Inland water optical classification method, system, equipment, medium and program

Abstract

The invention discloses an inland water optical classification method, system, equipment, medium and program, and belongs to the technical field of water optical classification. The method comprises the following steps: acquiring multi-period remote sensing data, and preprocessing the multi-period remote sensing data to obtain preprocessed remote sensing image data; performing quantization processing on the preprocessed remote sensing image data to obtain the remote sensing image data after quantization processing; based on the quantized remote sensing image data, color features of the water body type are extracted; processing the color features of the extracted water body type to obtain feature components; and processing the feature components by using a classifier to obtain a classification result of the water body types. Based on the computer vision technology, the collected water body samples are preprocessed to form the color feature parameter library, and the inland water body optical classification precision is improved according to the inland water body optical classification model.

Description

Technical Field

[0001] The present invention relates to the technical field of water body optical classification, and particularly relates to an inland water body optical classification method, system, device, medium and program based on computer vision technology. Background Art

[0002] As an important part of remote sensing monitoring of water environment, the development process and technological evolution of water body optical classification are closely related to the global water environment monitoring requirements and the rapid development of remote sensing technology. Since the 1950s, scientists have begun to pay attention to and study the optical differences between different water bodies. The pioneering work in this field can be traced back to the research of Jerlov and Koczy in 1951. By collecting voyage observation data worldwide, they first revealed the significant differences in the optical properties of water bodies in different sea areas, especially the key optical index of diffuse attenuation coefficient. The change of this index is mainly attributed to the spatial distribution differences of suspended particulate matter and colored dissolved organic matter in different water areas. This discovery laid a theoretical foundation for subsequent water body optical classification research and also pointed out the internal connection between water body optical properties and water quality parameters.

[0003] With the continuous progress of remote sensing technology, water body optical classification methods have gradually evolved from theoretical exploration to practical application, forming a variety of classification systems. Among them, the classification method based on the inherent optical properties of water bodies is the most fundamental one. This type of method directly measures or uses remote sensing data to calculate the inherent optical properties of water bodies (such as absorption coefficient, scattering coefficient, etc.), and then classifies water bodies according to these properties. However, the calculation process of inherent optical properties is complex and involves the interaction of multiple optical processes. Therefore, error accumulation is likely to occur in actual operation, limiting the wide application of this type of method in remote sensing inversion of water environment parameters.

[0004] To overcome the above limitations, a classification method based on the waveform characteristics of remote sensing reflectance has emerged. This type of method identifies different types of water bodies by analyzing the variation characteristics of remote sensing reflectance with wavelength, that is, the remote sensing reflectance waveform. Since the remote sensing reflectance directly reflects the spectral characteristics of the light reflected from the water body surface, this method can more intuitively reflect the optical properties of water bodies. However, this method has certain limitations in actual application. On the one hand, due to the remote sensing reflectance waveform being affected by multiple factors, such as solar angle, water depth, atmospheric conditions, etc., its stability and accuracy are often challenged. On the other hand, this method focuses more on the inversion of specific water environment indicators, such as chlorophyll a concentration, suspended sediment concentration, etc., resulting in the obtained water body types may not be suitable for the model construction of other specific parameters. In addition, this method usually relies on specific datasets to build models, and there are differences in water body types in different study areas, which limits the universality and transferability of the models. Summary of the Invention

[0005] In view of the problems existing in the prior art in the optical classification of water bodies, mainly including complex calculations, error accumulation, poor model universality, and limited feature extraction, resulting in low accuracy of the optical classification of inland water bodies. The present invention provides an optical classification method for inland water bodies based on computer vision technology. Based on computer vision technology, the collected water body samples are preprocessed to form a color feature parameter library, and according to the optical classification model of inland water bodies, the accuracy of the optical classification of inland water bodies is improved.

[0006] To achieve the above object, the present invention provides the following technical solutions.

[0007] In a first aspect, the present invention provides an optical classification method for inland water bodies based on computer vision technology, including: Obtain multi-temporal remote sensing data, and preprocess the multi-temporal remote sensing data to obtain preprocessed remote sensing image data; Perform quantization processing on the preprocessed remote sensing image data to obtain quantized remote sensing image data; Based on the quantized remote sensing image data, extract the color features of the water body type; Process the extracted color features of the water body type to obtain feature components; Use a classifier to process the feature components to obtain the classification result of the water body type.

