A multi-source information-based remote sensing image intelligent segmentation management system and method

Abstract

The application discloses a kind of based on multi-source information's remote sensing image intelligent segmentation management system and method, the present system includes preprocessing module, positioning module, merging module and segmentation module;The preprocessing module is used to carry out pre-segmentation to given image, obtains multiple sub-regions;The positioning module is used to determine the linking area of each sub-region;The merging module is used to merge according to the same feature respectively to the multiple sub-regions, obtains the sub-segmentation result corresponding to each feature;The segmentation module is used to carry out clustering integration to the sub-segmentation result of the same feature, to obtain the segmentation result of given image.Meanwhile still provide a kind of based on multi-source information's remote sensing image intelligent segmentation management method, can better apply image segmentation technology in remote sensing image ground feature information automatic extraction.

Description

Technical Field

[0001] This invention relates to the field of intelligent segmentation of remote sensing images, specifically to an intelligent segmentation management system and method for remote sensing images based on multi-source information. Background Technology

[0002] Remote sensing image segmentation is a crucial technology, forming the foundation of digital image processing and holding significant importance in image engineering. In scientific research and practical applications of image engineering, researchers are generally only interested in localized content within an image; these localized contents are called the foreground, and they are areas with specific properties and characteristics. The content outside the foreground is called the background. Sometimes, for research and application purposes, the foreground needs to be extracted from the entire image to prepare for further work. Image segmentation technology is the process of processing an image based on certain features to separate the desired foreground from the background of the entire image. Image segmentation technology is now applied in numerous industries, and its application scope is constantly expanding, including medical image analysis, image recognition, remote sensing image analysis, military, and agriculture. Remote sensing image segmentation involves dividing a remote sensing image into smaller intervals based on certain features and then separating the specific background as needed. In land use remote sensing monitoring, it is necessary to extract information on land features and land types, which currently relies mainly on human interpretation. However, remote sensing image data is massive, making manual interpretation time-consuming and labor-intensive. Therefore, finding better ways to apply image segmentation technology to the automatic extraction of ground feature information from remote sensing images is crucial and urgent. Scholars both domestically and internationally have conducted extensive research in this area, proposing numerous remote sensing image segmentation algorithms. However, due to the inherent characteristics of remote sensing images, such as their large information content, complex image content, and unclear boundaries, the versatility of these algorithms is poor. There is no reliable model to fully represent and guide the entire process, thus hindering the practical application of segmentation technology in the field of remote sensing image segmentation.

[0003] In conclusion, deepening the research on image segmentation technology so that it can be better applied to the extraction of ground feature information from remote sensing images has significant practical implications for the future development of image segmentation technology and the understanding and analysis of remote sensing images.

[0004] Therefore, a remote sensing image intelligent segmentation management system and method based on multi-source information is needed to solve the above problems. Summary of the Invention

[0005] The purpose of this invention is to provide a remote sensing image intelligent segmentation management system and method based on multi-source information, so as to solve the problems mentioned in the background art.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] A remote sensing image intelligent segmentation management system based on multi-source information, the system includes: a preprocessing module, a positioning module, a merging module and a segmentation module;

[0008] The preprocessing module is used to pre-segment a given image to obtain multiple sub-regions;

[0009] The positioning module is used to determine the link area of ​​each sub-region;

[0010] The merging module is used to merge the multiple sub-regions according to the same features to obtain the sub-segmentation result corresponding to each feature;

[0011] The segmentation module is used to cluster and integrate sub-segmentation results with the same features to obtain the segmentation result of a given image;

[0012] The output of the preprocessing module is connected to the input of the positioning module; the output of the positioning module is connected to the input of the merging module; and the output of the merging module is connected to the input of the segmentation module.

[0013] According to the above technical solution, the positioning module includes an extraction unit, a first calculation unit, and a determination unit;

[0014] The extraction unit is used to extract feature values ​​of a sub-region and its adjacent sub-regions. The extraction of these feature values ​​is an existing technology. Image features can be selected according to requirements and are generally divided into point, line, and surface features, texture features, color features, and statistical features of the image.

[0015] The first calculation unit is used to calculate the feature similarity between the sub-region and its neighboring sub-regions based on the extracted feature values;

[0016] The determining unit is used to identify adjacent sub-regions whose feature similarity is greater than the similarity feature value and whose region area is greater than the sub-region as the linking regions of the sub-region.

