CN113837074B - Remote sensing image change detection method combining posterior probability and space neighborhood information - Google Patents
Remote sensing image change detection method combining posterior probability and space neighborhood information Download PDFInfo
- Publication number
- CN113837074B CN113837074B CN202111118251.6A CN202111118251A CN113837074B CN 113837074 B CN113837074 B CN 113837074B CN 202111118251 A CN202111118251 A CN 202111118251A CN 113837074 B CN113837074 B CN 113837074B
- Authority
- CN
- China
- Prior art keywords
- surface coverage
- change
- neighborhood information
- intensity
- pixel point
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
A remote sensing image change detection method combining posterior probability and space neighborhood information utilizes a double-phase remote sensing image to obtain training samples and change intensity images of various earth surface coverage types, and calculates the posterior probability of the change intensity of the earth surface coverage types. And (3) calculating posterior probability by considering the space neighborhood information, and constructing a space curved surface of the earth surface coverage type-space neighborhood information. A threshold value is adaptively determined for each pixel according to a spatial curvature, and a change detection result image is acquired. For the multi-temporal remote sensing image change detection research, the false change of farmlands can be effectively reduced, a change area with lower gray value in a change intensity image is identified, and the possibility of missing report and false report is reduced to the greatest extent.
Description
Technical Field
The invention relates to the field of multi-temporal remote sensing image change detection, in particular to a remote sensing image change detection method combining posterior probability and space neighborhood information.
Background
The multi-temporal remote sensing change detection plays an increasingly important role in the fields of resource management, environmental protection, surface dynamic monitoring and the like. The direct comparison method is a common and simple change detection method, and consists of two steps: (1) calculating a variation intensity image; (2) obtaining a change detection result. The calculation of the change intensity image is mainly to compare the difference degree between the remote sensing image spectrum values at the time T1 and the time T2. The common method for calculating the change intensity image is as follows: difference method, ratio method, change vector analysis method, etc. The change threshold selection method is a common method in the generation of a change detection result and is also one of important steps directly influencing the detection result. Notably, the selection of the change threshold is a relatively complex process, subject to a variety of factors. In the change detection, different surface coverage types have different magnitudes of change. For example, the spectral variation values of grasslands and woodlands are relatively small, while the spectral variation values of farmlands to construction lands are relatively large. Furthermore, different surface coverage exhibit different spatial heterogeneities depending on the region of investigation and the resolution of the remote sensing image. For example, when the area of investigation is entirely farmland, the spatial heterogeneity is relatively low. All these factors lead to difficulties in the selection of the change threshold.
Over the last three decades, scholars have proposed various methods of change threshold selection, which can be categorized into global change threshold selection and local change threshold selection. The global change threshold selection method generally divides the change intensity map into a change region and an unchanged region using one or more indexes according to the gray histogram distribution situation, for example, an Otsu method, a Kapur method, a Kittler method, a Expectation Maximum (EM) method, and the like. Although convenient and effective, the global variation threshold selection method is to determine the threshold as a whole, and lacks consideration for the type of surface coverage and spatial heterogeneity. The local change threshold selection method generally sets a suitable window (for example, 3×3), divides the change intensity image into sub-areas with the same size, and performs threshold segmentation in the sub-areas. This approach accounts for spatial heterogeneity to some extent, but does not integrate the earth coverage type information. Notably, if the spectral variation differences between the earth coverage types are ignored, the determination of the threshold is prone to over-detection and over-detection problems. In recent years, some scholars have proposed the introduction of surface coverage type information in the change threshold selection method. However, most methods are insufficient in consideration of spatial heterogeneity, and cannot effectively solve the problem of selecting a change threshold of a complex surface coverage type. Therefore, considering the surface coverage type and spatial heterogeneity is important to improve the accuracy of the change detection result.
Disclosure of Invention
The invention provides a remote sensing image change detection method combining posterior probability and space neighborhood information in order to overcome the defects of the technology.
The technical scheme adopted for overcoming the technical problems is as follows:
a remote sensing image change detection method combining posterior probability and space neighborhood information comprises the following steps:
a) Acquiring training samples and change intensity images of all earth surface coverage types by using the double-time-phase remote sensing image, and calculating posterior probability of the change intensity of the earth surface coverage types;
b) Calculating posterior probability of the earth surface coverage type change intensity by utilizing the space neighborhood information, and constructing a space curved surface of the earth surface coverage type-space neighborhood information;
c) And adaptively selecting a threshold according to the surface coverage type-space neighborhood information space curved surface, and determining whether the change intensity image changes.
Further, in step a), a change intensity image is obtained by using a change vector analysis method.
Further, in step a) the formula is passedCalculating to obtain posterior probability I of earth surface coverage type change intensity, wherein M is T 1 Time and T 2 The number of earth surface coverage types at the moment, M is a constant, and the value of M ranges from 1 to M, < ->Is T 1 Class probability of time mth earth surface coverage type,/-)>Is T 2 Class probability of mth earth surface coverage type of time,/->And->Obtained by support vector machine, ">Is T 1 Mth earth coverage type of time, < >>Is T 2 The mth earth surface coverage type at the moment, N is the wave band number of the remote sensing image, N is a constant, and the value range of N is 1 to N,>is T 1 Remote sensing image pixel value of nth wave band at moment, < >>Is T 2 The pixel value of the remote sensing image in the nth wave band at the moment.
Further, in step b) the formula is passedCalculating to obtain the spatial surface value of the earth surface coverage type-spatial neighborhood information of the corresponding pixel point p>In which I bf Spatial surface for earth surface coverage type-spatial neighborhood information, wherein +.>For the normalized weight sum of the corresponding pixel point p,s is the spatial domain, r is the range domain, < >>As a gaussian decreasing function of the spatial domain, +.>P is a pixel point in the center of the field, q is a pixel point adjacent to the pixel point p, and I p-q I is the Euclidean distance between the pixel point p and the pixel point q p Is the intensity value of the pixel point p, I q Is the intensity value of the pixel point q, |I p -I q I is I p And I q A measure of the distance between them.
Further, in step c) the formula is passed
Calculating to obtain a threshold T of the corresponding pixel point p p Where κ is the coefficient for increasing the threshold differentiation, ++> Standard deviation, sigma, of a spatial surface for earth surface coverage type-spatial neighborhood information CMM For varying standard deviation of intensity images +.>Mu, mean value of space curved surface of surface coverage type-space neighborhood information CMM For the average value of the variable intensity image, when the pixel point of the variable intensity image is greater than or equal to the corresponding threshold value T p When the pixel point of the image with changed intensity is smaller than the corresponding threshold value T p And when the pixel point is not changed, obtaining a final change detection result. The beneficial effects of the invention are as follows: acquiring training samples and changes of various earth surface coverage types by using double-time-phase remote sensing imagesAnd (5) calculating the posterior probability of the intensity of the change of the surface coverage type according to the intensity image. And (3) calculating posterior probability by considering the space neighborhood information, and constructing a space curved surface of the earth surface coverage type-space neighborhood information. A threshold value is adaptively determined for each pixel according to a spatial curvature, and a change detection result image is acquired. For the multi-temporal remote sensing image change detection research, the false change of farmlands can be effectively reduced, a change area with lower gray value in a change intensity image is identified, and the possibility of missing report and false report is reduced to the greatest extent.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
fig. 2 is a reference change chart obtained by the present invention, in which (a) is a 2013 remote sensing image, (b) is a 2018 remote sensing image, and (c) is a reference change chart;
FIG. 3 is a graph of the change detection results of the present invention;
FIG. 4 is a graph showing the comparison of the change detection results of the present invention and the conventional method.
Detailed Description
The invention is further described with reference to fig. 1 and 2.
As shown in fig. 1, a remote sensing image change detection method combining posterior probability and spatial neighborhood information includes the following steps:
a) Acquiring training samples and change intensity images of all earth surface coverage types by using the double-time-phase remote sensing image, and calculating posterior probability of the change intensity of the earth surface coverage types;
b) Calculating posterior probability of the earth surface coverage type change intensity by utilizing the space neighborhood information, and constructing a space curved surface of the earth surface coverage type-space neighborhood information;
c) And adaptively selecting a threshold according to the surface coverage type-space neighborhood information space curved surface, and determining whether the change intensity image changes.
As shown in FIG. 2, landsat8-Operational Land Imager (OLI) remote sensing images of 2013, 5, 21 and 2018, 5, 3 were selected, and the study area image size was 376X 350 pixels, located in Leling city, shandong province, china.
And acquiring training samples and change intensity images of each earth surface coverage type by using the double-time-phase remote sensing image, and calculating posterior probability of the change intensity of the earth surface coverage type. And (3) calculating posterior probability by considering the space neighborhood information, and constructing a space curved surface of the earth surface coverage type-space neighborhood information. A threshold value is adaptively determined for each pixel according to a spatial curvature, and a change detection result image is acquired. For the multi-temporal remote sensing image change detection research, the false change of farmlands can be effectively reduced, a change area with lower gray value in a change intensity image is identified, and the possibility of missing report and false report is reduced to the greatest extent.
Example 1:
in step a) a variation intensity image is obtained by means of a variation vector analysis (CVA) method.
Example 2:
in step a) by the formulaCalculating to obtain posterior probability I of earth surface coverage type change intensity, wherein M is T 1 Time and T 2 The number of earth surface coverage types at the moment, M is a constant, and the value of M ranges from 1 to M, < ->Is T 1 Class probability of time mth earth surface coverage type,/-)>Is T 2 Class probability of mth earth surface coverage type of time,/->And->Obtained by support vector machine, ">Is T 1 Mth earth surface of momentType of overlay->Is T 2 The mth earth surface coverage type at the moment, N is the wave band number of the remote sensing image, N is a constant, and the value range of N is 1 to N,>is T 1 Remote sensing image pixel value of nth wave band at moment, < >>Is T 2 The pixel value of the remote sensing image in the nth wave band at the moment. And calculating the posterior probability of the intensity of the change of the surface coverage type of the research area according to the formula.
Example 3:
and integrating the space neighborhood information on the posterior probability of the surface coverage type change intensity through a bilateral filtering formula, so as to construct a space curved surface of the surface coverage type-space neighborhood information. In a specific step b) by the formulaCalculating to obtain the spatial surface value of the earth surface coverage type-spatial neighborhood information of the corresponding pixel point p>In which I bf Spatial surface for earth surface coverage type-spatial neighborhood information, wherein +.>Is the sum of normalized weights of the corresponding pixel point p, +.>s is the spatial domain, r is the range domain, < >>As a gaussian decreasing function of the spatial domain, +.>P is a pixel point in the center of the field, q is a pixel point adjacent to the pixel point p, and I p-q I is the Euclidean distance between the pixel point p and the pixel point q p Is the intensity value of the pixel point p, I q Is the intensity value of the pixel point q, |I p -I q I is I p And I q A measure of the distance between them. And calculating and obtaining the space curved surface of the earth surface coverage type-space neighborhood information according to the formula.
Example 4:
further, in step c), the variation intensity image is normalized to 0-255, and the final threshold is obtained by multiplying 255 by the coefficient κ minus the spatial surface of the earth coverage type-spatial neighborhood information. Thus, the proposed method enables a relatively small threshold to be obtained for a region of variation in the varying intensity image, so that variation is more easily detected, and a relatively large threshold to be obtained for an unchanged region, so that the possibility of detecting a false variation is reduced. The method can better distinguish the changed area from the unchanged area, and improves the adaptability degree of the change threshold selection. The final threshold value corresponding to each pixel point of the variable intensity image is as follows:calculating to obtain a threshold T of the corresponding pixel point p p Where κ is a coefficient for increasing threshold distinguishability, is the sum of discrete coefficients of the spatial surface of the change intensity image and the earth surface coverage type-spatial neighborhood information, +.> Standard deviation, sigma, of a spatial surface for earth surface coverage type-spatial neighborhood information CMM For varying standard deviation of intensity images +.>Mu, mean value of space curved surface of surface coverage type-space neighborhood information CMM For the average value of the variable intensity image, when the pixel point of the variable intensity image is greater than or equal to the corresponding threshold value T p When the pixel point of the image with changed intensity is smaller than the corresponding threshold value T p When the pixel is not changed, the final change detection result (as shown in fig. 3) is obtained.
In order to demonstrate the effectiveness of the proposed method, a comparison analysis was performed with a conventional change detection method, as shown in fig. 4. The comparison method comprises the following steps: CVA-Otsu method, CVA-Kittler method, CVA-EM method. The change detection result is evaluated using the Overall Accuracy (OA), kappa coefficient, omission Ratio (OR), and overstock ratio (CR). The change detection results are shown in the following table:
| CVA-Otsu | CVA-Kittler | CVA-EM | the method is provided | |
| OA(％) | 83.86 | 83.86 | 82.22 | 91.62 |
| kappa coefficient | 0.65 | 0.65 | 0.62 | 0.82 |
| CR(％) | 35.11 | 35.11 | 40.44 | 18.52 |
| OR(％) | 1.36 | 1.36 | 0.14 | 0.48 |
Experiments prove that in a research area, the CVA-Otsu method and the CVA-Kittler method have the same selected threshold value and have obvious overstock phenomenon. There is also more overdetection with the CVA-EM method, but the method has the lowest miss rate. Although the omission ratio of the proposed method is not the lowest, it can be seen in conjunction with fig. 4 that the proposed method is closest to the real variation. In the research of multi-temporal remote sensing image change detection, the invention can effectively reduce pseudo-change of farmland, identify a change region with lower gray value in a change intensity image, and furthest reduce the possibility of missing report and false report.
Finally, it should be noted that: the foregoing description is only a preferred embodiment of the present invention, and the present invention is not limited thereto, but it is to be understood that modifications and equivalents of some of the technical features described in the foregoing embodiments may be made by those skilled in the art, although the present invention has been described in detail with reference to the foregoing embodiments. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (2)
1. A remote sensing image change detection method combining posterior probability and space neighborhood information is characterized by comprising the following steps:
a) Acquiring training samples and change intensity images of all earth surface coverage types by using the double-time-phase remote sensing image, and calculating posterior probability of the change intensity of the earth surface coverage types;
b) Calculating posterior probability of the earth surface coverage type change intensity by utilizing the space neighborhood information, and constructing a space curved surface of the earth surface coverage type-space neighborhood information;
c) According to the surface coverage type-space neighborhood information space curved surface, adaptively selecting a threshold value, and determining whether the change intensity image changes or not;
in step a) by the formulaCalculating to obtain posterior probability I of earth surface coverage type change intensity, wherein M is T 1 Time and T 2 The number of earth surface coverage types at the moment, M is a constant, and the value of M ranges from 1 to M, < ->Is T 1 Class probability of time mth earth surface coverage type,/-)>Is T 2 Class probability of mth earth surface coverage type of time,/->And->Obtained by support vector machine, ">Is T 1 Mth earth coverage type of time, < >>Is T 2 The mth earth surface coverage type at the moment, N is the wave band number of the remote sensing image, N is a constant, and the value range of N is 1 to N,>is T 1 Remote sensing image pixel value of nth wave band at moment, < >>Is T 2 The pixel value of the remote sensing image in the nth wave band at the moment;
in step b) by the formulaCalculating to obtain the spatial surface value of the earth surface coverage type-spatial neighborhood information of the corresponding pixel point p>In which I bf Spatial surface for earth surface coverage type-spatial neighborhood information, wherein +.>Is the sum of normalized weights of the corresponding pixel point p, +.>s is the spatial domain, r is the range domain, < >>As a gaussian decreasing function of the spatial domain, +.>Is Gaussian decreasing function of the range domain, p is pixel point of the domain center, q is pixel point adjacent to the pixel point p, and p-q is pixelEuclidean distance between point p and pixel point q, I p Is the intensity value of the pixel point p, I q Is the intensity value of the pixel point q, I p -I q Is I p And I q A measure of the distance between them;
in step c) by the formula
Calculating to obtain a threshold T of the corresponding pixel point p p Where κ is the coefficient for increasing the threshold differentiation, ++> Standard deviation, sigma, of a spatial surface for earth surface coverage type-spatial neighborhood information CMM For varying standard deviation of intensity images +.>Mu, mean value of space curved surface of surface coverage type-space neighborhood information CMM For the average value of the variable intensity image, when the pixel point of the variable intensity image is greater than or equal to the corresponding threshold value T p When the pixel point of the image with changed intensity is smaller than the corresponding threshold value T p And when the pixel point is not changed, obtaining a final change detection result.
2. The method for detecting the change of the remote sensing image by combining posterior probability and spatial neighborhood information according to claim 1, wherein the method is characterized by comprising the following steps of: and a step a) of obtaining a variation intensity image by using a variation vector analysis method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111118251.6A CN113837074B (en) | 2021-09-24 | 2021-09-24 | Remote sensing image change detection method combining posterior probability and space neighborhood information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111118251.6A CN113837074B (en) | 2021-09-24 | 2021-09-24 | Remote sensing image change detection method combining posterior probability and space neighborhood information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113837074A CN113837074A (en) | 2021-12-24 |
| CN113837074B true CN113837074B (en) | 2023-08-11 |
Family
ID=78969738
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111118251.6A Active CN113837074B (en) | 2021-09-24 | 2021-09-24 | Remote sensing image change detection method combining posterior probability and space neighborhood information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113837074B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115410096B (en) * | 2022-11-03 | 2023-01-24 | 成都国星宇航科技股份有限公司 | Satellite remote sensing image multi-scale fusion change detection method, medium and electronic device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101694719A (en) * | 2009-10-13 | 2010-04-14 | 西安电子科技大学 | Method for detecting remote sensing image change based on non-parametric density estimation |
| CN102169584A (en) * | 2011-05-28 | 2011-08-31 | 西安电子科技大学 | Remote sensing image change detection method based on watershed and treelet algorithms |
| CN103488968A (en) * | 2012-06-14 | 2014-01-01 | 株式会社日立制作所 | Device and method for constituting fining decomposer through mixed pixel materials of remote sensing image |
| CN111008644A (en) * | 2019-10-30 | 2020-04-14 | 西安电子科技大学 | Ecological change monitoring method based on local dynamic energy function FCN-CRF model |
| CN111259784A (en) * | 2020-01-14 | 2020-06-09 | 西安理工大学 | SAR image change detection method based on transfer learning and active learning |
| CN112419266A (en) * | 2020-11-23 | 2021-02-26 | 山东建筑大学 | Remote sensing image change detection method based on surface coverage category constraint |
-
2021
- 2021-09-24 CN CN202111118251.6A patent/CN113837074B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101694719A (en) * | 2009-10-13 | 2010-04-14 | 西安电子科技大学 | Method for detecting remote sensing image change based on non-parametric density estimation |
| CN102169584A (en) * | 2011-05-28 | 2011-08-31 | 西安电子科技大学 | Remote sensing image change detection method based on watershed and treelet algorithms |
| CN103488968A (en) * | 2012-06-14 | 2014-01-01 | 株式会社日立制作所 | Device and method for constituting fining decomposer through mixed pixel materials of remote sensing image |
| CN111008644A (en) * | 2019-10-30 | 2020-04-14 | 西安电子科技大学 | Ecological change monitoring method based on local dynamic energy function FCN-CRF model |
| CN111259784A (en) * | 2020-01-14 | 2020-06-09 | 西安理工大学 | SAR image change detection method based on transfer learning and active learning |
| CN112419266A (en) * | 2020-11-23 | 2021-02-26 | 山东建筑大学 | Remote sensing image change detection method based on surface coverage category constraint |
Non-Patent Citations (1)
| Title |
|---|
| Histogram thresholding for unsupervised change detection of remote sensing images;Swarnajyoti Patra et al.;International Journal of Remote Sensing;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113837074A (en) | 2021-12-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102968798B (en) | SAR (Synthetic Aperture Radar) image sea-land segmentation method based on wavelet transform and OTSU threshold | |
| CN107292310B (en) | Visual positioning and automatic reading method for circular pointer type dial plate | |
| CN111783583B (en) | SAR image speckle suppression method based on non-local mean algorithm | |
| CN107123150A (en) | The method of global color Contrast Detection and segmentation notable figure | |
| CN108564085A (en) | A kind of method of automatic reading pointer type instrument reading | |
| CN109741356A (en) | A kind of sub-pixel edge detection method and system | |
| CN116188496B (en) | Remote sensing image self-adaptive segmentation method based on land utilization type | |
| CN108830899B (en) | Circle center detection method based on geometric feature and capable of realizing rapid self-adaption corner point combination | |
| CN116012364B (en) | SAR image change detection method and device | |
| CN107256409B (en) | High-resolution SAR image change detection method based on SAE and significance detection | |
| CN113628234B (en) | Method for detecting salient polarization SAR image change based on comprehensive neighborhood information | |
| CN101976436A (en) | Pixel-level multi-focus image fusion method based on correction of differential image | |
| CN109461132A (en) | SAR image automatic registration method based on feature point geometric topological relation | |
| CN107392887B (en) | Heterogeneous remote sensing image change detection method based on homogeneous pixel point conversion | |
| CN113837074B (en) | Remote sensing image change detection method combining posterior probability and space neighborhood information | |
| CN112308873A (en) | Edge detection method for multi-scale Gabor wavelet PCA fusion image | |
| CN114677525B (en) | Edge detection method based on binary image processing | |
| CN104680536A (en) | Method for detecting SAR image change by utilizing improved non-local average algorithm | |
| CN117911408B (en) | Road pavement construction quality detection method and system | |
| CN117541654B (en) | Detail enhancement method for high-resolution remote sensing image | |
| CN106940782A (en) | High score SAR based on variogram increases construction land newly and extracts software | |
| CN110070545B (en) | Method for automatically extracting urban built-up area by urban texture feature density | |
| CN108986083B (en) | SAR image change detection method based on threshold optimization | |
| CN107369163B (en) | Rapid SAR image target detection method based on optimal entropy dual-threshold segmentation | |
| CN116740579A (en) | Intelligent collection method for territorial space planning data |
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 |