[0008] As a further improvement of the present invention, the step of obtaining multi-temporal remote sensing data and preprocessing the multi-temporal remote sensing data to obtain preprocessed remote sensing image data includes: Use remote sensing data to obtain water body color parameters and multi-temporal remote sensing data; Perform atmospheric correction, cloud removal, flare correction, water body identification, and water-leaving reflectance correction on the multi-temporal remote sensing data to obtain preprocessed remote sensing image data.

[0009] As a further improvement of the present invention, the step of performing quantization processing on the preprocessed remote sensing image data to obtain quantized remote sensing image data includes: Use the HSV color space model to perform quantization processing on the preprocessed remote sensing image data; Obtain quantized remote sensing image data; The HSV color space model includes:

[0010] In the formula, H is the hue; a, b, c, d, e, f, g are the boundaries of the predetermined hue intervals;

[0011] Where S is the saturation; h, i, and j are preset thresholds;

[0012] Where V is the brightness; k, l, and m are the quantization boundaries of the brightness.

[0013] As a further improvement of the present invention, the color features of the water body type are extracted from the remotely sensed image data after quantization processing, including: Select three types of water bodies: oligotrophic, mesotrophic, and eutrophic. Based on the remotely sensed image data after quantization processing, construct standard templates for the three types of water bodies: oligotrophic, mesotrophic, and eutrophic. Convert the image data of the oligotrophic, mesotrophic, and eutrophic water bodies into the HSV space. Quantize each component in the HSV space to obtain a discretized result for each component. Represent the HSV color information of each water body as a one-dimensional vector, that is, the color feature vector. Calculate the color histogram through the quantized HSV color information. Use the corresponding frequencies that appear in the color histogram as the components of the color feature index of the image to perform similarity measurement with the template to obtain color feature data.

[0014] As a further improvement of the present invention, the processing of the color features of the water body type to obtain feature components includes: The processing of the color features of the water body type to obtain feature components includes: Count each value in the color features of the water body type extracted to obtain the frequency of each value in the color feature dataset. Remove the color features with a frequency of 0 and the color features with a frequency less than the set threshold to obtain the feature components after removal.

[0015] As a further improvement of the present invention, the processing of the feature components using a classifier to obtain the classification result of the water body type includes: Based on the feature components after removal, construct a color feature database. Input the constructed color feature database into the K-Means classifier for water body classification to obtain the classification result.

[0016] In a second aspect, the present invention provides an inland water optical classification system based on computer vision technology, including: Preprocessing data module: used to obtain multi-temporal remotely sensed data and preprocess the multi-temporal remotely sensed data to obtain the preprocessed remotely sensed image data. Quantization processing data module: used to perform quantization processing on the preprocessed remotely sensed image data to obtain the remotely sensed image data after quantization processing. Color feature extraction module: used to extract the color features of water body types based on the remotely sensed image data after quantization processing; Feature component obtaining module: used to process the color features of the extracted water body types to obtain feature components; Classification result acquisition module: used to process the feature components using a classifier to obtain the classification results of water body types.

[0017] In a third aspect, this aspect provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the optical classification method for inland water bodies based on computer vision technology are implemented.

[0018] In a fourth aspect, this aspect provides a computer-readable storage medium. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the steps of the optical classification method for inland water bodies based on computer vision technology are implemented.

[0019] In a fifth aspect, this aspect provides a computer program product, characterized in that it includes computer instructions. When the computer instructions are executed by a processor, the steps of the optical classification method for inland water bodies based on computer vision technology are implemented.