[0017] According to the above technical solution, the merging module includes a computing unit and a merging unit;

[0018] The calculation unit is used to determine the merging weight of the linked regions of each sub-region based on the feature similarity and semantic similarity evaluation function between each sub-region and its linked regions;

[0019] The merging unit is used to merge each sub-region with the linked region that has the largest merging weight.

[0020] According to the above technical solution, the segmentation module includes a processing unit, a setting unit, and a segmentation subunit;

[0021] The processing unit is used to calculate the feature similarity of the sub-segmentation results corresponding to each feature, and to normalize the feature similarity of the sub-segmentation results with the same feature;

[0022] The setting unit is used to take the number of clusters of the sub-segmentation results corresponding to the minimum value of the normalized feature similarity as the final number of clusters.

[0023] The segmentation subunit is used to perform spectral clustering integration of the sub-segmentation results with the same features according to the final number of clusters, so as to obtain the segmentation result of the given image.

[0024] A method for intelligent segmentation and management of remote sensing images based on multi-source information, comprising the following steps:

[0025] S1. Pre-segment the given image to obtain multiple sub-regions;

[0026] S2. Determine the link areas for each sub-region;

[0027] S3. Merge the multiple sub-regions according to the same features to obtain the sub-segmentation result corresponding to each feature;

[0028] S4. Cluster and integrate the sub-segmentation results with the same features to obtain the segmentation result of the given image.

[0029] According to the above technical solution, in S2, for the linked regions of the sub-regions, the feature similarity extracted from the feature values ​​of the sub-regions is used to determine the merging weight of the linked regions of each sub-region, and the feature value set of each sub-region is extracted by the extraction unit as U = {U1, U2, ..., U...} n}, where, for the required sub-region p i The set of features belonging to the same region is a subset of the set of feature values. The feature similarity between sub-regions is calculated according to the following formula:

[0030]

[0031] Where pi and pj are subsets of the feature value sets of two different sub-regions, and the similarity feature values ​​can be set separately for different features according to requirements, σ(P i P j ) is a subregion p i and p j Feature similarity, sim(p i ,p jLet be the feature similarity between two linked regions pi and pj, and let E be a similarity evaluation function. Compare the feature similarity between the two linked regions pi and pj: if the similarity functions are equal, it means that the two sub-regions belong to the same feature region; if the similarity functions are not equal, it means that the two sub-regions do not belong to the same feature region.

[0032] According to the above technical solution, in S2, the determining unit uses the adjacent sub-regions whose feature similarity is greater than the similarity feature value and whose area is greater than that of the sub-region as the linking region of the sub-region. When the feature similarity between a sub-region and its adjacent sub-regions is greater than the set similarity feature value, it is considered that there is a linking relationship between the sub-region and its adjacent sub-regions. The area of ​​the two sub-regions is compared, and the adjacent sub-region with an area greater than that of the sub-region is identified as the linking region of the sub-region. The linking direction of the sub-region points to its linking region. The similarity feature value can be set separately for different features according to requirements. For example, the similarity feature value corresponding to the color feature can be set to the average feature similarity of HSV colors of all sub-regions; the similarity feature value corresponding to the texture feature can be set to the average feature similarity of the texture co-occurrence matrix of all sub-regions; and the similarity feature value corresponding to the SIFT feature can be set to the average SIFT feature similarity of all sub-regions.

[0033] According to the above technical solution, in S3, the merging module is used to merge the multiple sub-regions with the same features. Merging sub-regions of the image can be seen as a process of jumping from one node to another according to the similarity probability of the linked nodes, thus obtaining the merging weight P of each sub-region. R ,Right now

[0034]

[0035] Wherein Pr(p j ) represents region P j The merging weights, ε is an adjustment constant, and n is the number of sub-regions obtained after pre-segmentation of the image, for example, set to 50 in the experiment, N(p j ) is a function that calculates the subregion p. j All linked regions, sub-regions p j Each link region is represented by p k This means that after obtaining the merging weight Pr of the linked regions of each sub-region based on the above formula, each sub-region is merged with the linked regions whose merging weights meet the conditions.

[0036] According to the above technical solution, in S4, the processing unit performs normalization processing on sub-segments with the same feature similarity, and defines the saliency value of a pixel by using the contrast between the color of a pixel and the colors of pixels in other sub-segments within the sub-segment. Pixel I in sub-segment I... k The significance of is defined as:

[0037]

[0038] Among them, C l For pixel I k The color value, n is the total number of colors in the image, f j The number of clusters in which Cj appears in image I is used to perform spectral clustering integration of the sub-segmentation results with the same features according to the final number of clusters, thereby obtaining the segmentation result of the given image.

