Rural style and feature intelligent analysis and evaluation system based on multi-dimensional features

Abstract

The invention discloses a multi-dimensional feature-based rural style and appearance intelligent analysis and evaluation system, and relates to the technical field of rural style and appearance analysis and evaluation. Comprising a data acquisition module which is responsible for collecting high-resolution remote sensing images, ground live photos and geographic information data of a rural area; the preprocessing module is used for denoising, correcting and standardizing the collected data; the feature extraction module is used for extracting multi-dimensional features from the image by adopting a deep learning network; the style and appearance classification module is used for classifying the rural style and appearance according to the extracted features by utilizing a classifier; and the quality evaluation module is used for evaluating the quality of the village style and appearance in combination with an expert system and fuzzy logic. The system provided by the invention can process and analyze high-resolution remote sensing images, ground live photos and geographic information data, and provides a comprehensive data view angle; automatic classification and quality evaluation of rural styles and features are realized by using machine learning technologies such as deep learning, a support vector machine and a random forest.

Description

Technical Field

[0001] The present invention relates to the technical field of rural landscape analysis and evaluation, and in particular to a rural landscape intelligent analysis and evaluation system based on multi-dimensional features. Background Art

[0002] Traditional landscape assessment methods often rely on manual surveys and expert evaluations, which is not only time-consuming and labor-intensive, but also difficult to achieve rapid assessments over large areas. With the rapid development of artificial intelligence and machine learning technologies, their applications in image recognition, data analysis, and predictive model building are becoming increasingly widespread. Therefore, how to use this technology to develop an intelligent system that can automatically, efficiently, and accurately assess rural landscape is a problem that needs to be considered.

[0003] After searching, the application scheme of Chinese patent application No. 202010418816.1 discloses a rural environmental quality comprehensive assessment and supervision system and method. The rural environmental quality comprehensive assessment and supervision system includes a communication network layer, a data center layer, an application support layer, an application service layer, and an application presentation layer. The rural environmental quality comprehensive assessment and supervision method includes the following steps: (1) obtaining the ecological environment quality parameter value of the target area; (2) normalizing the data; (3) calculating the rural environmental quality comprehensive index value; (4) grading the environmental quality of the target area according to the rural environmental quality comprehensive index value, analyzing the reasons for the excessive environmental quality parameters, and diagnosing environmental problems; (5) the system outputs the corresponding environmental supervision plan based on the diagnosed environmental problems. The comprehensive assessment and supervision system in the above patent has the following shortcomings: lack of ability to extract and analyze deep-level features of image and text data, and still lacks in data processing efficiency and intelligence level. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a rural landscape intelligent analysis and evaluation system based on multi-dimensional features.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A rural landscape intelligent analysis and evaluation system based on multi-dimensional features, including:

[0007] Data collection module: responsible for collecting high-resolution remote sensing images, ground truth photos and geographic information data of rural areas;

[0008] Preprocessing module: denoising, correction and standardization of collected data;

[0009] Feature extraction module: uses deep learning network to extract multi-dimensional features from images;

[0010] Landscape classification module: Use the classifier to classify the rural landscape according to the extracted features;

[0011] Quality assessment module: Combines expert system and fuzzy logic to assess the quality of rural landscape and outputs suggestions for landscape protection and development;

[0012] Time series analysis module: Use Python's statsmodels library to perform time series analysis, and predict future trends based on long-term series data on rural landscape changes; identify the periodicity, trend, and seasonal characteristics of landscape changes, and provide a basis for long-term planning of rural landscape;

[0013] Deep learning optimization module: By integrating the deep learning framework of TensorFlow or PyTorch, the extracted multi-dimensional features can be learned and understood at a deeper level; the image features can be extracted and analyzed again using a deep convolutional neural network to achieve higher classification and evaluation accuracy;

[0014] Natural language processing module: Introducing the NLTK natural language processing tool to analyze and process the collected descriptive text data about rural landscapes; including text classification, sentiment analysis, and keyword extraction tasks;

[0015] Reinforcement learning module: Introducing reinforcement learning algorithms to help the system optimize evaluation strategies through trial and error, thereby achieving better results in future evaluations;

[0016] Visual display module: output the analysis results visually.

[0017] Preferably: the data acquisition module comprises:

[0018] Satellite image acquisition unit: responsible for acquiring the latest satellite remote sensing data, including high-resolution panchromatic and multispectral images;

[0019] UAV aerial photography unit: Use drones to carry out aerial photography missions in specific areas and collect high-definition ground truth photos;

[0020] Ground measurement data unit: collects specific indicator data of terrain and vegetation coverage through ground measurement equipment;

[0021] Historical Data Arrangement Unit: Collect and organize historical materials and data, including old maps, historical images and documentary records.