[0020] Compared with the prior art, the present invention has the following beneficial effects: The present invention adopts a multi - period remote sensing data acquisition method. By integrating the data obtained from different sensors and platforms, it can comprehensively reflect the optical characteristics of inland waters. These remote sensing data can include optical images such as MODIS, Landsat, Sentinel - 2, VIIRS, etc. After obtaining these data, precise pre - processing steps are carried out. During the pre - processing process, considering the differences between different remote sensing data and the influence of different geographical environments on optical images, means such as denoising, geometric correction, and radiometric correction are adopted to ensure the quality of the image data and provide an accurate data basis for subsequent processing. After obtaining the pre - processed remote sensing image data, the present invention further conducts quantization processing. The purpose of quantization processing is to convert the original remote sensing image data into a digital format that is convenient for processing and analysis, and eliminate some interference factors caused by data inconsistency, scale differences, etc. The quantized remote sensing data is convenient for extracting the optical characteristics of water bodies, especially color characteristics, because color is one of the most intuitive manifestations of the optical properties of water bodies. By using computer vision technology to extract the color characteristics of water body types, it can accurately reflect different states of water bodies, such as the clarity of water bodies and the distribution of algae. The core advantage of the present invention lies in combining the powerful capabilities of computer vision technology, using color characteristics and machine learning technology to efficiently classify inland waters. Compared with traditional classification methods, the present invention not only significantly improves the classification accuracy, but also can effectively avoid the accumulation of errors, reduce the computational complexity, and improve the universality and stability of the model. In particular, for this specific application scenario of inland waters, the present invention can automatically adjust the classification model according to different geographical, climatic conditions, and water body types, thereby improving the flexibility and accuracy of classification.

[0021] Furthermore, the extracted color characteristics will be converted into feature components. This process involves the analysis and extraction of the color of water bodies to form a set of efficient feature parameters. These feature components will be used as inputs and provided to the classifier for further processing. The classifier can learn through training data, identify the optical characteristics of different types of water bodies, and conduct classification. This classification process adopts advanced machine learning technologies such as support vector machines (SVM), deep learning networks, etc. Through the training of a large number of sample data, an efficient and accurate water body classification model can be constructed. Brief Description of the Drawings

[0022] The drawings described herein are for illustrative purposes only and are not intended to limit the scope of the disclosure of the present invention in any way. In the drawings: Figure 1 It is a schematic flow chart of an optical classification method for inland waters based on computer vision technology of the present invention; Figure 2 It is a schematic process diagram of the color feature extraction technology process of the present invention; Figure 3Schematic diagram of a structure of an inland water optical classification system based on computer vision technology according to the present invention; Figure 4 Schematic diagram of an electronic device in an embodiment of the present invention. Detailed implementation manners

[0023] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. The described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

[0024] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field of the present invention. The terms used in the description of the present invention herein are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0025] Aiming at the problems existing in the prior art in water body optical classification, mainly including complex calculation, error accumulation, poor model universality, and limited feature extraction, resulting in low accuracy of inland water body optical classification. The present invention provides an inland water body optical classification method based on computer vision technology, and the method includes: Aiming at the problems existing in the prior art in water body optical classification, mainly including complex calculation, error accumulation, poor model universality, and limited feature extraction, resulting in low accuracy of inland water body optical classification. The present invention provides an inland water body optical classification method based on computer vision technology, as Figure 1 shown, the method includes: S100: Obtain multi-period remote sensing data, and perform preprocessing on the multi-period remote sensing data to obtain preprocessed remote sensing image data; S200: Perform quantization processing on the preprocessed remote sensing image data to obtain quantized remote sensing image data; S300: Based on the quantized remote sensing image data, extract the color features of the water body type; S400: Process the extracted color features of the water body type to obtain feature components; S500: Use a classifier to process the feature components to obtain the classification result of the water body type.

[0026] Based on computer vision technology, the present invention preprocesses the collected water body samples to form a color feature parameter library, and improves the optical classification accuracy of inland water bodies according to the optical classification model of inland water bodies.

[0027] The following will further explain and illustrate the present invention with reference to the Figure 2 accompanying drawings as shown below.

[0028] An optical classification method for inland water bodies based on computer vision technology, the specific steps include: S1: Data acquisition: water body color parameters and multi-temporal remote sensing data Water body color parameters are usually obtained through remote sensing technology, mainly through satellite images or UAV image data. The common method for obtaining water body color parameters is to use the spectral reflectance in satellite remote sensing data, such as the multi-spectral images of satellites like Landsat, Sentinel-2 or MODIS. By analyzing the reflectance of the water body in different bands (such as blue, green and red), the color information of the water body can be extracted, such as the turbidity, chlorophyll concentration and suspended matter of the water body. Specific methods include reflectance calculation, ratio indices (such as NDWI, Normalized Difference Water Index) and the application of other color indices.