[0039] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0040] This invention pre-segments a given image using a preprocessing module to obtain multiple sub-regions. A feature value extraction module, which identifies the link regions of each sub-region and its neighboring sub-regions, is then used to locate these regions. Feature value extraction is a prior art technique; image features can be selected according to requirements, generally including point, line, and surface features, texture features, color features, and statistical features. A merging module merges the multiple sub-regions according to the same features, obtaining sub-segmentation results corresponding to each feature. A segmentation module then clusters and integrates the sub-segmentation results with the same features to obtain the segmentation result of the given image. This allows remote sensing images to be clustered and clearly labeled according to individual needs during segmentation, significantly reducing human resources. Attached Figure Description

[0041] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0042] Figure 1 This is a structural diagram of a remote sensing image intelligent segmentation management system based on multi-source information according to the present invention;

[0043] Figure 2 This is a flowchart of a remote sensing image intelligent segmentation and management method based on multi-source information according to the present invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] Please see Figures 1-2 The present invention provides a technical solution: a remote sensing image intelligent segmentation management system based on multi-source information, the system comprising: a preprocessing module, a positioning module, a merging module and a segmentation module;

[0046] The preprocessing module is used to pre-segment a given image to obtain multiple sub-regions;

[0047] The positioning module is used to determine the link area of ​​each sub-region;

[0048] The merging module is used to merge the multiple sub-regions according to the same features to obtain the sub-segmentation result corresponding to each feature;

[0049] The segmentation module is used to cluster and integrate sub-segmentation results with the same features to obtain the segmentation result of a given image;

[0050] The output of the preprocessing module is connected to the input of the positioning module; the output of the positioning module is connected to the input of the merging module; and the output of the merging module is connected to the input of the segmentation module.

[0051] According to the above technical solution, the positioning module includes an extraction unit, a first calculation unit, and a determination unit;

[0052] The extraction unit is used to extract feature values ​​of a sub-region and its adjacent sub-regions. The extraction of these feature values ​​is an existing technology. Image features can be selected according to requirements and are generally divided into point, line, and surface features, texture features, color features, and statistical features of the image.

[0053] The first calculation unit is used to calculate the feature similarity between the sub-region and its neighboring sub-regions based on the extracted feature values;

[0054] The determining unit is used to identify adjacent sub-regions whose feature similarity is greater than the similarity feature value and whose region area is greater than the sub-region as the linking regions of the sub-region.

[0055] According to the above technical solution, the merging module includes a computing unit and a merging unit;

[0056] The calculation unit is used to determine the merging weight of the linked regions of each sub-region based on the feature similarity and semantic similarity evaluation function between each sub-region and its linked regions;

[0057] The merging unit is used to merge each sub-region with the linked region that has the largest merging weight.

[0058] According to the above technical solution, the segmentation module includes a processing unit, a setting unit, and a segmentation subunit;

[0059] The processing unit is used to calculate the feature similarity of the sub-segmentation results corresponding to each feature, and to normalize the feature similarity of the final cluster number sub-segmentation results with the same feature.

[0060] The setting unit is used to take the number of clusters of the sub-segmentation results corresponding to the minimum value of the normalized feature similarity as the number of clusters;

[0061] The segmentation subunit is used to perform spectral clustering integration of the sub-segmentation results with the same features according to the final cluster number to obtain the segmentation result of the given image.

[0062] A method for intelligent segmentation and management of remote sensing images based on multi-source information, comprising the following steps:

[0063] S1. Pre-segment the given image to obtain multiple sub-regions;

[0064] S2. Determine the link areas for each sub-region;

[0065] S3. Merge the multiple sub-regions according to the same features to obtain the sub-segmentation result corresponding to each feature;

[0066] S4. Cluster and integrate the sub-segmentation results with the same features to obtain the segmentation result of the given image.

[0067] In S2, for the linked regions of the sub-regions, the feature similarity extracted based on the feature values ​​of the sub-regions is used to determine the merging weight of the linked regions of each sub-region. The feature value set of each sub-region is extracted using the extraction unit as U = {U1, U2, ..., U...} n}, where, for the required sub-region p i The set of features belonging to the same region is a subset of the set of feature values. The feature similarity between sub-regions is calculated according to the following formula:

[0068]

[0069] Where pi and pj are subsets of the feature sets of two different sub-regions, σ(P i Pj ) is a subregion p i and p j Feature similarity, sim(p i ,p j Let be the feature similarity between two linked regions pi and pj, and let E be a similarity evaluation function. Compare the feature similarity between the two linked regions pi and pj: if the similarity functions are equal, it means that the two sub-regions belong to the same feature region; if the similarity functions are not equal, it means that the two sub-regions do not belong to the same feature region.

