | 2012-12-14 | 2013-04-03 | 西安电子科技大学 | SAR (synthetic aperture radar) image change detection method combining multi-threshold segmentation with fuzzy clustering |
| CN103971362A (en) * | 2013-12-24 | 2014-08-06 | 西安电子科技大学 | Synthetic aperture radar (SAR) imagine change detection based on histogram and elite genetic clustering algorithm |
| CN104268574A (en) * | 2014-09-25 | 2015-01-07 | 西安电子科技大学 | SAR image change detecting method based on genetic kernel fuzzy clustering |
| CN104751185A (en) * | 2015-04-08 | 2015-07-01 | 西安电子科技大学 | SAR image change detection method based on mean shift genetic clustering |
| CN105374047A (en) * | 2015-12-15 | 2016-03-02 | 西安电子科技大学 | Improved bilateral filtering and clustered SAR based image change detection method |
| CN109409292A (en) * | 2018-10-26 | 2019-03-01 | 西安电子科技大学 | The heterologous image matching method extracted based on fining characteristic optimization |
| CN110135322A (en) * | 2019-05-09 | 2019-08-16 | 航天恒星科技有限公司 | A kind of time series forest change monitoring method based on IFI |
| CN111666981A (en) * | 2020-05-13 | 2020-09-15 | 云南电网有限责任公司信息中心 | System data anomaly detection method based on genetic fuzzy clustering |
| CN112149492A (en) * | 2020-07-06 | 2020-12-29 | 北京航空航天大学 | Remote sensing image accurate cloud detection method based on reinforcement genetic learning |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7625824B2 (en) * | 2005-06-16 | 2009-12-01 | Oerlikon Usa, Inc. | Process change detection through the use of evolutionary algorithms |
| US9946931B2 (en) * | 2015-04-20 | 2018-04-17 | Los Alamos National Security, Llc | Change detection and change monitoring of natural and man-made features in multispectral and hyperspectral satellite imagery |
| US11785877B2 (en) * | 2018-10-31 | 2023-10-17 | Climate Llc | Automated sample collection and tracking system |
| CN110472525B (en) * | 2019-07-26 | 2021-05-07 | 浙江工业大学 | Noise detection method for time series remote sensing vegetation index |
-
2021
- 2021-05-31 CN CN202110605694.1A patent/CN113408370B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101561882A (en) * | 2009-05-22 | 2009-10-21 | 中国科学院地理科学与资源研究所 | Sub-pixel spatial mapping method based on spatial correlation |
| CN103020978A (en) * | 2012-12-14 | 2013-04-03 | 西安电子科技大学 | SAR (synthetic aperture radar) image change detection method combining multi-threshold segmentation with fuzzy clustering |
| CN103971362A (en) * | 2013-12-24 | 2014-08-06 | 西安电子科技大学 | Synthetic aperture radar (SAR) imagine change detection based on histogram and elite genetic clustering algorithm |
| CN104268574A (en) * | 2014-09-25 | 2015-01-07 | 西安电子科技大学 | SAR image change detecting method based on genetic kernel fuzzy clustering |
| CN104751185A (en) * | 2015-04-08 | 2015-07-01 | 西安电子科技大学 | SAR image change detection method based on mean shift genetic clustering |
| CN105374047A (en) * | 2015-12-15 | 2016-03-02 | 西安电子科技大学 | Improved bilateral filtering and clustered SAR based image change detection method |
| CN109409292A (en) * | 2018-10-26 | 2019-03-01 | 西安电子科技大学 | The heterologous image matching method extracted based on fining characteristic optimization |
| CN110135322A (en) * | 2019-05-09 | 2019-08-16 | 航天恒星科技有限公司 | A kind of time series forest change monitoring method based on IFI |
| CN111666981A (en) * | 2020-05-13 | 2020-09-15 | 云南电网有限责任公司信息中心 | System data anomaly detection method based on genetic fuzzy clustering |
| CN112149492A (en) * | 2020-07-06 | 2020-12-29 | 北京航空航天大学 | Remote sensing image accurate cloud detection method based on reinforcement genetic learning |
Non-Patent Citations (7)
| Title |
|---|
| Brain Hemisphere Analysis Using Genetic Algorithm and Fuzzy Clustering in Alzheimer Disease;D. Chitradevi 等;《2018 International Conference on Communication and Signal Processing (ICCSP)》;20181108;0901-0905 * |