[0029] The acquisition of multi-temporal remote sensing data usually depends on remote sensing satellites and sensors from different sources, and these data can be obtained through open platforms (such as ESA, NASA Earthdata, USGS Earth Explorer) or commercial remote sensing data providers (such as DigitalGlobe). These platforms provide multi-temporal, multi-resolution and multi-band data, which are suitable for different remote sensing analysis tasks. When acquiring, it is necessary to select appropriate satellites (such as Landsat, Sentinel-2, MODIS) and related remote sensing products (such as images, reflectance, vegetation indices, etc.) according to the research purpose.

[0030] S2: Preprocess the remote sensing images by performing atmospheric correction, cloud removal, flare correction, water body identification, and water-leaving reflectance correction.

[0031] The preprocessing of remote sensing images is a very important step in remote sensing data analysis, aiming to improve the image quality and eliminate the influence of various environmental factors for more accurate subsequent analysis. The following is a detailed description of each preprocessing step: 1) Atmospheric correction Atmospheric correction is to eliminate the influence of the atmosphere on remote sensing images, mainly the influence of atmospheric scattering, absorption, and water vapor and dust in the atmosphere on electromagnetic waves. The goal of atmospheric correction is to restore the ground reflectance of remote sensing data from the images, so that the image data can reflect the true reflectance characteristics of ground objects.

[0032] The specific steps include: Step 1: Obtain the radiance value (DN value) of the remote sensing image; Step 2: Calculate the atmospheric radiation using the atmospheric model and the observation conditions of the remote sensing image (such as the bands of the sensor, imaging angle, etc.); Step 3: Remove the atmospheric effects through the radiative transfer model and atmospheric parameters (such as aerosols, gas content); Step 4: Restore to the surface reflectance.

[0033] 2) Cloud and fog removal Clouds and haze have a great interference on remote sensing images, affecting the extraction of ground object information. The purpose of cloud and fog removal is to identify and remove the cloud and haze areas in the image.

[0034] The specific steps include: Step 1: Detect clouds and fog according to spectral characteristics, applying thresholds or classification algorithms; Step 2: Replace the cloud and fog areas with pixel values and fill the areas blocked by clouds and fog through interpolation.

[0035] 3) Flare correction The flare phenomenon usually occurs when sunlight shines on the water surface or smooth surface, resulting in high-light reflection and affecting the image quality. The goal of flare correction is to eliminate or reduce the flare phenomenon caused by solar reflection.

[0036] The specific steps include: Step 1: Mark the flare area in the image and identify it through the area with higher brightness; Step 2: Use the spectral information of the image, such as the short-wave infrared band, to evaluate the impact of the flare; Step 3: Apply the de-flare algorithm to adjust the reflectance of the image.

[0037] 4) Water body identification Water body identification mainly identifies the water body area through the reflection characteristics of different bands in the remote sensing image. Water bodies usually show lower reflectance in the visible light and near-infrared bands, while showing higher reflectance in the short-wave infrared band.

[0038] The specific steps include: Step 1: Calculate the NDWI or MNDWI value; Normalized Difference Water Index (NDWI): Using the green band (visible light) and the near-infrared band, areas with larger NDWI values usually indicate water bodies. The formula is:

[0039] In the formula, is the Normalized Difference Water Index; is the reflectance of the green band; is the near-infrared band.

[0040] The Modified Normalized Difference Water Index (MNDWI): replacing the near-infrared band with the short-wave infrared band can improve the accuracy of water body identification.

[0041]

[0042] In the formula, MNDWI is the Modified Normalized Difference Water Index, Green is the reflectance of the green light band; SWIR is the reflectance of the short-wave infrared band.

[0043] Step 2: Identify the water body area according to the preset threshold.

[0044] Step 3: Post-process the identified water body area to correct the wrong water body identification results.

[0045] 5) Leaving-water Reflectance Correction Leaving-water Reflectance Correction is to eliminate the influence of reflectance in non-water body areas, so that all ground objects in the image can more truly reflect their spectral characteristics.

[0046] The specific steps include: Step 1: Use different bands in the remote sensing image for spectral feature analysis; Step 2: Adjust the reflectance of non-water body areas based on the reflectance difference between water bodies and non-water bodies.