[0022] Preferably: the preprocessing module comprises:

[0023] Image correction unit: performs radiation correction and geometric correction on the collected images to ensure the accuracy of the data;

[0024] Data fusion unit: fuses multi-source data to improve the effectiveness and accuracy of data application;

[0025] Noise removal unit: uses filtering technology to remove noise from images and signals;

[0026] Data standardization unit: Standardize all data to facilitate subsequent analysis and model training.

[0027] Preferably: the feature extraction module comprises:

[0028] Color feature extraction unit: analyze the color composition and distribution characteristics of rural landscape;

[0029] Texture feature extraction unit: identifies and calculates the texture features of various surface covers;

[0030] Shape feature extraction unit: extracts the shape and edge features of geographic objects such as buildings and water bodies;

[0031] Spatial relationship feature extraction unit: Analyze the spatial relationship between different geographical elements.

[0032] Preferably: the style classification module includes:

[0033] Supervised learning classification unit: Use labeled samples to train classification models;

[0034] Unsupervised learning classification unit: Use unsupervised methods to perform preliminary identification of landscape areas;

[0035] Deep learning classification unit: Use deep neural networks to learn and classify complex features;

[0036] Model evaluation unit: Evaluate the performance of the classification model through precision and recall indicators, and perform optimization.

[0037] Preferably, the style classification module performs preliminary identification of style areas based on the K-means algorithm, and the specific process is as follows:

[0038] S1: Initialization, randomly select K data points as the initial cluster centers;

[0039] S2: Assignment, for each data point in the data set, calculate its distance from each cluster center and assign it to the cluster center closest to it; use Euclidean distance as the distance metric, the specific formula is:

[0040]

[0041] Where x is the data point, c i is the i-th cluster center, d is the dimension of the data, x j and cij are x and c respectively i The value in the i-th dimension;

[0042] S3: For each cluster, recalculate its cluster center; the new cluster center is the mean of all data points in the cluster. The specific formula is:

[0043]

[0044] Among them, S i is the set of data points of the ith cluster, |S i | is the number of data points in the set;

[0045] S4: Iteration: Repeat the assignment and update steps until the termination condition is met.

[0046] Preferably: the quality assessment module comprises:

[0047] Landscape scoring unit: quantitatively scoring the rural landscape based on the extracted features and classification results;

[0048] Development Proposal Generation Unit: Combine the scores and regional characteristics to propose specific protection and development suggestions;

[0049] Risk assessment unit: Identify possible factors and risks of landscape destruction and provide a basis for risk management;

[0050] Expert system integration unit: Integrate expert knowledge to improve the professionalism and accuracy of assessment.

[0051] Preferably: the time series analysis module comprises:

[0052] Periodic analysis unit: Identify the periodic change characteristics of rural landscape in the time dimension;

[0053] Trend forecasting unit: predict future trends based on historical data to provide reference for long-term planning;

[0054] Y t =μ+φ1Y t-1 +∈ t

[0055] Among them, Y t represents the observed value of the time series at time t, μ is the mean, φ1 is the autoregressive coefficient, ∈ t is the error term.

[0056] Preferably: the deep learning optimization module comprises:

[0057] Network structure adjustment unit: adjusts the structure of the deep learning network according to data characteristics and task requirements;

[0058] Hyperparameter optimization unit: Use grid search and random search methods to optimize hyperparameters;

[0059] The operation of the deep learning optimization module on the convolutional layer is expressed as follows:

[0060]

[0061] in, is the current layer output, σ is the activation function, b l is the bias term, is the convolution kernel weight, and * represents the convolution operation.

[0062] Preferably: the natural language processing module comprises:

[0063] Text parsing unit: Analyze and process descriptive texts about rural landscape and extract key information;

[0064] Sentiment Analysis Unit: Evaluate the sentiment in the text and understand the public’s attitude and feelings towards rural landscape;

[0065] Keyword extraction unit: extract keywords from descriptive text to assist in feature understanding and style classification.

[0066] The beneficial effects of the present invention are:

[0067] 1. The system of the present invention can process and analyze high-resolution remote sensing images, ground-truth photos and geographic information data, providing a comprehensive data perspective; by utilizing machine learning technologies such as deep learning, support vector machines, and random forests, it can achieve automatic classification and quality assessment of rural landscapes.

