Image processing method and image processing system based on remote sensing image
By using the method of ring analysis domain and grayscale symbiosis matrix feature values in remote sensing image processing, the problem of insufficient adaptability of remote sensing image recognition model in the new environment is solved, and high accuracy distinction between weeds and crops is achieved.
Patent Information
- Application Number
- CN202510263144.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-06
- Publication Date
- 2025-06-20
AI Technical Summary
The existing remote sensing image recognition model is insufficiently adaptable to the new environment, especially when weeds and crop plants are mixed with high levels, and the recognition accuracy is difficult to ensure.
The image processing method based on remote sensing images is adopted to accurately distinguish plant objects and weed objects by specifying the range color extraction, separation processing, establishing ring analysis domains, calculating the feature values of grayscale symbiosis matrix and classification statistics.
Improves the accuracy of weeds and crops, and enhances the model's adaptability in new environments, especially when weeds and crops are well mixed.
Smart Images
Figure CN120182850A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the technical field of data processing, and in particular, to an image processing method and an image processing system based on remote sensing images. Background Art
[0002] Remote sensing images are images made after sensors obtain electromagnetic wave information of various ground objects on the ground surface at high altitudes (such as on platforms like satellites and airplanes). These sensors can capture the electromagnetic waves reflected or emitted by the ground surface, convert them into digital or analog signals, and then form images after processing. With the development of small low-altitude aircraft in recent years, a low-altitude remote sensing image imaging method with unmanned aerial vehicles as the carrier has emerged, which can further reduce the cost of remote sensing images and make them more convenient to use.
[0003] In the process of large-scale agricultural production, remote sensing images can be used for the identification and processing of weeds, accurately identifying the types and quantities of weeds and providing precise weeding guidance. The advantages of this method are very obvious, but there are also certain problems at present.
[0004] One problem is the limitation of the generalization use of the recognition model, which is mainly manifested in that the recognition model is restricted by the initial design and training process and has insufficient adaptability to new environments; when the mixture degree of weeds and crop plants is relatively high, special recognition conditions (growth stage, addition of specified substances) are required for assistance, and there are certain difficulties in implementation. Summary of the Invention
[0005] This application provides an image processing method and an image processing system based on remote sensing images, which can distinguish plant objects and weed objects through regionalized type recognition and statistical analysis methods, and at the same time use a multi-region assisted recognition processing method for the same position, which can further improve the accuracy of discrimination.
[0006] The above object of this application is achieved through the following technical solutions:
[0007] In a first aspect, this application provides an image processing method based on remote sensing images, including:
[0008] Using a specified range color extraction method to extract analysis objects in the remote sensing image, where the analysis objects include plant objects and weed objects;
[0009] Separating the analysis objects and determining the central positions of the analysis objects;
[0010] Establishing a plurality of circular analysis domains based on the central positions and obtaining feature images located on the circular analysis domains;
[0011] Calculating the gray-level co-occurrence matrix of the feature images and obtaining the eigenvalues of the gray-level co-occurrence matrix;
[0012] Classify the feature images using eigenvalues, and preliminarily determine the type of the analysis object according to the classification results;
[0013] Statistically classify the classification results belonging to the same central position and determine the type of the analysis object according to the statistical results;
[0014] Among them, in the direction far from the central position, the diameter of the annular analysis domain belonging to the same central position tends to increase.
[0015] In a possible implementation manner of the first aspect, determining the central position of the analysis object includes:
[0016] Divide the extracted remote sensing image into multiple regions;
[0017] Perform edge extraction in each region to obtain edge extraction objects;
[0018] Calculate the center line of each edge extraction object and determine the convergence region of the center lines;
[0019] Obtain the center point of the convergence region and use the center point of the convergence region as the central position of the analysis object.
[0020] In a possible implementation manner of the first aspect, after obtaining the feature image located on the annular analysis domain, it further includes:
[0021] Perform segmentation processing on the feature image to obtain multiple fragment images;
[0022] Use the attribution relationship to screen the fragment images, retain the fragment images associated with the central position, and the retained fragment images are recorded as the feature images on the annular analysis domain.