[0070] In S2, using the determining unit, adjacent sub-regions with feature similarity greater than the similarity feature value and area greater than the sub-region are identified as linked regions of the sub-region. When the feature similarity between a sub-region and its adjacent sub-regions is greater than the set similarity feature value, it is considered that there is a link relationship between the sub-region and its adjacent sub-regions. The area of ​​the two sub-regions is compared, and the adjacent sub-region with an area greater than that of the sub-region is identified as the linked region of the sub-region. The link direction of the sub-region points to its linked region. The similarity feature value can be set separately for different features according to requirements. For example, the similarity feature value corresponding to the color feature can be set to the average feature similarity of HSV colors of all sub-regions; the similarity feature value corresponding to the texture feature can be set to the average feature similarity of the texture co-occurrence matrix of all sub-regions; and the similarity feature value corresponding to the SIFT feature can be set to the average SIFT feature similarity of all sub-regions.

[0071] In S3, the merging module is used to merge the multiple sub-regions with the same features. Merging sub-regions of an image can be viewed as a process of jumping from one node to another according to the similarity probability of linked nodes, thus obtaining the merging weight P for each sub-region. R ,Right now

[0072]

[0073] Wherein Pr(p j ) represents region P j The merging weights, ε is an adjustment constant, and n is the number of sub-regions obtained after pre-segmentation of the image, for example, set to 50 in the experiment, N(p j ) is a function that calculates the subregion p. j All linked regions, sub-regions p j Each link region is represented by p k This means that after obtaining the merging weight Pr of the linked regions of each sub-region based on the above formula, each sub-region is merged with the linked regions whose merging weights meet the conditions.

[0074] In S4, the processing unit normalizes sub-segments with the same feature similarity, defining the saliency value of a pixel by the contrast between its color and the colors of pixels in other sub-segments. Pixel I in sub-segment I... k The significance of is defined as:

[0075]

[0076] Among them, C l For pixel I k The color value, n is the total number of colors in the image, f j Given the number of clusters in image I where Cj appears, the sub-segmentation results with the same features are spectral clustered and integrated according to the final number of clusters to obtain the segmentation result of the given image.

[0077] Example 1:

[0078] The given image is pre-segmented to obtain multiple sub-regions. The merging weight of the linking regions of each sub-region is determined based on the feature similarity extracted from the feature values ​​of the sub-regions. The feature value set of each sub-region is extracted using the extraction unit as U = {U1, U2, ..., U...}. n}, where, for the required sub-region p i The set of features belonging to the same region is a subset of the set of feature values. The feature similarity between sub-regions is calculated according to the following formula:

[0079]

[0080] Where pi and pj are two distinct sub-regions, σ(P i P j ) is a subregion p i and p j Feature similarity, sim(p i ,p j Let ) represent the feature similarity between two linked regions pi and pj, and E be a similarity evaluation function. Comparing the feature similarity of pi and pj, if the similarity functions are equal, it indicates that the two sub-regions belong to the same feature region. Adjacent sub-regions with a feature similarity greater than the similarity feature value and a region area greater than the sub-region are considered as linked regions of that sub-region. The similarity feature value can be set according to different features as needed. For example, if the requirement is farmland, the similarity value of farmland is used for searching. Normalization is performed on sub-segments with the same feature similarity. The saliency value of a pixel is defined by the contrast between the color of a pixel and the colors of pixels in other sub-segments. Finally, spectral clustering is performed on the final number of clusters to obtain the segmentation result of the given image.