[0068] 2. The present invention introduces a time series analysis module and uses the statsmodels library to predict future trends in rural landscape and assist in long-term planning; it integrates tools such as NLTK to process descriptive text data and provide richer analysis dimensions.

[0069] 3. The deep learning optimization module of the present invention allows the adjustment of network structure and hyperparameter optimization, thereby improving model performance; the evaluation strategy is continuously optimized using the reinforcement learning algorithm, thereby improving the evaluation effect of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 This is a flow chart for preliminary identification of landscape areas based on the K-means algorithm in the present invention. DETAILED DESCRIPTION

[0071] The technical solution of the present invention is further described in detail below in conjunction with specific implementation methods.

[0072] Embodiment 1:

[0073] A rural landscape intelligent analysis and evaluation system based on multi-dimensional features, including:

[0074] Data collection module: responsible for collecting high-resolution remote sensing images, ground truth photos and geographic information data of rural areas;

[0075] Preprocessing module: denoising, correcting and standardizing the collected data to improve the accuracy of subsequent analysis;

[0076] Feature extraction module: uses deep learning networks, including convolutional neural networks (CNN), to extract multi-dimensional features from images, including but not limited to color features, texture features, shape features, and spatial relationship features;

[0077] Landscape classification module: Use classifiers such as support vector machines or random forests to classify rural landscapes according to the extracted features, including natural landscapes, cultural landscapes, architectural landscapes, etc.

[0078] Quality assessment module: Combines expert system and fuzzy logic to assess the quality of rural landscape and outputs suggestions for landscape protection and development;

[0079] Time series analysis module: Use Python's statsmodels library to perform time series analysis, and predict future trends based on long-term series data on rural landscape changes; identify the periodicity, trend, and seasonal characteristics of landscape changes, and provide a basis for long-term planning of rural landscape;

[0080] Deep learning optimization module: By integrating the deep learning framework of TensorFlow or PyTorch, the extracted multi-dimensional features can be learned and understood at a deeper level; the image features can be extracted and analyzed again using a deep convolutional neural network to achieve higher classification and evaluation accuracy;

[0081] Natural language processing module: Introducing the NLTK natural language processing tool to analyze and process the collected descriptive text data about rural landscape; including tasks such as text classification, sentiment analysis, and keyword extraction, so as to assist image data and provide a more comprehensive landscape assessment;

[0082] Reinforcement learning module: Introducing reinforcement learning algorithms to help the system optimize evaluation strategies through trial and error, thereby achieving better results in future evaluations;

[0083] Visual display module: The analysis results are displayed intuitively in the form of charts, heat maps, etc., which is easy for decision makers to understand and use.

[0084] Wherein, the data acquisition module includes:

[0085] Satellite image acquisition unit: responsible for acquiring the latest satellite remote sensing data, including high-resolution panchromatic and multispectral images;

[0086] Drone aerial photography unit: Arrange drones to carry out aerial photography missions in specific areas and collect high-definition ground-truth photos;

[0087] Ground measurement data unit: collects specific indicator data such as terrain and vegetation coverage through ground measurement equipment;

[0088] Historical Data Arrangement Unit: Collect and organize historical materials and data, including old maps, historical images and documentary records.

[0089] Wherein, the preprocessing module includes:

[0090] Image correction unit: performs radiation correction and geometric correction on the collected images to ensure the accuracy of the data;

[0091] Data fusion unit: fuses multi-source data to improve the effectiveness and accuracy of data application;

[0092] Noise removal unit: uses filtering technology to remove noise from images and signals;

[0093] Data standardization unit: Standardize all data to facilitate subsequent analysis and model training.

[0094] Wherein, the feature extraction module includes:

[0095] Color feature extraction unit: analyze the color composition and distribution characteristics of rural landscape;

[0096] Texture feature extraction unit: identifies and calculates the texture features of various surface covers;

[0097] Shape feature extraction unit: extracts the shape and edge features of geographic objects such as buildings and water bodies;

[0098] Spatial relationship feature extraction unit: analyzes the spatial relationship between different geographical elements, such as distance, direction, etc.

[0099] Wherein, the style classification module includes:

[0100] Supervised learning classification unit: Use labeled samples to train classification models such as SVM and random forest;

[0101] Unsupervised learning classification unit: Use unsupervised methods such as K-means and hierarchical clustering to perform preliminary identification of landscape areas;

[0102] Deep Learning Classification Unit: Uses deep neural networks, such as convolutional neural networks (CNNs), to learn and classify complex features;

[0103] Model evaluation unit: Evaluate the performance of the classification model through indicators such as precision and recall, and perform optimization.

[0104] The style classification module performs preliminary identification of style areas based on the K-means algorithm. The specific process is as follows:

[0105] S1: Initialization, randomly select K data points as the initial cluster centers;

[0106] S2: Assignment, for each data point in the data set, calculate its distance from each cluster center and assign it to the cluster center closest to it; use Euclidean distance as the distance metric, the specific formula is:

[0107]

[0108] Where x is the data point, c i is the i-th cluster center, d is the dimension of the data, x j and c ij are x and c respectively i The value in the jth dimension;

[0109] S3: For each cluster, recalculate its cluster center; the new cluster center is the mean of all data points in the cluster. The specific formula is:

[0110]

[0111] Among them, S i is the set of data points of the ith cluster, |S i | is the number of data points in the set;

[0112] S4: Iteration: Repeat the assignment and update steps until the termination condition is met.

[0113] Wherein, the quality assessment module includes:

[0114] Landscape scoring unit: quantitatively scoring the rural landscape based on the extracted features and classification results;

[0115] Development Proposal Generation Unit: Combine the scores and regional characteristics to propose specific protection and development suggestions;

[0116] Risk assessment unit: Identify possible factors and risks of landscape destruction and provide a basis for risk management;

[0117] Expert system integration unit: Integrate expert knowledge to improve the professionalism and accuracy of assessment.

[0118] Wherein, the time series analysis module includes:

[0119] Periodic analysis unit: Identify the periodic change characteristics of rural landscape in the time dimension;

[0120] Trend forecasting unit: predict future trends based on historical data to provide reference for long-term planning;

[0121] Y t =μ+φ1Y t-1 +∈ t

[0122] Among them, Y t represents the observed value of the time series at time t, μ is the mean, φ1 is the autoregressive coefficient, ∈ t is the error term.

[0123] Wherein, the deep learning optimization module includes:

[0124] Network structure adjustment unit: adjusts the structure of the deep learning network according to data characteristics and task requirements;

[0125] Hyperparameter optimization unit: Use grid search, random search and other methods to optimize hyperparameters;

[0126] The operation of the deep learning optimization module on the convolutional layer is expressed as follows:

[0127]

[0128] in, is the current layer output, σ is the activation function, b l is the bias term, is the convolution kernel weight, and * represents the convolution operation.

[0129] Wherein, the natural language processing module includes:

[0130] Text parsing unit: Analyze and process descriptive texts about rural landscape and extract key information;

[0131] Sentiment Analysis Unit: Evaluate the emotional tendencies in the text and understand the public's attitudes and feelings towards rural landscape;

[0132] Keyword extraction unit: extract keywords from descriptive text to assist in feature understanding and style classification.

[0133] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A rural landscape intelligent analysis and evaluation system based on multi-dimensional features, characterized in that: include: Data collection module: responsible for collecting high-resolution remote sensing images, ground truth photos and geographic information data of rural areas; Preprocessing module: denoising, correction and standardization of collected data; Feature extraction module: uses deep learning network to extract multi-dimensional features from images; Landscape classification module: Use the classifier to classify the rural landscape according to the extracted features; Quality assessment module: Combines expert system and fuzzy logic to assess the quality of rural landscape and outputs suggestions for landscape protection and development; Time series analysis module: Use Python's statsmodels library to perform time series analysis, and predict future trends based on long-term series data on rural landscape changes; identify the periodicity, trend, and seasonal characteristics of landscape changes, and provide a basis for long-term planning of rural landscape; Deep learning optimization module: By integrating the deep learning framework of TensorFlow or PyTorch, the extracted multi-dimensional features can be learned and understood at a deeper level; the image features can be extracted and analyzed again using a deep convolutional neural network to achieve higher classification and evaluation accuracy; Natural language processing module: introduce NLTK natural language processing tools to analyze and process the collected descriptive text data about rural landscape; Including text classification, sentiment analysis, and keyword extraction tasks; Reinforcement learning module: Introducing reinforcement learning algorithms to help the system optimize evaluation strategies through trial and error, thereby achieving better results in future evaluations; Visual display module: output the analysis results visually.

2. According to claim 1, a rural landscape intelligent analysis and evaluation system based on multidimensional features is characterized in that: The data acquisition module comprises: Satellite image acquisition unit: responsible for acquiring the latest satellite remote sensing data, including high-resolution panchromatic and multispectral images; UAV aerial photography unit: Use drones to carry out aerial photography missions in specific areas and collect high-definition ground truth photos; Ground measurement data unit: collects specific indicator data of terrain and vegetation coverage through ground measurement equipment; Historical Data Arrangement Unit: Collect and organize historical materials and data, including old maps, historical images and documentary records.

3. The rural landscape intelligent analysis and evaluation system based on multi-dimensional features according to claim 1 is characterized in that: The preprocessing module comprises: Image correction unit: performs radiation correction and geometric correction on the collected images to ensure the accuracy of the data; Data fusion unit: fuses multi-source data to improve the effectiveness and accuracy of data application; Noise removal unit: uses filtering technology to remove noise from images and signals; Data standardization unit: Standardize all data to facilitate subsequent analysis and model training.

4. The rural landscape intelligent analysis and evaluation system based on multi-dimensional features according to claim 1 is characterized in that: The feature extraction module comprises: Color feature extraction unit: analyze the color composition and distribution characteristics of rural landscape; Texture feature extraction unit: identifies and calculates the texture features of various surface covers; Shape feature extraction unit: extracts the shape and edge features of geographic objects such as buildings and water bodies; Spatial relationship feature extraction unit: Analyze the spatial relationship between different geographical elements.

5. The rural landscape intelligent analysis and evaluation system based on multi-dimensional features according to claim 1 is characterized in that: The style classification module comprises: Supervised learning classification unit: Use labeled samples to train classification models; Unsupervised learning classification unit: Use unsupervised methods to perform preliminary identification of landscape areas; Deep learning classification unit: Use deep neural networks to learn and classify complex features; Model evaluation unit: Evaluate the performance of the classification model through precision and recall indicators, and perform optimization.

6. The rural landscape intelligent analysis and evaluation system based on multi-dimensional features according to claim 5 is characterized in that: The style classification module performs preliminary identification of style areas based on the K-means algorithm. The specific process is as follows: S1: Initialization, randomly select K data points as the initial cluster centers; S2: Assignment, for each data point in the data set, calculate its distance from each cluster center and assign it to the cluster center closest to it; use Euclidean distance as the distance metric, the specific formula is: Where x is the data point, c i is the i-th cluster center, d is the dimension of the data, x j and c ij are x and c respectively i The value in the jth dimension; S3: For each cluster, recalculate its cluster center; the new cluster center is the mean of all data points in the cluster. The specific formula is: Among them, S i is the set of data points of the ith cluster, |S i | is the number of data points in the set; S4: Iteration: Repeat the assignment and update steps until the termination condition is met.

7. The rural landscape intelligent analysis and evaluation system based on multi-dimensional features according to claim 1 is characterized in that: The quality assessment module comprises: Landscape scoring unit: quantitatively scoring the rural landscape based on the extracted features and classification results; Development Proposal Generation Unit: Combine the scores and regional characteristics to propose specific protection and development suggestions; Risk assessment unit: Identify possible factors and risks of landscape destruction and provide a basis for risk management; Expert system integration unit: Integrate expert knowledge to improve the professionalism and accuracy of assessment.

8. The rural landscape intelligent analysis and evaluation system based on multi-dimensional features according to claim 1 is characterized in that: The time series analysis module comprises: Periodic analysis unit: Identify the periodic change characteristics of rural landscape in the time dimension; Trend forecasting unit: predict future trends based on historical data to provide reference for long-term planning; Y t =μ+φ1Y t-1 +∈ t Among them, Y t represents the observed value of the time series at time t, μ is the mean, φ1 is the autoregressive coefficient, ∈ t is the error term.

9. The rural landscape intelligent analysis and evaluation system based on multi-dimensional features according to claim 1 is characterized in that: The deep learning optimization module comprises: Network structure adjustment unit: adjusts the structure of the deep learning network according to data characteristics and task requirements; Hyperparameter optimization unit: Use grid search and random search methods to optimize hyperparameters; The operation of the deep learning optimization module on the convolutional layer is expressed as follows: in, is the current layer output, σ is the activation function, b l is the bias term, is the convolution kernel weight, and * represents the convolution operation.

10. The rural landscape intelligent analysis and evaluation system based on multi-dimensional features according to claim 1 is characterized in that: The natural language processing module comprises: Text parsing unit: Analyze and process descriptive texts about rural landscape and extract key information; Sentiment Analysis Unit: Evaluate the sentiment in the text and understand the public’s attitude and feelings towards rural landscape; Keyword extraction unit: extract keywords from descriptive text to assist in feature understanding and style classification.

Images