[0023] In a possible implementation manner of the first aspect, when it is difficult to determine the attribution relationship of the fragment images, it further includes:
[0024] Determine the edge of the fragment image whose attribution relationship is difficult to determine, and record it as the analysis edge;
[0025] Establish a reference center line based on the analysis edge;
[0026] Extend the reference center line and obtain the analysis objects suspected of being associated with the analysis edge on the extended reference center line, and the number of the associated analysis objects is multiple;
[0027] Expand the analysis edge based on the reference center line so that the analysis edge establishes a relationship with at least one of the associated analysis objects.
[0028] In a possible implementation of the first aspect, when establishing multiple circular analysis domains based on the central position, the circular analysis domains belonging to different central positions do not overlap.
[0029] In a possible implementation of the first aspect, after determining the type of the analysis object according to the statistical result, it further includes adjusting the diameter of the circular analysis domain and determining the type of the analysis object again;
[0030] When reducing the diameter of the circular analysis domain, the statistical result is greater than or equal to the previous statistical result;
[0031] When enlarging the diameter of the circular analysis domain, the statistical result is not greater than the previous statistical result.
[0032] In a possible implementation of the first aspect, when the statistical result is less than the previous statistical result, it further includes:
[0033] Determining the circular analysis domain whose result changes after being enlarged or reduced, and recording it as an abnormal circular analysis domain;
[0034] Moving the abnormal circular analysis domain and recording the moving range of the abnormal circular analysis domain. When moving the abnormal circular analysis domain, the result of the abnormal circular analysis domain remains unchanged;
[0035] Recording the moving range of the abnormal circular analysis domain as the suspected range and determining the central position of the suspected range;
[0036] Taking the central position of the suspected range as the new central position.
[0037] In the second aspect, the present application provides an image processing device based on remote sensing images, including:
[0038] An image extraction unit, configured to extract an analysis object in the remote sensing image by using a specified range color extraction method, where the analysis object includes a plant object and a weed object;
[0039] An image separation unit, configured to perform separation processing on the analysis object and determine the central position of the analysis object;
[0040] An image acquisition unit, configured to establish multiple circular analysis domains based on the central position and obtain a feature image located on the circular analysis domain;
[0041] A calculation unit, configured to calculate the gray-level co-occurrence matrix of the feature image and obtain the eigenvalue of the gray-level co-occurrence matrix;
[0042] An image classification unit, configured to classify the feature image by using the eigenvalue and preliminarily determine the type of the analysis object according to the classification result;
[0043] A category determination unit, configured to count classification results belonging to the same central position and determine the category of the analysis object according to the statistical results;
[0044] Wherein, in the direction away from the central position, the diameter of the annular analysis domain belonging to the same central position tends to increase.
[0045] In a third aspect, the present application provides an image processing system based on remote sensing images, the system comprising:
[0046] One or more memories, configured to store instructions; and
[0047] One or more processors, configured to call and run the instructions from the memory and execute the method described in the first aspect and any possible implementation manner of the first aspect.
[0048] In a fourth aspect, the present application provides a computer-readable storage medium, the computer-readable storage medium comprising:
[0049] A program, when the program is run by a processor, the method described in the first aspect and any possible implementation manner of the first aspect is executed.
[0050] In a fifth aspect, the present application provides a computer program product, comprising program instructions, when the program instructions are run by a computing device, the method described in the first aspect and any possible implementation manner of the first aspect is executed.
[0051] In a sixth aspect, the present application provides a chip system, the chip system comprising a processor, configured to implement the functions involved in the above aspects, for example, generating, receiving, sending, or processing the data and / or information involved in the above method.
[0052] The chip system may be composed of chips or may include chips and other discrete devices.
[0053] In a possible design, the chip system further comprises a memory, the memory being configured to store necessary program instructions and data. The processor and the memory may be decoupled and disposed on different devices, connected by wire or wirelessly, or the processor and the memory may also be coupled on the same device.
[0054] Description of the drawings
[0055] Figure 1 is a schematic block diagram of the steps of an image processing method based on remote sensing images provided by the present application.
[0056] Figure 2 is a schematic diagram of establishing an annular analysis domain provided by the present application.
[0057] Figure 3 It is a schematic diagram provided by this application that gives results using an annular analysis domain.
[0058] Figure 4 It is a schematic diagram provided by this application for classification and statistics based on the finally obtained results.
[0059] Figure 5 It is a schematic diagram provided by this application for dividing multiple regions.
[0060] Figure 6 It is provided by this application based on Figure 5 A schematic diagram of re - region division.
[0061] Figure 7 It is a schematic diagram provided by this application that the center line of a fragment image passes through a certain central position.
[0062] Figure 8 It is a schematic diagram provided by this application that the center line of a fragment image does not pass through the central position.
[0063] Figure 9 It is a schematic diagram provided by this application for determining the relationship of fragment images using the relationship of analysis objects. Detailed implementation manners
[0064] The following further elaborates on the technical solutions in this application with reference to the accompanying drawings.
[0065] This application discloses an image - processing method based on remote - sensing images. In some examples, please refer to Figure 1 , the image - processing method based on remote - sensing images disclosed in this application includes the following steps:
[0066] S101, extracting analysis objects in the remote - sensing image using a specified - range color extraction method, where the analysis objects include plant objects and weed objects;
[0067] S102, separating the analysis objects and determining the central positions of the analysis objects;
[0068] S103, establishing multiple annular analysis domains based on the central positions and obtaining characteristic images located on the annular analysis domains;
[0069] S104, calculating the gray - level co - occurrence matrix of the characteristic images and obtaining the eigenvalues of the gray - level co - occurrence matrix;
[0070] S105, classifying the characteristic images using the eigenvalues and preliminarily determining the types of the analysis objects according to the classification results;
[0071] S106. Statistically classify the classification results belonging to the same central position and determine the type of the analysis object according to the statistical results;
[0072] Among them, in the direction away from the central position, the diameter of the annular analysis domain belonging to the same central position tends to increase.
[0073] Generally speaking, in step S101, the analysis objects in the remote sensing image are first extracted by using the specified range color extraction method. The analysis objects include plant objects and weed objects, and the specified range color refers to the colors of the plant objects and weed objects.
[0074] The reason for being able to use the specified range color is that different colors correspond to different wavelengths of light. Under the limiting conditions of the specified range color extraction, the plant objects and weed objects can be directly extracted from the remote sensing image. Generally speaking, the specified range color refers to green. Of course, the green here refers to a wavelength range.
[0075] Then in step S102, the analysis objects are separated and the central position of the analysis object is determined. On the plane parallel to the ground, the central position generally refers to the position where the root system of the analysis object is located. The purpose of determining the central position of the analysis object is to conduct quantity statistics.
[0076] It should be understood that when the difference between the plant object and the weed object is large, classification can be carried out by means of shape distinction or even color distinction. However, when the difference between the plant object and the weed object is small, direct distinction cannot be made.
[0077] The processing method of this application is to establish an annular analysis domain based on the central position. Figure 2 As shown, then classification is carried out through the features on the annular analysis domain. At this time, two classification results will be obtained, namely plant objects and weed objects. The number of weed objects may be one or multiple.
[0078] Then, the way of quantity statistics is used to determine which are weeds, because the number of plant objects is significantly more than that of weed objects.
[0079] In step S103, multiple annular analysis domains are established based on the central position and the feature images located on the annular analysis domain are obtained. Here, the feature images on the annular analysis domain belong to the central position. The number of annular analysis domains is multiple, and the diameters of these annular analysis domains are different but all use the central position as the center of the circle.
[0080] In step S104, the gray-level co-occurrence matrix of the feature image will be calculated and the eigenvalues of the gray-level co-occurrence matrix will be obtained. Then, the eigenvalues are used to classify the feature image, and the type of the analysis object is initially determined according to the classification results ( Figure 3 as shown), that is, the content in step S105.
[0081] Finally, in step S106, the classification results belonging to the same central position are counted, and the type of the analysis object is determined according to the statistical results. Here, the number of classification results is the same as the number of circular analysis domains. The reason for counting the classification results belonging to the same central position is that the classification results may not be consistent. For example, among three classification results, there are two plant object classifications and one weed object classification.
[0082] The reason why the classification results cannot guarantee consistency is that when establishing the circular analysis domain based on the central position, the feature image located on the circular analysis domain may not correspond to the central position corresponding to the circular analysis domain. However, when the number of classification results increases, this non-correspondence can be masked by statistical means to minimize the negative impact on the final result.
[0083] Generally speaking, the image processing method based on remote sensing images provided by this application distinguishes plant objects and weed objects by using the central position, establishing a circular analysis domain based on the central position, and calculating the eigenvalues of the gray-level co-occurrence matrix of the circular analysis domain. This method does not require separate modeling for each type of plant, but instead classifies and counts according to the final obtained results. As Figure 4 shown, it can conveniently obtain plant objects and weed objects by quantity.
[0084] The relevant explanations for the gray-level co-occurrence matrix are as follows:
[0085] The gray-level co-occurrence matrix is defined as the probability that the gray value at a point that is at a distance d and in a direction θ from a pixel with gray value i is j. All estimated values can be represented in the form of a matrix, so it is called the gray-level co-occurrence matrix. It captures the texture features of the image by analyzing the spatial relationship between pixels and the statistical distribution of gray levels in the image.
[0086] The calculation method is as follows:
[0087] Gray-level quantization: Quantize the gray levels of the image. Usually, the gray levels are divided into 256 levels (0 - 255).
[0088] Determine the distance and direction: Select appropriate distance d and direction θ to calculate the gray-level co-occurrence matrix.
[0089] Statistical co-occurrence frequency: Traverse each pixel of the image and count the frequency of occurrence between the gray levels of adjacent pixels in the given direction.
[0090] Construct the matrix: According to the statistical results, construct a gray-level co-occurrence matrix, where the elements in the matrix represent the co-occurrence frequency of the corresponding gray-level pairs.
[0091] Normalization: Usually, the gray-level co-occurrence matrix is normalized so that the sum of all elements in the matrix is equal to 1.
[0092] The eigenvalues of the gray-level co-occurrence matrix refer to:
[0093] Angular Second Moment (ASM), the calculation formula is ASM = sum(p(i,j).^2), where p(i,j) is the element of the normalized gray-level co-occurrence matrix, which reflects the evenness of the image gray-level distribution and the coarseness of the texture. When the image texture is uniform and fine, and the gray-level distribution is uniform, the ASM value is larger; on the contrary, the ASM value is smaller.
[0094] Entropy (ENT), the calculation formula is ENT = -sum(p(i,j)*log(p(i,j))), which describes the measure of the amount of information in the image and indicates the complexity of the image. When the image complexity is high, the entropy value is larger; on the contrary, the entropy value is smaller.
[0095] Inverse Differential Moment (IDM), which reflects the clarity and regularity of the texture. For an image with clear texture, strong regularity, and easy to describe, the IDM value is larger; for a chaotic image, the IDM value is smaller.
[0096] Contrast, which describes the amplitude of the gray-level change in the image. The higher the contrast, the more obvious the texture change. The larger the number of pixel pairs with a large gray-level difference (i.e., contrast), the larger this value.
[0097] Homogeneity, which reflects the evenness of the gray-level distribution in the image. The higher the homogeneity, the smoother the texture.
[0098] Correlation, which is used to measure the similarity degree of the gray levels of the image in the row or column direction. The higher the correlation, the more ordered the texture.
[0099] Through one or several of the above eigenvalues, the annular analysis domain can be classified.
[0100] In some examples, the specific method for determining the center position of the analysis object is:
[0101] S201, divide the extracted remote sensing image into multiple regions;
[0102] S202, perform edge extraction in each region to obtain edge extraction objects;
[0103] S203, calculate the center line of each edge extraction object and determine the convergence region of the center lines;
[0104] S204, obtain the center point of the convergence area and use the center point of the convergence area as the central position of the analysis object.
[0105] The content from step S201 to step S204 is that first, the extracted remote sensing image is divided into multiple regions, then edge extraction is performed in each region to obtain edge extraction objects. Here, the edge extraction objects refer to the edges of the leaves. Then, calculate the center line of each edge extraction object and determine the convergence area of the center line. Finally, calculate the center point of the convergence area and use the center point of the convergence area as the central position of the analysis object.
[0106] The purpose of dividing the extracted remote sensing image into multiple regions is to narrow the processing scope. A suitable division method is to determine according to the occupied areas of the plant objects and weed objects, and the occupied areas are obtained through sampling and statistical analysis. Generally, the area of a divided region is 2 - 3 times the occupied area.
[0107] Meanwhile, during the process of dividing into multiple regions, a partially overlapping division method is used to ensure that the center point of the convergence area can be fully obtained. As Figure 5 and Figure 6 shown, this method can ensure that the center point of the convergence area located on the dividing line ( Figure 5 and Figure 6 the dashed line in) can also be found.
[0108] In some examples, after obtaining the feature image on the annular analysis domain, the following content is also added:
[0109] S301, perform segmentation processing on the feature image to obtain multiple fragmented images;
[0110] S302, use the attribution relationship to screen the fragmented images, retain the fragmented images associated with the central position, and the retained fragmented images are recorded as the feature images on the annular analysis domain.
[0111] In step S301 and step S302, the feature image will be segmented to obtain multiple fragmented images. The fragmented images refer to Figure 3 the individual leaves located between two circles in, and here the individual leaves are part of a complete leaf.
[0112] Then use the attribution relationship to screen the fragmented images, retain the fragmented images associated with the central position, and the retained fragmented images are recorded as the feature images on the annular analysis domain.
[0113] The specific method of using the attribution relationship to screen the fragmented images is to determine the center line of the fragmented image, and then extend the obtained center line. The extension will result in the center line passing through a certain central position (Figure 7 As shown, the dashed line in the figure represents the center line) and does not pass through the central position ( Figure 8 As shown, the dashed line in the figure represents the center line) of the case where when the center line passes through a certain central position, the fragment image is associated with the central position.
[0114] Of course, considering the calculation error, when the center line does not pass through the central position but the minimum straight-line distance from the central position is less than or equal to the set distance value, it is also considered that the center line passes through the central position.
[0115] When the attribution relationship of the fragment image is difficult to determine, the following method is used for processing:
[0116] S401, determine the edge of the fragment image whose attribution relationship is difficult to determine, denoted as the analysis edge;
[0117] S402, establish a reference center line based on the analysis edge;
[0118] S403, extend the reference center line and obtain analysis objects on the extended reference center line that are suspected to be associated with the analysis edge, and the number of the associated analysis objects is multiple;
[0119] S404, expand the analysis edge based on the reference center line so that the analysis edge establishes a relationship with at least one of the associated analysis objects.
[0120] In steps S401 to S404, use the extended reference center line and obtain analysis objects on the extended reference center line that are suspected to be associated with the analysis edge to determine the associated analysis objects. At this time, the number of the associated analysis objects may be one or multiple.
[0121] Then expand the analysis edge based on the reference center line so that the analysis edge establishes a relationship with at least one of the associated analysis objects, that is, associate the fragment image whose attribution relationship is difficult to determine with an analysis object, and then use the relationship of the analysis object to determine the relationship of the fragment image that is difficult to determine, as Figure 9 shown.
[0122] In some examples, when establishing multiple circular analysis domains based on the central position, the circular analysis domains belonging to different central positions do not overlap, aiming to avoid interference between the circular analysis domains.
[0123] In some examples, after determining the type of the analysis object according to the statistical result, it further includes adjusting the diameter of the circular analysis domain and determining the type of the analysis object again. The adjustment method is as follows:
[0124] When reducing the diameter of the circular analysis domain, the statistical result is greater than or equal to the previous statistical result;
[0125] When the diameter of the enlarged annular analysis domain is increased, the statistical result is not greater than the previous statistical result.
[0126] When the statistical result is less than the previous statistical result, the specific processing method is as follows:
[0127] S501, determine the annular analysis domain whose result has changed after enlargement or reduction, and record it as the abnormal annular analysis domain;
[0128] S502, move the abnormal annular analysis domain and record the moving range of the abnormal annular analysis domain. When moving the abnormal annular analysis domain, the result of the abnormal annular analysis domain remains unchanged;
[0129] S503, record the moving range of the abnormal annular analysis domain as the suspected range and determine the central position of the suspected range;
[0130] S504, use the central position of the suspected range as the new central position.
[0131] Specifically, determine the annular analysis domain whose result has changed after enlargement or reduction, and record it as the abnormal annular analysis domain. At this time, the number of abnormal annular analysis domains may be one or more. The reason for the appearance of the abnormal annular analysis domain is that there is an omission in determining the central position in the early stage.
[0132] At this time, move the abnormal annular analysis domain and record the moving range of the abnormal annular analysis domain. It is necessary to ensure that the result of the abnormal annular analysis domain remains unchanged when moving the abnormal annular analysis domain. After obtaining the moving range of the abnormal annular analysis domain, record the moving range of the abnormal annular analysis domain as the suspected range and determine the central position of the suspected range. Finally, use the central position of the suspected range as the new central position.
[0133] For the newly obtained central position, the method described in the foregoing content is still used for processing.
[0134] This application also provides an image processing device based on remote sensing images, including:
[0135] An image extraction unit for extracting analysis objects in the remote sensing image using a specified range color extraction method. The analysis objects include plant objects and weed objects;
[0136] An image separation unit for separating the analysis objects and determining the central position of the analysis objects;
[0137] An image acquisition unit for establishing a plurality of annular analysis domains based on the central position and obtaining the feature images located on the annular analysis domains;
[0138] A calculation unit for calculating the gray-level co-occurrence matrix of the feature image and obtaining the eigenvalues of the gray-level co-occurrence matrix;
[0139] An image classification unit for classifying a feature image using eigenvalue and preliminarily determining the type of the analysis object according to the classification result;
[0140] A type determination unit for counting the classification results belonging to the same central position and determining the type of the analysis object according to the statistical result;
[0141] Wherein, in the direction away from the central position, the diameter of the annular analysis domain belonging to the same central position tends to increase.
[0142] Furthermore, it further includes:
[0143] A region division unit for dividing the extracted remote sensing image into multiple regions;
[0144] An edge extraction unit for performing edge extraction in each region to obtain an edge extraction object;
[0145] A converging region determination unit for calculating the center line of each edge extraction object and determining the converging region of the center line;
[0146] A central position determination unit for obtaining the center point of the converging region and taking the center point of the converging region as the central position of the analysis object.
[0147] Furthermore, it further includes:
[0148] A segmentation processing unit for performing segmentation processing on the feature image to obtain multiple fragment images;
[0149] A screening processing unit for screening the fragment images using the attribution relationship, retaining the fragment images associated with the central position, and the retained fragment images are recorded as the feature images on the annular analysis domain.
[0150] Furthermore, it further includes:
[0151] An analysis edge determination unit for determining the edge of the fragment image whose attribution relationship is difficult to determine, denoted as the analysis edge;
[0152] A reference center line establishment unit for establishing a reference center line based on the analysis edge;
[0153] An association relationship processing unit for extending the reference center line and obtaining the analysis objects suspected of having an association with the analysis edge on the extended reference center line, and the number of the associated analysis objects is multiple;
[0154] An analysis edge expansion unit for expanding the analysis edge based on the reference center line so that the analysis edge establishes a relationship with at least one of the associated analysis objects.
[0155] Further, when establishing multiple annular analysis domains based on the central position, the annular analysis domains belonging to different central positions do not overlap.
[0156] Further, after determining the type of the analysis object according to the statistical result, it further includes adjusting the diameter of the annular analysis domain and determining the type of the analysis object again;
[0157] When reducing the diameter of the annular analysis domain, the statistical result is greater than or equal to the previous statistical result;
[0158] When enlarging the diameter of the annular analysis domain, the statistical result is not greater than the previous statistical result.
[0159] Further, it further includes:
[0160] An abnormal annular analysis domain determination unit, configured to determine the annular analysis domain whose result changes after being enlarged or reduced, and record it as an abnormal annular analysis domain;
[0161] An abnormal annular analysis domain moving unit, configured to move the abnormal annular analysis domain and record the moving range of the abnormal annular analysis domain. When moving the abnormal annular analysis domain, the result of the abnormal annular analysis domain remains unchanged;
[0162] A process processing unit, configured to record the moving range of the abnormal annular analysis domain as a suspected range and determine the central position of the suspected range;
[0163] A result determination unit, configured to use the central position of the suspected range as the new central position.
[0164] In one example, the units in any of the above devices can be one or more integrated circuits configured to implement the above methods. For example: one or more application specific integrated circuits (ASICs), or, one or more digital signal processors (DSPs), or, one or more field programmable gate arrays (FPGAs), or a combination of at least two of these integrated circuit forms.
[0165] Again, when the units in the device can be implemented in the form of a processing element scheduler, the processing element can be a general-purpose processor, such as a central processing unit (CPU) or other processors that can call programs. Again, these units can be integrated together to be implemented in the form of a system-on-a-chip (SOC).
[0166] In this application, names may be given to various objects such as various messages / information / devices / network elements / systems / devices / actions / operations / processes / concepts, etc. It can be understood that these specific names do not constitute a limitation on the relevant objects, and the given names may change with factors such as scenarios, contexts, or usage habits. The understanding of the technical meaning of the technical terms in this application should be mainly determined from the functions and technical effects reflected / executed in the technical solutions.
[0167] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.
[0168] In several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.
[0169] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0170] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or by a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of this application.
[0171] It should also be understood that in each embodiment of this application, the first, second, etc. are only used to indicate that multiple objects are different. For example, the first time window and the second time window are only used to indicate different time windows. And it should not have any impact on the time window itself. The above first, second, etc. should not impose any limitations on the embodiments of this application.
[0172] It should also be understood that in various embodiments of the present application, if there is no special description and logical conflict, the terms and / or descriptions between different embodiments are consistent and can be referenced to each other, and the technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationships.
[0173] If the above-mentioned function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a computer-readable storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present application. The aforementioned computer-readable storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical discs that can store program codes.
[0174] The present application also provides an image processing system based on remote sensing images, and the system includes:
[0175] One or more memories for storing instructions; and
[0176] One or more processors for calling and running the instructions from the memory and executing the methods described above.
[0177] The present application also provides a computer program product, which includes instructions that, when executed, cause the terminal device and the network device to perform the operations of the terminal device and the network device corresponding to the above methods.
[0178] The present application also provides a chip system, which includes a processor for implementing the functions involved above. For example, generating, receiving, sending, or processing the data and / or information involved in the above methods.
[0179] This chip system can be composed of chips or can include chips and other discrete devices.
[0180] The processor mentioned anywhere above can be a CPU, a microprocessor, an ASIC, or an integrated circuit for controlling the execution of a program of the above method for transmitting feedback information.
[0181] In a possible design, the chip system further includes a memory for storing necessary program instructions and data. The processor and the memory can be decoupled and disposed on different devices respectively, and connected by wired or wireless means to support the chip system to implement various functions in the above embodiments. Alternatively, the processor and the memory can also be coupled on the same device.
[0182] Optionally, the computer instructions are stored in the memory.
[0183] Optionally, the memory is a storage unit within the chip, such as a register, a cache, etc. The memory can also be a storage unit outside the chip within the terminal, such as a ROM or other types of static storage devices that can store static information and instructions, a RAM, etc.
[0184] It can be understood that the memory in the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories.
[0185] The non-volatile memory can be a ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), or a flash memory.
[0186] The volatile memory can be a RAM, which is used as an external cache. There are various different types of RAM, such as a static random access memory (SRAM), a dynamic random access memory (DRAM), a synchronous DRAM (SDRAM), a double data rate SDRAM (DDR SDRAM), an enhanced SDRAM (ESDRAM), a synch link DRAM (SLDRAM), and a direct memory bus random access memory.
[0187] The embodiments of the present specific implementation manners are all preferred embodiments of the present application, and do not limit the protection scope of the present application accordingly. Therefore, all equivalent changes made according to the structure, shape, and principle of the present application should be covered within the protection scope of the present application.
Claims
1. An image processing method based on remote sensing images, characterized in that: include: Extract the analysis objects in the remote sensing image using the specified range color extraction method, and the analysis objects include plant objects and weed objects; Separate the analysis object and determine the center position of the analysis object; Establishing multiple annular analysis domains based on the center position and obtaining characteristic images located on the annular analysis domains; Calculate the gray level co-occurrence matrix of the feature image and obtain the eigenvalue of the gray level co-occurrence matrix; Use the eigenvalues to classify the feature images, and preliminarily determine the type of the analysis object according to the classification results; Count the classification results belonging to the same central position and determine the type of analysis object based on the statistical results; Among them, in the direction away from the central position, the diameter of the annular analysis domain belonging to the same central position tends to increase.
2. The image processing method based on remote sensing images according to claim 1, characterized in that: Determining the center position of the analysis object includes: Divide the extracted remote sensing image into multiple regions; Perform edge extraction in each region to obtain an edge extraction object; Calculate the center line of each edge extraction object and determine the collection area of the center lines; The center point of the collection area is obtained and used as the center position of the analysis object.
3. The image processing method based on remote sensing images according to claim 1 or 2, characterized in that: After obtaining the characteristic image located in the annular analysis domain, it also includes: Segment the feature image to obtain multiple fragment images; The fragment images are screened using the attribution relationship, and the fragment images associated with the central position are retained. The retained fragment images are recorded as feature images on the annular analysis domain.
4. The image processing method based on remote sensing images according to claim 2, characterized in that: When the ownership of the fragmented images is difficult to determine, it also includes: Determine the edges of the fragmented images whose attribution is difficult to determine, and record them as analysis edges; Establish a reference centerline based on the analyzed edge; Extending the reference center line and acquiring analysis objects on the extended reference center line that are suspected to be associated with the analysis edge, wherein the number of the associated analysis objects is multiple; The analysis edge is expanded based on the reference center line, so that the analysis edge is related to at least one of the associated analysis objects.
5. The image processing method based on remote sensing images according to claim 1, characterized in that: When multiple annular analysis domains are established based on the central position, the annular analysis domains belonging to different central positions do not overlap.
6. The image processing method based on remote sensing images according to claim 1 or 5, characterized in that: After the type of the analysis object is determined according to the statistical results, the method further includes adjusting the diameter of the annular analysis domain and determining the type of the analysis object again; When the diameter of the annular analysis domain is reduced, the statistical result is greater than or equal to the previous statistical result; When the diameter of the annular analysis domain is enlarged, the statistical result is no greater than the previous statistical result.
7. The image processing method based on remote sensing images according to claim 6, characterized in that: When the statistical result is less than the previous statistical result, it also includes: Determine the annular analysis domain whose results change after being enlarged or reduced, and record it as an abnormal annular analysis domain; Move the abnormal annular analysis domain and record the moving range of the abnormal annular analysis domain. When the abnormal annular analysis domain is moved, the result of the abnormal annular analysis domain remains unchanged. The moving range of the abnormal annular analysis domain is recorded as the suspected range and the center position of the suspected range is determined; The center position of the suspected range is taken as the new center position.
8. An image processing device based on remote sensing images, characterized in that: include: An image extraction unit is used to extract analysis objects in the remote sensing image using a specified range color extraction method, where the analysis objects include plant objects and weed objects; An image separation unit, used for separating the analysis object and determining the center position of the analysis object; An image acquisition unit, used to establish multiple annular analysis domains based on the center position and obtain characteristic images located on the annular analysis domains; A calculation unit, used for calculating the gray level co-occurrence matrix of the feature image and obtaining the eigenvalue of the gray level co-occurrence matrix; An image classification unit, used to classify the feature image using the feature value, and preliminarily determine the type of the analysis object according to the classification result; A category determination unit, used to count the classification results belonging to the same central position and determine the category of the analysis object according to the statistical results; Among them, in the direction away from the central position, the diameter of the annular analysis domain belonging to the same central position tends to increase.
9. An image processing system based on remote sensing images, characterized in that: The system comprises: one or more memories for storing instructions; and One or more processors, configured to call and execute the instructions from the memory to perform the method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that: The computer readable storage medium comprises: The program, when the program is executed by a processor, the method according to any one of claims 1 to 7 is executed.