[0081] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0082] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A remote sensing image intelligent segmentation management system based on multi-source information, characterized in that: The system includes: a preprocessing module, a positioning module, a merging module, and a segmentation module; The preprocessing module is used to pre-segment a given image to obtain multiple sub-regions; The positioning module is used to determine the link area of ​​each sub-region; The merging module is used to merge the multiple sub-regions according to the same features to obtain the sub-segmentation result corresponding to each feature; The segmentation module is used to cluster and integrate sub-segmentation results with the same features to obtain the segmentation result of a given image; The output of the preprocessing module is connected to the input of the positioning module; the output of the positioning module is connected to the input of the merging module; and the output of the merging module is connected to the input of the segmentation module. The positioning module includes an extraction unit, a first calculation unit, and a determination unit; The extraction unit is used to extract the feature values ​​of the sub-region and its adjacent sub-regions; The first calculation unit is used to calculate the feature similarity between the sub-region and its neighboring sub-regions based on the extracted feature values; The determining unit is used to identify adjacent sub-regions with a feature similarity greater than the similarity feature value and a region area greater than the sub-region as the linking regions of the sub-region; The merging module includes a calculation unit and a merging unit; The calculation unit is used to determine the merging weight of the linked regions of each sub-region based on the feature similarity and semantic similarity evaluation function between each sub-region and its linked regions; The merging unit is used to merge each sub-region with the linked region that has the largest merging weight; The merging module is used to merge the multiple sub-regions with the same features. Merging sub-regions of an image can be viewed as a process of jumping from one node to another according to the similarity probability of the linked nodes, thus obtaining the merging weight P for each sub-region. R ,Right now Wherein Pr(p j ) represents region P j The merging weights, ε is an adjustment constant, n is the number of sub-regions obtained after pre-segmentation of the image, and N(p j ) is a function that calculates the subregion p. j All linked regions, sub-regions p j Each link region is represented by p k This means that after obtaining the merging weight Pr of the linked regions of each sub-region based on the above formula, each sub-region is merged with the linked regions whose merging weights meet the conditions.

2. The remote sensing image intelligent segmentation management system based on multi-source information according to claim 1, characterized in that: The segmentation module includes a processing unit, a setting unit, and a segmentation subunit; The processing unit is used to calculate the feature similarity of the sub-segmentation results corresponding to each feature, and to normalize the feature similarity of the sub-segmentation results with the same feature; The setting unit is used to take the number of clusters of the sub-segmentation results corresponding to the minimum value of the normalized feature similarity as the final number of clusters. The segmentation subunit is used to perform spectral clustering integration of the sub-segmentation results with the same features according to the final number of clusters, so as to obtain the segmentation result of the given image.

3. A remote sensing image intelligent segmentation management method based on multi-source information, used in the remote sensing image intelligent segmentation management system based on multi-source information as described in any one of claims 1-2, characterized in that: The method includes the following steps: S1. Pre-segment the given image to obtain multiple sub-regions; S2. Determine the link areas for each sub-region; S3. Merge the multiple sub-regions according to the same features to obtain the sub-segmentation result corresponding to each feature; S4. Cluster and integrate the sub-segmentation results with the same features to obtain the segmentation result of the given image.

4. The intelligent segmentation and management method for remote sensing images based on multi-source information according to claim 3, characterized in that: In S2, for the linked regions of the sub-regions, the feature similarity extracted based on the feature values ​​of the sub-regions is used to determine the merging weight of the linked regions of each sub-region. The feature value set of each sub-region is then extracted using the extraction unit as U = {U1, U2, ..., U...}. n }, where, for the required sub-region p i The set of features belonging to the same region is a subset of the set of feature values. The feature similarity between sub-regions is calculated according to the following formula: Where pi and pj are subsets of the feature sets of two different sub-regions, σ(P i P j ) is a subregion p i and p j Feature similarity, sim(p i ,p j Let be the feature similarity between two linked regions pi and pj, and let E be a similarity evaluation function. Compare the feature similarity between the two linked regions pi and pj: if the similarity functions are equal, it means that the two sub-regions belong to the same feature region; if the similarity functions are not equal, it means that the two sub-regions do not belong to the same feature region.

5. The intelligent segmentation and management method for remote sensing images based on multi-source information according to claim 4, characterized in that: In S2, using the determining unit, adjacent sub-regions with feature similarity greater than the similarity feature value and area greater than the sub-region are identified as the linked regions of the sub-region. When the feature similarity between a sub-region and its adjacent sub-regions is greater than the set similarity feature value, it is considered that there is a link relationship between the sub-region and its adjacent sub-regions. The area of ​​the two sub-regions is compared, and the adjacent sub-region with an area greater than that of the sub-region is identified as the linked region of the sub-region. The linking direction of the sub-region points to its linked region. The similarity feature value can be set separately for different features according to requirements.

6. The intelligent segmentation and management method for remote sensing images based on multi-source information according to claim 5, characterized in that: In S4, the processing unit normalizes sub-segments with the same feature similarity. The saliency value of a pixel is defined by the contrast between its color and the colors of pixels in other sub-segments within the sub-segment. Pixel I in sub-segment I... k The significance of is defined as: Among them, C l For pixel I k The color value, n is the total number of colors in the image, f j The number of clusters in which Cj appears in image I is used to perform spectral clustering integration of the sub-segmentation results with the same features according to the final number of clusters, thereby obtaining the segmentation result of the given image.

Images