[0047] Step 3: Correct the image against the ground measured data (such as ground reflectance).

[0048] S3: Quantize the colors in the remote sensing image using the HSV (hue, saturation, value) color space model.

[0049] Quantizing the colors in the remote sensing image using the HSV color space model involves the extraction and conversion of color information. The purpose is to simplify the image color information through quantization, making subsequent processing and analysis more convenient, especially useful in remote sensing applications such as classification, feature extraction, and pattern recognition.

[0050]

[0051] The quantization rules for hue H can be divided into multiple intervals. According to the value range of H, it is divided into several segments, and each segment corresponds to a specific color. In the formula, a, b, c, d, e, f, g are the boundaries of the predetermined hue intervals. These intervals divide the hue H values into several parts, and each part represents a different color range. For example, 0 to 30 may represent red, and 30 to 90 represents yellow, etc.

[0052]

[0053] The quantization of saturation S can be divided into multiple intervals. According to the value range of H, it is divided into several segments, and each segment corresponds to a specific saturation. In the formula, h, i, j are the preset thresholds used to quantify different degrees of saturation. Low saturation means the color is close to gray, and high saturation means the color is more pure and vivid.

[0054]

[0055] The quantization of brightness V can be divided into multiple intervals. According to the value range of H, it is divided into several segments, and each segment corresponds to a specific brightness. In the formula, k, l, m are the quantization boundaries of brightness, which control the light and dark degree of the image. Low brightness represents dark colors, and higher brightness represents bright colors.

[0056] This quantization process classifies and processes each pixel in the image through the three parameters of HSV (hue, saturation, and brightness). For remote sensing images, performing HSV quantization can help reduce the complexity of the image, thereby improving the analysis and processing efficiency. By quantifying and segmenting the hue, saturation, and brightness, specific color information can be extracted according to needs, so as to perform further analysis such as image classification and feature extraction.

[0057] S4: Select three types of water bodies: oligotrophic, mesotrophic, and eutrophic. Construct a standard template for each type of water body, and use the HSV space model to quantify the color, form a one-dimensional vector, and obtain the relevant color histogram. Use the corresponding frequency in the color histogram as the component of the color feature index of the image to perform similarity measurement with the standard template to obtain color feature data.

[0058] S41: Select three types of water bodies: oligotrophic, mesotrophic, and eutrophic, and obtain the image data of oligotrophic, mesotrophic, and eutrophic water bodies; S42: Convert the image data of oligotrophic, mesotrophic, and eutrophic water bodies into the HSV space; S43: Quantify each component in the HSV space to obtain a result where each component is discretized. Represent the HSV color information of each water body as a one-dimensional vector, that is, the color feature vector; Among them, three color components are synthesized into a one-dimensional feature component:

[0059] Among them, l The value range of is discrete integer values, and the value range is from 0 to 71, with a total of 72 possible values; Is the product of the hue component and the saturation, reflecting the comprehensive characteristics of the color; Is the saturation of a specific value; Is the product of the saturation and the lightness, further characterizing the color intensity of the water body; Is the lightness, indicating the brightness of the color; S44: Calculate the color histogram through the HSV color information obtained after quantization. The histogram represents the frequency of occurrence of each color interval and reflects the color distribution in the image.

[0060] For each water body type, calculate the frequency distribution of its HSV color components and construct a histogram.

[0061] S45: Evaluate the similarity between different water body types by calculating the color feature vectors of each water body type through a similarity measurement method to obtain color feature data.

[0062] S5: Remove the feature components with small frequencies and frequencies of 0, arrange the frequencies from large to small, and use the K-Means classifier for water body type classification.

[0063] S51: Data preprocessing S511: Remove features with small frequencies and frequencies of 0 Calculate the frequency of each feature: First, count the values of each color feature and calculate their frequencies in the dataset.

[0064] Remove features with a frequency of 0: If a color feature does not appear in all samples, the frequency is 0, and this feature can be directly removed because it has no information.

[0065] Remove features with small frequencies: A threshold can be set to remove color features with frequencies less than the threshold. For example, if the frequency of a certain category of a color feature is less than 5 times, this category of the color feature may not be sufficient to represent the changes in water body types and can thus be removed to form a color feature database.

[0066] S512: Feature standardization For K-Means, features of different scales will affect the results, so the data needs to be standardized. Commonly used standardization methods include Z-score standardization or Min-Max standardization.

[0067]

[0068] Among them, is the mean of each feature; is the standard deviation of each feature. After standardization, the mean of all features is 0 and the standard deviation is 1, ensuring that they are on the same scale and avoiding the influence of feature scale on distance calculation in K-Means.

[0069] S52: Feature Sorting S521: Sorting According to Frequency Count the frequency of each feature in the color feature database, and then sort them from largest to smallest according to the frequency. Identify those features that may have more information for classification.

[0070] Select important features: Features with larger frequencies may contain more sample information, which is more beneficial for K-Means clustering. According to the sorting results, the first few features with larger frequencies can be selected for classification.

[0071] S53: K-Means Clustering K-Means is a commonly used unsupervised learning algorithm, aiming to divide the data into K clusters, making the samples within each cluster as similar as possible, while the samples between different clusters are quite different.

[0072] S531: Initialize Cluster Centers: Randomly select K initial cluster centers (intelligent initialization methods such as K-Means++ can be selected).

[0073] S532: Assign Data Points to Clusters: According to the distance metric (usually using Euclidean distance), assign each data point to the nearest cluster center.

[0074] S533: Update Cluster Centers: Calculate the mean of all data points within each cluster and update the cluster center to this mean.

[0075] S534: Repeat Steps S532 and S533: Until the cluster centers no longer change or reach the preset maximum number of iterations.

[0076] When selecting the value of K, the "elbow method" or other methods (such as silhouette coefficient) can be used. By plotting the sum of squared errors (SSE) curve for different values of K, select the value of K at the inflection point of the growth rate of SSW.

[0077] S54: Model Training and Evaluation S541: Train the K-Means Model Use the training set to perform K-Means clustering training to obtain K clusters.

[0078] Model training: Use the K-Means clustering algorithm to train the processed data to obtain the division of each cluster.

[0079] S542: Classification results Clustering labels: Each data point will be assigned a clustering label indicating the water body type (cluster) it belongs to. These labels can be used as prediction results.

[0080] S543: Evaluate the model Evaluation of clustering results: Since K-Means is an unsupervised learning method, evaluation is usually based on the compactness within clusters and the separation between clusters. Common evaluation metrics include: Silhouette Coefficient: Measures the compactness and separation of samples assigned to each cluster.

[0081] Within-Cluster Sum of Squares (WSS): Measures the total distance between samples within a cluster and the cluster center, and the smaller the better.

[0082] Through these evaluation methods, it can be determined whether the model is effective and further analyze the clustering results.

[0083] S55: Post-processing and interpretation S551: Analyze the clustering results Interpret the center of each cluster (the characteristics of the water body type represented by the cluster) to infer different water body types based on different characteristics of the water bodies.

[0084] Based on the clustering labels and the characteristics of the water body types, further analyze the water body categories represented by each cluster.

[0085] S552: Visualization For high-dimensional data, the K-Means clustering results can be visualized through dimensionality reduction techniques (such as PCA or t-SNE) to more intuitively understand the clustering effect.

[0086] In summary, by collecting the space-ground synchronous measured data of the water area to be classified, the present invention ensures the timeliness and accuracy of the data, laying a solid foundation for subsequent analysis. Compared with the traditional method that only relies on a single data source, the space-ground synchronous measured data can more comprehensively reflect the spatio-temporal changes of water body optical properties, effectively avoiding classification errors caused by insufficient data or data deviation. Moreover, by using remote sensing technology for preprocessing, not only the speed and efficiency of data processing are greatly improved, but also a detailed and accurate color feature parameter library is constructed by extracting the color feature parameters of water body samples. Overcoming the limitations of feature extraction in the traditional method, the classification model can capture more subtle and key water body optical features, thus significantly improving the classification fineness and accuracy. Secondly, by inputting the color feature parameter library into the color mode classifier for classification analysis, the advantages of computer vision technology are utilized to achieve fast and accurate processing of a large amount of data. The color mode classifier can automatically identify and distinguish different types of water body optical features based on advanced machine learning algorithms such as deep learning and support vector machines, avoiding the subjectivity and error accumulation problems of manual classification. At the same time, by continuously optimizing the classifier parameters and model structure, the universality and generalization ability of the classifier are ensured, enabling it to be applicable to a variety of complex and changing inland water body environments, effectively solving the problem of poor model universality. Furthermore, the step of judging whether the classification data meets the set conditions introduced in the present invention further enhances the reliability and practicality of the classification results. By setting reasonable classification thresholds and verification rules, the classification results can be strictly screened and verified to ensure that only the classification results that meet the conditions are output. This step not only improves the classification accuracy, but also makes the classification results more in line with the actual application requirements, providing strong technical support for water quality monitoring, environmental protection, water resource management, etc.

[0087] As Figure 3 shown, the second object of the present invention is to propose an inland water body optical classification system based on computer vision technology, including: Preprocessing data module 101: used to obtain multi-phase remote sensing data and preprocess the multi-phase remote sensing data to obtain preprocessed remote sensing image data; Quantization processing data module 201: used to perform quantization processing on the preprocessed remote sensing image data to obtain quantized remote sensing image data; Extract color feature module 301: used to extract the color features of water body types based on the quantized remote sensing image data; Obtain feature component module 401: used to process the extracted color features of water body types to obtain feature components; Obtain classification result module 501: used to process the feature components by using a classifier to obtain the classification results of water body types.

[0088] As shown Figure 4 in the figure, the third object of the present invention is to provide an electronic device, which includes: a processor 601, a memory 602, and a display screen 603. Among them, the memory 602 and the display screen 603 are both connected to the processor 601, such as being connected through a bus 604. Optionally, the electronic device may further include a transceiver 605. It should be noted that in practical applications, the transceiver 605 is not limited to one, and the structure of the electronic device does not constitute a limitation on the embodiments of the present application.

[0089] The processor 601 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute various exemplary logic blocks, modules, and circuits described in combination with the disclosure of the present application. The processor 601 may also be a combination that implements a computing function, such as a combination including one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0090] The bus 604 may include a path for transmitting information between the above components. The bus 604 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 604 may be divided into an address bus, a data bus, a control bus, etc.

[0091] The memory 602 can be a ROM (Read Only Memory), or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory), or other types of dynamic storage devices that can store information and instructions. It can also be an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory), or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media, or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0092] The memory 602 is used to store the application program code for implementing the solution of this application and is controlled by the processor 601 for execution. The processor 601 is used to execute the application program code stored in the memory 602 to implement the content shown in the foregoing method embodiments.

[0093] Figure 4 The illustrated electronic device is merely an example and should not impose any limitations on the functions and usage scope of the embodiments of this application.

[0094] The fourth objective of the present invention is to provide a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the program is executed by a processor, it implements each process of the method embodiment as described above. Figure 1 For example, a memory including instructions, and the above instructions can be executed by the processor of the electronic device to complete the above method.

[0095] A computer-readable storage medium can be a tangible device that holds and stores instructions used by an instruction execution device. A computer-readable storage medium can be, but is not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any combination of the above. Specifically, a computer-readable storage medium can be a portable computer disk, a hard disk, a USB flash drive, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disc (DVD), a memory stick, a floppy disk, an optical disc, a magnetic disk, a mechanical encoding device, and any combination of the above.

[0096] The fifth object of the present invention is to provide a computer program product, including computer instructions, which implement the various processes of the method embodiments shown above when executed by a processor, and can achieve the same technical effects. To avoid repetition, it will not be elaborated here. Figure 1 The various processes of the method embodiments shown above, and can achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0097] Upon reading the above description, many embodiments and many applications beyond the provided examples will be obvious to those skilled in the art. Therefore, the scope of this teaching should not be determined with reference to the above description, but rather should be determined with reference to the full scope of the foregoing claims and the equivalents thereof. For the sake of completeness, all articles and references, including patent applications and publications, are incorporated herein by reference. The omission of any aspect of the subject matter disclosed herein in the foregoing claims is not intended to abandon such subject matter, nor should it be considered that the applicant has not considered such subject matter to be part of the disclosed inventive subject matter.

[0098] The above is a further detailed description of the present invention. It cannot be determined that the specific embodiments of the present invention are limited thereto. For those of ordinary skill in the technical field to which the present invention pertains, without departing from the concept of the present invention, several simple deductions or substitutions can be made, which should all be regarded as falling within the protection scope determined by the claims submitted for the present invention.

Claims

1. An optical classification method for inland water bodies based on computer vision technology, characterized in that, Including: Obtain multi - period remote sensing data, and pre - process the multi - period remote sensing data to obtain pre - processed remote sensing image data; Perform quantization processing on the pre - processed remote sensing image data to obtain quantized remote sensing image data; Based on the quantized remote sensing image data, extract the color features of the water body type; Process the extracted color features of the water body type to obtain feature components; Use a classifier to process the feature components to obtain the classification result of the water body type.

2. The optical classification method for inland water bodies based on computer vision technology according to claim 1, characterized in that The obtaining of multi - period remote sensing data and the pre - processing of the multi - period remote sensing data to obtain pre - processed remote sensing image data include: Use remote sensing data to obtain water body color parameters and multi - period remote sensing data; Perform atmospheric correction, cloud removal, flare correction, water body identification, and water - leaving reflectance correction on the multi - period remote sensing data to obtain pre - processed remote sensing image data.

3. The optical classification method for inland water bodies based on computer vision technology according to claim 1, characterized in that The performing of quantization processing on the pre - processed remote sensing image data to obtain quantized remote sensing image data includes: Use the HSV color space model to perform quantization processing on the pre - processed remote sensing image data; Obtain quantized remote sensing image data; The HSV color space model includes: In the formula, H is hue; a, b, c, d, e, f, g are the boundaries of the predetermined hue interval; In the formula, S is saturation; h, i, j are preset thresholds; In the formula, V is brightness; k, l, m are the quantization boundaries of brightness.

4. A method for optical classification of inland water bodies based on computer vision technology according to claim 1, characterized in that, The extracting of the color features of the water body type based on the quantized remote sensing image data includes: Select three types of water bodies: oligotrophic, mesotrophic, and eutrophic. Based on the quantized remote sensing image data, construct standard templates for the oligotrophic, mesotrophic, and eutrophic water bodies; Convert the image data of the oligotrophic, mesotrophic, and eutrophic water bodies into the HSV space; Quantize each component in the HSV space to obtain a discretized result for each component, and represent the HSV color information of each water body as a one - dimensional vector, that is, a color feature vector; Calculate the color histogram through the quantized HSV color information; Use the corresponding frequency occurrences in the color histogram as components of the color feature index of the image to perform similarity measurement with the template to obtain color feature data.

5. A method for optical classification of inland water bodies based on computer vision technology according to claim 1, characterized in that, The processing of the extracted color features of the water body type to obtain feature components includes: The processing of the extracted color features of the water body type to obtain feature components includes: Count each value in the extracted color features of the water body type to obtain the frequency of each value in the color feature dataset; Remove the color features with a frequency of 0 and the color features with a frequency less than the set threshold to obtain the removed feature components.

6. The optical classification method for inland water bodies based on computer vision technology according to claim 1, characterized in that, The using of a classifier to process the feature components to obtain the classification result of the water body type includes: Based on the removed feature components, construct a color feature database; Input the constructed color feature database into the K - Means classifier for water body classification to obtain the classification result.

7. An optical classification system for inland waters based on computer vision technology, characterized in that, Including: Pre - processing data module: used to obtain multi - period remote sensing data and pre - process the multi - period remote sensing data to obtain pre - processed remote sensing image data; Quantization processing data module: used to perform quantization processing on the preprocessed remote sensing image data to obtain the quantized remote sensing image data; Color feature extraction module: used to extract the color features of the water body type based on the quantized remote sensing image data; Feature component obtaining module: used to process the extracted color features of the water body type to obtain feature components; Classification result acquisition module: used to process the feature components by using a classifier to obtain the classification result of the water body type.

8. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the method for inland water body optical classification based on computer vision technology according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, it implements the steps of the method for inland water body optical classification based on computer vision technology according to any one of claims 1-6.

10. A computer program product, characterized in that, It includes computer instructions. When the computer instructions are executed by a processor, it implements the steps of the method for inland water body optical classification based on computer vision technology according to any one of claims 1-6.

Images