| Change detection in SAR images based on the salient map guidance and an accelerated genetic algorithm;Caihong Mu 等;《2017 IEEE Congress on Evolutionary Computation (CEC)》;20170707;1150-1157 * |
| 刘骊 等.显著特征提取.《计算机服装建模及仿真》.昆明：云南大学出版社,2018, * |
| 基于差异图和块分类的雷达图像变化检测;王平;《长春师范大学学报》;第39卷(第8期);42-49 * |
| 基于进化优化及卷积网络的遥感图像变化检测;李成洲;《中国优秀硕士学位论文全文数据库 信息科技辑》;20190215;第2019年卷(第2期);摘要、第2.2.2、2.3.1、2.3.2、2.4、2.6、3.1、3.4.2节 * |
| 王桥.植被指数法.《区域生态保护的遥感监管方法与应用》.北京：中国环境科学出版社,2015, * |
| 面向对象的高分辨率影像城市多特征变化检测研究;汤玉奇;《中国博士学位论文全文数据库 信息科技辑》;20140715;第2014年卷(第7期);I140-23 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113408370A (en) | 2021-09-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111767801B (en) | Remote sensing image water area automatic extraction method and system based on deep learning | |
| US11521380B2 (en) | Shadow and cloud masking for remote sensing images in agriculture applications using a multilayer perceptron | |
| CN111898688B (en) | Airborne LiDAR data tree classification method based on three-dimensional deep learning | |
| CN108428220B (en) | Automatic geometric correction method for ocean island reef area of remote sensing image of geostationary orbit satellite sequence | |
| CN109815979B (en) | Weak label semantic segmentation calibration data generation method and system | |
| CN115170979B (en) | Mining area fine land classification method based on multi-source data fusion | |
| CN109063660B (en) | Crop identification method based on multispectral satellite image | |
| CN112950780B (en) | Intelligent network map generation method and system based on remote sensing image | |
| CN113486975A (en) | Ground object classification method, device, equipment and storage medium for remote sensing image | |
| CN111291818B (en) | Non-uniform class sample equalization method for cloud mask | |
| CN113408370B (en) | Forest change remote sensing detection method based on adaptive parameter genetic algorithm | |
| CN116091937A (en) | High-resolution remote sensing image ground object recognition model calculation method based on deep learning | |
| CN113936214B (en) | Karst wetland vegetation community classification method based on fusion of aerospace remote sensing images | |
| CN107292268A (en) | The SAR image semantic segmentation method of quick ridge ripple deconvolution Structure learning model | |
| CN108416297B (en) | A kind of vegetation information method for quickly identifying based on chlorophyll fluorescence | |
| CN108509835B (en) | PolSAR image ground object classification method based on DFIC super-pixels | |
| CN116977960A (en) | Rice seedling row detection method based on example segmentation | |
| CN109726679B (en) | Remote sensing classification error spatial distribution mapping method | |
| CN111882573A (en) | Cultivated land plot extraction method and system based on high-resolution image data | |
| CN104732246B (en) | A kind of semi-supervised coorinated training hyperspectral image classification method | |
| CN109344837B (en) | SAR image semantic segmentation method based on deep convolutional network and weak supervised learning | |
| CN115965867A (en) | Remote sensing image earth surface coverage classification method based on pseudo label and category dictionary learning | |
| Zhao et al. | Improving object-oriented land use/cover classification from high resolution imagery by spectral similarity-based post-classification | |
| CN114419443A (en) | Automatic remote-sensing image cultivated land block extraction method and system | |
| CN113743373A (en) | High-resolution remote sensing image cropland change detection device and method based on deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |