CN111160199A - Highway disaster information detection method based on high-resolution remote sensing image - Google Patents
Highway disaster information detection method based on high-resolution remote sensing image Download PDFInfo
- Publication number
- CN111160199A CN111160199A CN201911340911.8A CN201911340911A CN111160199A CN 111160199 A CN111160199 A CN 111160199A CN 201911340911 A CN201911340911 A CN 201911340911A CN 111160199 A CN111160199 A CN 111160199A
- Authority
- CN
- China
- Prior art keywords
- road
- disaster
- remote sensing
- sensing image
- damage
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
- G06V10/464—Salient features, e.g. scale invariant feature transforms [SIFT] using a plurality of salient features, e.g. bag-of-words [BoW] representations
Abstract
The invention relates to a highway disaster information detection method based on a high-resolution remote sensing image, which comprises the following steps: acquiring images before and after a disaster; performing road extraction on the optical remote sensing image before the disaster; registering the pre-disaster optical remote sensing image and the post-disaster SAR remote sensing image to achieve a prior condition of accurate registration; road range mapping and road damage area extraction. The method solves the problems that the existing method has high requirements on near-infrared wave bands and the road extraction precision on post-disaster images cannot be guaranteed. The method solves the problems that the radiation and geometric characteristics of an optical image and an SAR image are obviously different, and the same-name ground objects before and after a disaster are obviously changed or even disappear, so that the traditional image registration method based on single measurement is increased in registration difficulty and insufficient in reliability.
Description
Technical Field
The invention relates to a highway disaster information detection method, in particular to a highway disaster information detection method based on a high-resolution remote sensing image.
Background
When an earthquake occurs in a plateau and mountain area, ground traffic infrastructures (roads, railways and the like) are usually damaged in different degrees, and rescue workers are difficult to get disaster information from the ground to a disaster area in the first time. The remote sensing technology has the advantages of fast imaging, wide coverage range, rich data information quantity, no restriction of ground conditions and the like, and becomes the most effective means for acquiring disaster data after an earthquake at the present stage. The highway is a gold line and a life line for earthquake-proof rescue in the disaster area, and the remote sensing technology is utilized to obtain the road damage condition of the earthquake disaster area and evaluate secondary disasters, so that scientific basis can be provided for road emergency and security, rescue route planning and post-disaster reconstruction, and the highway has important significance and value.
At present, most of remote sensing detection methods for road disaster information directly extract road information from post-disaster remote sensing images, and then detect damage conditions, and the specific methods are divided into 2 types:
(1) and directly extracting road damage information from the remote sensing image after the disaster. For example, based on post-disaster optical remote sensing images, damaged road sections are extracted through an object-oriented classification and change detection method; based on the post-disaster SAR remote sensing image, extracting road damage information and the like by using methods such as threshold segmentation, morphological analysis, wavelet transformation and the like;
(2) the method comprises the steps of extracting the distribution condition of undamaged roads from post-disaster remote sensing images, combining prior GIS road information in a disaster area range, and indirectly obtaining disaster road damage information by adopting a superposition spatial analysis method.
The above method mainly has the following problems:
(1) whichever method is adopted, the road extraction operation needs to be carried out on the post-disaster remote sensing image. However, the road is seriously damaged after a disaster (such as an earthquake) occurs, and compared with a complete road, the radiation, geometric and spectral characteristics of the damaged road on a remote sensing image are obviously changed, the extraction difficulty is higher, and the precision and the accuracy cannot be ensured;
(2) most disasters (such as earthquakes) occur in areas with rare people, the updating frequency of GIS road information in the areas is low, the situation and the precision cannot be guaranteed, and the accuracy and the reliability of road damage detection assisted by the GIS data are low;
(3) the existing remote sensing image road extraction method has uncertainty in accuracy, generally only utilizes spectral information of the remote sensing image, and ignores a lot of other useful auxiliary information. In addition, the existing method has high requirements on near-infrared bands, and no effective method is formed for road extraction on RGB remote sensing images without near-infrared bands;
with the rapid development of global earth observation technology and the implementation of 'high-grade important special items' in China in recent years, high-quality optical images before disasters occur are easy to obtain. However, after a disaster occurs, meteorological conditions in a disaster area are generally severe, and rainy, cloudy and foggy weather frequently occurs, so that the Synthetic Aperture Radar (SAR) with all-weather earth observation capability has obvious advantages in acquiring data after the disaster in all days. Therefore, the optical remote sensing image before the disaster and the SAR remote sensing image after the disaster are comprehensively utilized to detect and evaluate the road disaster information, the respective advantages of the two data can be fully exerted, and the accuracy of detection and evaluation is effectively improved.
Disclosure of Invention
In order to solve the technical problems, the invention provides a highway disaster information detection method based on a high-resolution remote sensing image.
Firstly, on the optical remote sensing image before disaster, only the road damage information detection of RGB wave band is relied on, and the problems that the requirement of the current method for near infrared wave band is high and the road extraction precision on the image after disaster can not be guaranteed are solved.
And then, mapping a road range extracted from the optical remote sensing image before the disaster to the SAR remote sensing image after the disaster based on the prior condition of accurate registration of the optical remote sensing image before the disaster and the SAR remote sensing image after the disaster, thereby reducing the detection area from the whole image to the road area. And then, carrying out road disaster information detection on the post-disaster SAR remote sensing image.
The technical scheme of the invention is as follows:
a highway disaster information detection method based on high-resolution remote sensing images comprises the following steps:
step (1), acquiring images before and after disaster
And acquiring a high-resolution optical remote sensing image before the disaster occurs in the disaster area and a high-resolution SAR remote sensing image after the disaster occurs.
Step (2), road extraction is carried out on the optical remote sensing image before disaster
And performing multi-scale segmentation on the optical remote sensing image by using an object-oriented method, selecting an optimal segmentation scale, judging the homogeneity of the image object, and combining the homogeneous image objects to obtain a multi-scale segmentation result. And based on the result of multi-scale segmentation, combining machine learning and a threshold classification model, and extracting the road by using the RGB wave band information of the remote sensing image.
Step (3) registering the optical remote sensing image before disaster and the SAR remote sensing image after disaster to achieve prior condition of accurate registration
Because the radiation and geometric characteristics of the optical image and the SAR image are obviously different, and the same-name ground objects before and after a disaster are obviously changed or even disappear, the image registration difficulty is greatly increased. Therefore, in order to realize high-reliability registration of the optical image and the SAR image, a coarse-fine two-stage progressive registration method is provided, and the registration reliability of the optical/SAR image is improved by comprehensively using the point, line and surface characteristics of the remote sensing image.
Step (4), road range mapping
And (3) mapping the road range extracted in the step (2) to the post-disaster SAR remote sensing image based on the prior condition of the accurate registration in the step (3), and reducing the detection area from the whole image to the road area.
Based on the prior condition that the pre-disaster optical remote sensing image and the post-disaster SAR remote sensing image are accurately registered, the road range extracted from the pre-disaster optical remote sensing image is mapped to the post-disaster SAR remote sensing image, the detection area is reduced from the whole image to the road area, and the problem that the accuracy and the reliability of road damage detection using GIS data assistance in the conventional method are low is solved. Next, on the SAR remote sensing image after the disaster, a small base line set method is adopted to generate an average terrain deformation rate diagram and a deformation time sequence diagram in a road domain range, a road damage area with obvious deformation is extracted, further, the road damage degrees (serious, moderate and mild) with different grades are identified section by section, the damage length is counted, and the road damage degree is quantitatively evaluated in a research area by combining the road grade and the earthquake intensity, so that the detection of the road disaster damage information is realized.
Step (5) extracting road damage area
And generating an average terrain deformation rate diagram and a deformation time sequence diagram within a road domain range by adopting a small base line set method, extracting road damage areas with obvious deformation, further identifying road damage degrees of different grades section by section, counting damage lengths, calculating damage evaluation scores in a research area by combining road grades and earthquake intensity, and further quantitatively evaluating the road damage degrees, namely completing detection of road disaster damage information.
Further, in the step (2), the factors of spectrum homogeneity and shape homogeneity are selected as segmentation factors, and a multi-scale segmentation result is obtained by setting an optimal segmentation scale, wherein the segmentation scale comprises 20, 31 and 39.
Further, in the step (2), the specific steps of extracting the road by using the remote sensing image RGB band information in combination with the machine learning and threshold classification model are as follows:
firstly, a support vector machine supervision classification algorithm is adopted, different samples are separated by utilizing a classification hyperplane model, an optimal linear classification surface is obtained in a new space, and preliminary classification of roads and other ground objects is realized; then, on the basis of the primary classification result, classification features are constructed by using RGB wave band information of the remote sensing image, a road interpretation extraction rule set is established, road optimization extraction is realized, and finally, roads are output by using morphological optimization.
Further, in the step (2), the classification features include spectral features, texture features, geometric features and context features; the spectral feature construction mode is as follows:
BSCC＝[Brightness]+B-G*2 (1)
in the formula, BSCC is a built self-defined waveband spectral feature, and only the spectrum and brightness value combination of RGB waveband information is used; brightness is Brightness, B is blue band, G is green band;
the texture feature construction method comprises the following steps: only RGB wave band information is used, and the homogeneity and contrast of the gray level co-occurrence matrix are used as classification conditions;
the geometrical characteristics are constructed as follows: identifying by adopting the length-width ratio and the density attribute of the shape characteristic, and extracting and refining the road;
the context feature is constructed as follows: and optimizing the road classification result by establishing the position relation between the road object and the adjacent object and the proportion of the public edge of the adjacent object to the total edge length.
Further, in the step (3), the registration of the pre-disaster optical remote sensing image and the post-disaster SAR remote sensing image comprises the following steps:
step (3.1), preliminary registration
The optical image and the SAR image have different imaging mechanisms, the difference of radiation and geometric characteristics is obvious, but some obvious planar ground object characteristics can be kept stable under different imaging modes, so the method can be used for preliminary registration of the optical image and the SAR image. Firstly, planar ground feature characteristics are respectively extracted from the pre-disaster optical remote sensing image and the post-disaster SAR remote sensing image. And then, describing the planar ground object characteristics through a characteristic descriptor operator, and realizing characteristic matching through similarity matching of descriptors, thereby realizing the integral preliminary registration of the optical remote sensing image before disaster and the SAR image after disaster.
Step (3.2), accurate registration
When the preliminary registration is over, most of the geometric differences in the images are eliminated, and point and line features can be used for accurate registration. Firstly, extracting actual characteristic points and characteristic lines from the pre-disaster optical remote sensing image and the post-disaster SAR image which are integrally and preliminarily registered, and taking the intersection points of the characteristic lines as virtual characteristic points to enable the distribution of the characteristic points in the whole image to be more uniform. And then, describing the point features and the line features through a feature descriptor operator, and respectively realizing the pairing of the point features and the line features through the similarity matching of the descriptors, thereby accurately registering the two images.
Further, in the step (3.1), processing the optical remote sensing image before the disaster, extracting three primary visual features of color, direction and brightness of the optical remote sensing image by using an Itti visual attention model, forming three feature maps by a central peripheral difference mechanism, and combining the three feature maps to generate a saliency map;
processing the SAR remote sensing image after the disaster, merging the feature map into a saliency map by utilizing iterative standardization and utilizing three primary visual features of color, direction and texture extracted by a TW-Itti visual attention model, and merging the saliency map into a final saliency map of the SAR image;
extracting planar ground object characteristics from the pre-disaster optical remote sensing image saliency map and the post-disaster SAR remote sensing image saliency map respectively;
carrying out nonsubsampled Contourlet transformation on the extracted planar ground feature, carrying out multi-scale description on the transformed directional sub-band coefficient, and identifying and eliminating the surface feature belonging to the noise according to the distribution characteristic of the noise in the directional sub-band;
describing the face features after noise removal through the direction sub-band coefficients, and matching the face feature features by combining the inherent characteristics of the face region, thereby realizing the overall preliminary registration of the optical remote sensing image before disaster and the SAR image after disaster; wherein the inherent characteristics of the planar area comprise area, perimeter and primary-secondary axis ratio.
Further, in step (5), according to the road grade RdRoad damage length LdThe method comprises the following steps of sending geological disaster point scale M and seismic intensity I along a road, and establishing a damage assessment model based on a quantitative fuzzy comprehensive evaluation method as shown in the following formula:
R＝W×A (2)
in the formula: r is a damage assessment score; w is a weight vector calculated by a judgment matrix formed by each evaluation factor; a is represented by road grade RdRoad damage length LdThe membership matrix formed by the sent geological disaster point scale M and the seismic intensity I is as follows:
wherein: wherein the content of the first and second substances,
indicates the maximum value of the detected road grade,
a minimum value representing the detected road grade; a is11、a12、a13、a14Road grades R respectively representing first roadsdRoad damage length LdThe scale M of the sent geological disaster point and the seismic intensity I; a isn1、an2、an3、an4Road grades R respectively representing nth roaddRoad damage length LdThe scale M of the sent geological disaster point and the seismic intensity I; a is(n+m)1、a(n+m)2、a(n+m)3、a(n+m)4(ii) a Road grade R respectively representing the last roaddRoad damage length LdThe scale M of the sent geological disaster point and the seismic intensity I; and obtaining the road damage grade according to the damage evaluation score.
Compared with the prior art, the invention has the following beneficial effects:
(1) the invention provides a road damage information detection method only depending on RGB wave bands on an optical remote sensing image before a disaster by combining machine learning and a threshold classification model, so as to solve the problems that the existing method has high requirements on near infrared wave bands and the extraction precision of roads on an image after the earthquake cannot be guaranteed.
(2) The invention provides a two-stage progressive registration method for an optical/SAR remote sensing image, which solves the problems that the radiation and geometric characteristics of the optical image and the SAR image are obviously different, and the same-name ground objects are likely to obviously change or even disappear before and after a disaster occurs by fully utilizing the point, line and surface characteristics of the remote sensing image, so that the traditional image registration method based on single measure is increased in registration difficulty and insufficient in reliability. Then, based on the prior condition that the pre-disaster optical remote sensing image and the post-disaster SAR remote sensing image are accurately registered, the road range extracted from the pre-disaster optical remote sensing image is mapped to the post-disaster SAR remote sensing image, so that the detection area of the post-disaster SAR remote sensing image is reduced from the whole image to the road area.
After the detection of the road disaster damage information is completed, a foundation is laid for secondary disaster detection along the road in the subsequent disaster area according to the obtained SAR image amplitude information, and the integration of the road disaster information detection and the potential secondary disaster evaluation is laid.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the embodiments of the present invention, and it is obvious that the described embodiments are only a part of examples of the present invention, and not all examples. All other embodiments, which can be obtained by a person skilled in the art without inventive effort based on the embodiments of the present invention, are within the scope of the present invention.
The highway disaster information detection method based on the high-resolution remote sensing image, which is related by the embodiment, is used for analyzing 5 earthquake roads and comprises the following steps:
step (1), acquiring images before and after disaster
And acquiring a high-resolution optical remote sensing image before the earthquake of 5 roads in the affected area and a high-resolution SAR remote sensing image after the earthquake.
Step (2), road extraction is carried out on the optical remote sensing image before disaster
Firstly, the optical remote sensing image is segmented in a multi-scale mode by an object-oriented method. According to the imaging characteristics of the optical remote sensing image, selecting factors of spectrum homogeneity and shape homogeneity as segmentation factors, and obtaining a multi-scale segmentation result by setting an optimal segmentation scale. The segmentation scale may be selected to be 20, 31, 39, etc., depending on the requirements of the particular application on the segmentation accuracy. The software used in this embodiment is ecogonition.
And (3) performing road extraction through a model combining machine learning and threshold classification based on the result of multi-scale segmentation. Firstly, a support vector machine supervision classification algorithm is adopted, different samples are separated by utilizing a classification hyperplane model, an optimal linear classification surface is obtained in a new space, and preliminary classification of roads and other ground objects is realized. And then, on the basis of the primary classification result, spectrum features, texture features, geometric features and context features are constructed by using the RGB wave band information of the remote sensing image, and a road interpretation extraction rule set is established to realize road optimization extraction. And finally, optimizing the output road by using morphology. The software used in this embodiment is ecogonition.
The construction mode of each classification characteristic criterion is as follows:
spectral characteristics: only the spectrum and brightness value combined characteristics (BSCC) of the RGB wave band information are used, and roads can be distinguished from remote sensing images containing ground objects such as vegetation, water bodies and the like. Aiming at the plateau mountain road, the calculation formula of BSCC is as follows:
BSCC＝[Brightness]+B-G*2 (1)
in the formula, BSCC is the spectral characteristic of the constructed self-defined wave band, Brightness is the Brightness, B is the blue wave band, and G is the green wave band.
Texture characteristics: only the RGB band information is used, and the homogeneity and contrast of the gray level co-occurrence matrix are adopted as the classification conditions.
Geometric characteristics: and identifying by adopting the aspect ratio and the density attribute of the shape characteristic, and extracting and refining the road.
Context characteristics: and optimizing the road classification result by establishing the position relation between the road object and the adjacent object and the proportion of the public edge of the adjacent object to the total edge length.
Step (3) registering the optical remote sensing image before disaster and the SAR remote sensing image after disaster to achieve prior condition of accurate registration
The registration of the optical remote sensing image before disaster and the SAR remote sensing image after disaster is realized by comprehensively utilizing two-stage progressive registration of remote sensing image points, lines, surface characteristics and the like.
Step (3.1), preliminary registration
Processing the optical remote sensing image before disaster, extracting three primary visual features of color, direction and brightness of the optical remote sensing image by using an Itti visual attention model, forming three feature maps by a central peripheral difference mechanism, and combining the three feature maps to generate a saliency map.
And processing the SAR remote sensing image after the disaster, merging the feature map into a saliency map by utilizing iterative standardization and utilizing three primary visual features of color, direction and texture extracted by a TW-Itti visual attention model, and merging the saliency map into a final saliency map of the SAR image.
And extracting surface features from the pre-disaster optical remote sensing image saliency map and the post-disaster SAR remote sensing image saliency map respectively.
And performing nonsubsampled Contourlet transformation (NSCT transformation) on the extracted surface features, performing multi-scale description on the transformed directional subband coefficients, and identifying and eliminating the surface features belonging to the noise according to the distribution characteristics of the noise in the directional subbands. Next, the surface features after the noise is removed are described through a direction sub-band coefficient, and meanwhile, the surface features are matched according to the area, the perimeter and the inherent characteristics of the primary-secondary axis ratio contained in the surface area, so that the overall primary registration of the optical remote sensing image before disaster and the SAR image after disaster is realized.
The embodiment respectively extracts planar ground object characteristics from the pre-disaster optical remote sensing image and the post-disaster SAR remote sensing image; describing the planar ground feature characteristics through a characteristic descriptor operator, realizing characteristic pairing through similarity matching of the descriptors, and fitting an affine transformation model or a low-order polynomial transformation model according to the number of the characteristic pairing to realize initial registration.
Step (3.2), accurate registration:
extracting operators on the basis of phase consistency characteristic on the basis of the whole preliminarily registered pre-disaster optical remote sensing image and the post-disaster SAR image, extracting actual characteristic points and characteristic lines, and taking intersection points of the characteristic lines as virtual characteristic points to enable the distribution of the characteristic points in the whole image to be more uniform. And realizing final accurate registration between images by a combined spectrogram matching method.
Step (4), road range mapping
And (3) mapping the road range extracted from the pre-disaster optical remote sensing image in the step (2) onto the post-disaster SAR remote sensing image based on the prior condition that the pre-disaster optical remote sensing image and the post-disaster SAR remote sensing image in the step (3) are accurately registered, and reducing the detection area from the whole image to the road area.
Step (5) extracting road damage area
On the SAR remote sensing image after the disaster, a small base line set method is adopted to generate an average terrain deformation rate graph and a deformation time sequence graph in a road domain range, a road damage area with obvious deformation is extracted, and then the road damage area is classified according to the road grade (R)d) Road damage length (L)d) The method comprises the following steps of sending geological disaster point scale (M) and seismic intensity (I) along a road, establishing a damage assessment model based on a quantitative fuzzy comprehensive evaluation method, and obtaining damage grade as shown in the following formula:
R＝W×A (2)
in the formula: r is a damage assessment score; w is the weight calculated by the judgment matrix formed by each evaluation factorVector quantity; a is the road grade (R)d) Road damage length (L)d) And the scale (M) of the sent geological disaster point and the seismic intensity (I) form a membership matrix.
The evaluation results of 5 link actual measurement values are shown in table 1:
table 1 evaluation of measured values of links
Based on the road section measured values in table 1, the evaluation factor membership matrix a of the present embodiment is as follows:
wherein the content of the first and second substances,
indicates the maximum value of the detected road grade,
represents the minimum value of the detected road grade.
Finally, the damage assessment score and the corresponding grade of the 5-segment road are calculated and shown in table 2:
TABLE 2 road segment damage rating
The higher the damage assessment score, the more severe the damage.
Wherein, the damage grades corresponding to the damage assessment scores are shown in table 3:
TABLE 3 road segment damage level
| Score range | Grade of damage |
| [0,0.25) | Light and slight |
| [0.25,0.4) | Mild degree of |
| [0.4,0.6) | Of moderate degree |
| [0.6,0.8) | Severe severity of disease |
| [0.8,1) | Severe and severe |
After the detection of the road disaster damage information is completed, based on the post-disaster SAR remote sensing image, the deformation, scale and distance from each disaster hidden danger point to a target road can be obtained according to the existing method, then the risk degree of the potential hidden danger point is determined by combining the seismic intensity and road grade and utilizing the quantitative fuzzy comprehensive evaluation method, and the evaluation of the secondary disasters along the road is realized.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (7)
1. A highway disaster information detection method based on high-resolution remote sensing images is characterized by comprising the following steps: the method comprises the following steps:
step (1), acquiring images before and after disaster
Acquiring a high-resolution optical remote sensing image before a disaster occurs in a disaster area and a high-resolution SAR remote sensing image after the disaster occurs;
step (2), road extraction is carried out on the optical remote sensing image before disaster
Carrying out multi-scale segmentation on the optical remote sensing image by adopting an object-oriented method, selecting an optimal segmentation scale, and combining homogeneous image objects by judging the homogeneity of the image objects to obtain a multi-scale segmentation result; based on the result of multi-scale segmentation, combining machine learning and a threshold classification model, and extracting roads by using remote sensing image RGB wave band information;
step (3) registering the optical remote sensing image before disaster and the SAR remote sensing image after disaster to achieve prior condition of accurate registration
Step (4), road range mapping
Based on the prior condition of the accurate registration in the step (3), mapping the road range extracted in the step (2) to the post-disaster SAR remote sensing image, and reducing the detection area from the whole image to the road area;
step (5) extracting road damage area
And generating an average terrain deformation rate diagram and a deformation time sequence diagram within a road domain range by adopting a small base line set method, extracting road damage areas with obvious deformation, further identifying road damage degrees of different grades section by section, counting damage lengths, calculating damage evaluation scores in a research area by combining road grades and earthquake intensity, and further quantitatively evaluating the road damage degrees, namely completing detection of road disaster damage information.
2. The road disaster information detection method based on the high-resolution remote sensing image according to claim 1, characterized in that: in the step (2), the factors of spectrum homogeneity and shape homogeneity are selected as segmentation factors, and a multi-scale segmentation result is obtained by setting an optimal segmentation scale, wherein the segmentation scale comprises 20, 31 and 39.
3. The road disaster information detection method based on the high-resolution remote sensing image according to claim 1, characterized in that: in the step (2), the specific steps of extracting the road by using the remote sensing image RGB waveband information in combination with the machine learning and threshold classification model are as follows:
firstly, a support vector machine supervision classification algorithm is adopted, different samples are separated by utilizing a classification hyperplane model, an optimal linear classification surface is obtained in a new space, and preliminary classification of roads and other ground objects is realized; then, on the basis of the primary classification result, classification features are constructed by using RGB wave band information of the remote sensing image, a road interpretation extraction rule set is established, road optimization extraction is realized, and finally, roads are output by using morphological optimization.
4. The road disaster information detection method based on the high-resolution remote sensing image according to claim 3, characterized in that: in the step (2), the classification features comprise spectral features, texture features, geometric features and context features; the spectral feature construction mode is as follows:
BSCC＝[Brightness]+B-G*2 (1)
in the formula, BSCC is a built self-defined waveband spectral feature, and only the spectrum and brightness value combination of RGB waveband information is used; brightness is Brightness, B is blue band, G is green band;
the texture feature construction method comprises the following steps: only RGB wave band information is used, and the homogeneity and contrast of the gray level co-occurrence matrix are used as classification conditions;
the geometrical characteristics are constructed as follows: identifying by adopting the length-width ratio and the density attribute of the shape characteristic, and extracting and refining the road;
the context feature is constructed as follows: and optimizing the road classification result by establishing the position relation between the road object and the adjacent object and the proportion of the public edge of the adjacent object to the total edge length.
5. The road disaster information detection method based on the high-resolution remote sensing image according to claim 1, characterized in that: in the step (3), the registration of the optical remote sensing image before disaster and the SAR remote sensing image after disaster comprises the following steps:
step (3.1), preliminary registration
Respectively extracting planar ground object characteristics from the pre-disaster optical remote sensing image and the post-disaster SAR remote sensing image; describing the planar ground object characteristics through a characteristic descriptor operator, and realizing characteristic matching through similarity matching of descriptors;
step (3.2), accurate registration
Extracting actual characteristic points and characteristic lines from the preliminarily registered pre-disaster optical remote sensing image and the post-disaster SAR remote sensing image, and taking the intersection points of the characteristic lines as virtual characteristic points to enable the distribution of the characteristic points in the whole image to be more uniform; and then, describing the point features and the line features through a feature descriptor operator, and respectively realizing the pairing of the point features and the line features through the similarity matching of the descriptors.
6. The road disaster information detection method based on the high-resolution remote sensing image according to claim 5, characterized in that: in the step (3.1), processing the optical remote sensing image before disaster, extracting three primary visual features of color, direction and brightness of the optical remote sensing image by using an Itti visual attention model, forming three feature maps by a central peripheral difference mechanism, and combining the three feature maps to generate a saliency map;
processing the SAR remote sensing image after the disaster, merging the feature map into a saliency map by utilizing iterative standardization and utilizing three primary visual features of color, direction and texture extracted by a TW-Itti visual attention model, and merging the saliency map into a final saliency map of the SAR image;
extracting planar ground object characteristics from the pre-disaster optical remote sensing image saliency map and the post-disaster SAR remote sensing image saliency map respectively;
carrying out nonsubsampled Contourlet transformation on the extracted planar ground feature, carrying out multi-scale description on the transformed directional sub-band coefficient, and identifying and eliminating the surface feature belonging to the noise according to the distribution characteristic of the noise in the directional sub-band;
describing the face features after noise removal through the direction sub-band coefficients, and matching the face feature features by combining the inherent characteristics of the face region, thereby realizing the overall preliminary registration of the optical remote sensing image before disaster and the SAR image after disaster; wherein the inherent characteristics of the planar area comprise area, perimeter and primary-secondary axis ratio.
7. The road disaster information detection method based on the high-resolution remote sensing image according to claim 1, characterized in that: in the step (5), according to the road grade RdRoad damage length LdThe method comprises the following steps of sending geological disaster point scale M and seismic intensity I along a road, and establishing a damage assessment model based on a quantitative fuzzy comprehensive evaluation method as shown in the following formula:
R＝W×A (2)
in the formula: r is a damage assessment score; w is a weight vector calculated by a judgment matrix formed by each evaluation factor; a is represented by road grade RdRoad damage length LdThe membership matrix formed by the sent geological disaster point scale M and the seismic intensity I is as follows:
wherein: wherein the content of the first and second substances,
indicates the maximum value of the detected road grade,
a minimum value representing the detected road grade; a is11、a12、a13、a14Road grades R respectively representing first roadsdRoad damage length LdThe scale M of the sent geological disaster point and the seismic intensity I; a isn1、an2、an3、an4Road grades R respectively representing nth roaddRoad damage length LdThe scale M of the sent geological disaster point and the seismic intensity I; a is(n+m)1、a(n+m)2、a(n+m)3、a(n+m)4(ii) a Road grade R respectively representing the last roaddRoad damage length LdThe scale M of the sent geological disaster point and the seismic intensity I; and obtaining the road damage grade according to the damage evaluation score.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911340911.8A CN111160199B (en) | 2019-12-23 | 2019-12-23 | Highway disaster information detection method based on high-resolution remote sensing image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911340911.8A CN111160199B (en) | 2019-12-23 | 2019-12-23 | Highway disaster information detection method based on high-resolution remote sensing image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111160199A true CN111160199A (en) | 2020-05-15 |
| CN111160199B CN111160199B (en) | 2022-09-13 |
Family
ID=70558023
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911340911.8A Active CN111160199B (en) | 2019-12-23 | 2019-12-23 | Highway disaster information detection method based on high-resolution remote sensing image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111160199B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111783686A (en) * | 2020-07-03 | 2020-10-16 | 中国交通通信信息中心 | Asphalt pavement health state monitoring system and method |
| CN111783700A (en) * | 2020-07-06 | 2020-10-16 | 中国交通通信信息中心 | Automatic recognition early warning method and system for road foreign matters |
| CN111985355A (en) * | 2020-08-01 | 2020-11-24 | 桂林理工大学 | Remote sensing building earthquake damage assessment method and system based on deep learning and cloud computing |
| CN112990100A (en) * | 2021-04-14 | 2021-06-18 | 中国科学院空天信息创新研究院 | Road disaster remote sensing intelligent detection method based on deep learning |
| CN113971505A (en) * | 2021-09-16 | 2022-01-25 | 杜敏齐 | Railway train emergency scheduling method, device, equipment and readable storage medium |
| CN114049568A (en) * | 2021-11-29 | 2022-02-15 | 中国平安财产保险股份有限公司 | Object shape change detection method, device, equipment and medium based on image comparison |
| CN114581771A (en) * | 2022-02-23 | 2022-06-03 | 南京信息工程大学 | High-resolution heterogeneous source remote sensing detection method for collapsed building |
| CN115620149A (en) * | 2022-12-05 | 2023-01-17 | 耕宇牧星(北京)空间科技有限公司 | Road detection method based on remote sensing image |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101614822A (en) * | 2009-07-17 | 2009-12-30 | 北京大学 | Detect the method for road damage based on post-disaster high-resolution remote sensing image |
| FR2934501A1 (en) * | 2008-08-04 | 2010-02-05 | Smart Packaging Solutions Sps | FIRE RISK PREVENTION SYSTEM |
| CN109544579A (en) * | 2018-11-01 | 2019-03-29 | 上海理工大学 | A method of damage building is assessed after carrying out calamity using unmanned plane |
| CN110287932A (en) * | 2019-07-02 | 2019-09-27 | 中国科学院遥感与数字地球研究所 | Route denial information extraction based on the segmentation of deep learning image, semantic |
-
2019
- 2019-12-23 CN CN201911340911.8A patent/CN111160199B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| FR2934501A1 (en) * | 2008-08-04 | 2010-02-05 | Smart Packaging Solutions Sps | FIRE RISK PREVENTION SYSTEM |
| CN101614822A (en) * | 2009-07-17 | 2009-12-30 | 北京大学 | Detect the method for road damage based on post-disaster high-resolution remote sensing image |
| CN109544579A (en) * | 2018-11-01 | 2019-03-29 | 上海理工大学 | A method of damage building is assessed after carrying out calamity using unmanned plane |
| CN110287932A (en) * | 2019-07-02 | 2019-09-27 | 中国科学院遥感与数字地球研究所 | Route denial information extraction based on the segmentation of deep learning image, semantic |
Non-Patent Citations (1)
| Title |
|---|
| 常泽岫: "基于单时相高分辨率遥感影像的震后道路提取研究", 《测绘与空间地理信息》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111783686A (en) * | 2020-07-03 | 2020-10-16 | 中国交通通信信息中心 | Asphalt pavement health state monitoring system and method |
| CN111783700A (en) * | 2020-07-06 | 2020-10-16 | 中国交通通信信息中心 | Automatic recognition early warning method and system for road foreign matters |
| CN111783700B (en) * | 2020-07-06 | 2023-11-24 | 中国交通通信信息中心 | Automatic recognition and early warning method and system for pavement foreign matters |
| CN111985355A (en) * | 2020-08-01 | 2020-11-24 | 桂林理工大学 | Remote sensing building earthquake damage assessment method and system based on deep learning and cloud computing |
| CN111985355B (en) * | 2020-08-01 | 2022-09-27 | 桂林理工大学 | Remote sensing building earthquake damage assessment method and system based on deep learning and cloud computing |
| CN112990100A (en) * | 2021-04-14 | 2021-06-18 | 中国科学院空天信息创新研究院 | Road disaster remote sensing intelligent detection method based on deep learning |
| CN113971505A (en) * | 2021-09-16 | 2022-01-25 | 杜敏齐 | Railway train emergency scheduling method, device, equipment and readable storage medium |
| CN113971505B (en) * | 2021-09-16 | 2023-10-27 | 杜敏齐 | Railway train emergency dispatching method, device, equipment and readable storage medium |
| CN114049568A (en) * | 2021-11-29 | 2022-02-15 | 中国平安财产保险股份有限公司 | Object shape change detection method, device, equipment and medium based on image comparison |
| CN114581771A (en) * | 2022-02-23 | 2022-06-03 | 南京信息工程大学 | High-resolution heterogeneous source remote sensing detection method for collapsed building |
| CN115620149A (en) * | 2022-12-05 | 2023-01-17 | 耕宇牧星(北京)空间科技有限公司 | Road detection method based on remote sensing image |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111160199B (en) | 2022-09-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111160199B (en) | Highway disaster information detection method based on high-resolution remote sensing image | |
| CN109919875B (en) | High-time-frequency remote sensing image feature-assisted residential area extraction and classification method | |
| US8913826B2 (en) | Advanced cloud cover assessment for panchromatic images | |
| Bouziani et al. | Rule-based classification of a very high resolution image in an urban environment using multispectral segmentation guided by cartographic data | |
| CN109447160B (en) | Method for automatically matching image and vector road intersection | |
| CN103235952B (en) | A kind of measure of the Lv Du space, city based on high-resolution remote sensing image | |
| CN105930772A (en) | City impervious surface extraction method based on fusion of SAR image and optical remote sensing image | |
| CN111191628B (en) | Remote sensing image earthquake damage building identification method based on decision tree and feature optimization | |
| CN110390255A (en) | High-speed rail environmental change monitoring method based on various dimensions feature extraction | |
| CN111339827A (en) | SAR image change detection method based on multi-region convolutional neural network | |
| CN111047695A (en) | Method for extracting height spatial information and contour line of urban group | |
| CN111598048A (en) | Urban village-in-village identification method integrating high-resolution remote sensing image and street view image | |
| Chen et al. | Detecting changes in high-resolution satellite coastal imagery using an image object detection approach | |
| US11941878B2 (en) | Automated computer system and method of road network extraction from remote sensing images using vehicle motion detection to seed spectral classification | |
| CN112669363B (en) | Method for measuring three-dimensional green space of urban green space | |
| CN107992856A (en) | High score remote sensing building effects detection method under City scenarios | |
| CN110889840A (en) | Effectiveness detection method of high-resolution 6 # remote sensing satellite data for ground object target | |
| CN113822141A (en) | Automatic glacier and snow extraction method and system based on remote sensing image | |
| Lefebvre et al. | Monitoring the morphological transformation of Beijing old city using remote sensing texture analysis | |
| CN115205528A (en) | Feature selection method for geographic object-oriented image analysis | |
| CN112347926B (en) | High-resolution image city village detection method based on building morphology distribution | |
| CN107657246B (en) | Remote sensing image building detection method based on multi-scale filtering building index | |
| CN112597936A (en) | Building rubbish separation method based on object-oriented hierarchical segmentation and related products | |
| Engstrom et al. | Evaluating the Relationship between Contextual Features Derived from Very High Spatial Resolution Imagery and Urban Attributes: A Case Study in Sri Lanka | |
| CN110929739A (en) | Automatic impervious surface range remote sensing iterative extraction method |
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 |