39 results about "Vegetation classification" patented technology

The invention discloses an urban building change detection method based on LiDAR point cloud spatial difference analysis. The method comprises the steps that 1, LiDAR point cloud data with different time phases are respectively preprocessed, and DSM and DTM of the different time phases are obtained through a linear interpolation algorithm; 2, two periods of the DSM obtained in the step 1 are subtracted to obtain an elevation difference model; 3, spatial difference analysis is carried out on the elevation difference model, and a candidate change area is extracted; 4, spatial features of a LiDAR point cloud are used for extracting direct features and elevation grain features; 5, building and vegetation classification is achieved; 6, superposition and overlay analysis is carried out on the candidate change area, and the building change attribute is determined. The advantages of the LiDAR data on three-dimensional representation of the building complex structure and the topological relation are fully achieved, and building three-dimensional change detection is carried out.

The invention relates to the field of ecological environment monitoring, and discloses a vegetation classification method based on a machine learning algorithm and multi-source remote sensing data fusion, which is used for efficiently realizing identification and classification of vegetation types in a target area. The method comprises the following steps: acquiring a low-altitude remote sensing image of terrestrial plants in a sample area by using an unmanned aerial vehicle, and acquiring a digital orthoimage and a digital surface model of the sample area based on the low-altitude remote sensing image; extracting elevation information of the digital surface model; acquiring an SAR image of a sample region corresponding to the aerial photography time of the unmanned aerial vehicle by utilizing satellite remote sensing; carrying out wave band and image fusion on the digital orthoimage, the elevation information and the SAR image; performing inversion model training and inversion model precision evaluation on the fused image through sample area actual measurement data and a machine learning algorithm to obtain an inversion model meeting requirements; and finally, classifying terrestrial plants in the target area based on the inversion model. The method is suitable for terrestrial plant ecological environment monitoring.

The invention relates to a method for estimating vegetation coverage degree in diggings based on a close-up photography mode. The method comprises the following steps of: photographing the diggings to be tested, extracting color features and texture features to form a feature vector according to a particular order, fast classifying vegetation and non-vegetation utilizing a support vector machine classifier, then gathering statistics of vegetation information, and then fast and accurately calculating the vegetation coverage degree for providing reliable basis for surveying the growth status of the vegetation in the diggings. The method also can be used as actual verification of quantitative remote sensing estimation of vegetation coverage algorithm at the same time. The invention also relates to a system for estimating the vegetation coverage degree in the diggings based on the close-up photography mode. The system comprises a feature extraction module, a classifier training module, a vegetation and non-vegetation classifying module and a vegetation coverage degree calculating module.

The invention brings forward an airborne laser radar data vegetation extraction method. The method comprises the following steps: 1, preprocessing of an airborne laser radar point cloud; 2, pre-segmentation of the airborne laser radar point cloud; 3, feature extraction of a segmentation unit; 4, soft interval SVM classification based on a kernel function; and 5, optimization of data vegetation classification results on the basis of prior knowledge. The method has the following advantages: 1, the method does not need fusion of other data sources such as multispectral images, high-spectral images and the like, thereby having quite high universality; 2, the method guarantees space self-correlation of the laser radar point cloud, effectively prevents a classification algorithm from damaging such a space attribute, ensures separability of vegetation and non-vegetation points and improves the vegetation detectability; and 3, the algorithm can effectively separate building circumference and stereo wall face points, building roof irregular object points and ground points of a vegetation-dense area from vegetation and realizes the purpose of accurately extracting the vegetation in an urban area.

The invention discloses a vegetation classification method capable of automatically selecting features in power transmission line corridor area. The method includes the following steps: a first step of extracting features of training samples and cross validation samples, a plurality of types of features of all training samples constituting a feature set; a second step of, based on the training samples and cross validation samples, adopting a cross validation method to conduct optimization selection on features so as to acquire optimized features; a third step of adopting the optimized features to conduct vegetation classification on remote sensing image test data. According to the invention, the method does not require a large amount of iterative computation in feature selection phase, and a combination of features obtained after optimization selection, have robustness. The combination of features serves to classify vegetation in remote sensing image power transmission line corridor area, and can explicitly increase algorithm computation efficiency and classification precision.

The invention provides a method for extracting a phytocoenosium spatial structure. The method comprises: performing multi-resolution segmentation of a to-be-tested remote-sensing image in a target area to obtain remote-sensing image objects with different resolutions; establishing a corresponding relation between an image resolution of the to-be-tested remote-sensing image and an ecological organization resolution to obtain an image resolution of each plant type in the to-be-tested remote-sensing image, wherein the plant types include a meadow, a shrub, an arbor, a population and a group, wherein the meadow, the shrub, and the arbor are plant individuals; performing vegetation classification of a pre-selected sample of the to-be-tested remote-sensing image in plant individual and population image resolution according to the plant individuals and the population image resolution; summing the classification result of each resolution to a grouped data layer; and calculating plant individuals and parameters of a population spatial structure in a group resolution object boundary. The method for extracting the phytocoenosium spatial structure is relatively accurate, and is low in monitoring cost and high in objectivity.

The invention discloses a method for pre-warning forest fire based on satellite remote sensing data, and solves the problem that existing forest fire danger forecasting is insufficient in precision. The method comprises the following steps: obtaining image data, and carrying out radiation correction, geometric correction and projection transformation; using processed image data to calculate surface temperature; using processed image data to calculate normalized differential vegetation index so as to calculate vegetation coverage rate; obtaining a panchromatic image and a multispectral image, carrying out geometric correction on the panchromatic image and the multispectral image, fusing the panchromatic image and the multispectral image for vegetation classification, constructing combustible moisture content models of different types of vegetations in combination with meteorological factors, and calculating the moisture contents of different types of vegetations; calculating dynamic hazard indexes of forest fire through the surface temperature, the vegetation coverage rate and the moisture contents of different types of vegetations, and classifying fire danger classes according to the dynamic hazard indexes of forest fire. According to the method, the precision of a combustible moisture content model is much higher than that of a model determined through remote sensing satelliteparameters directly.

The invention relates to the field of remote-sensing data fusion and intelligent application, namely a multi-source sensor remote-sensing fusion is used for realizing fine extraction of the land vegetation coverage, more specifically, the invention relates to a vegetation classification and identification method of hyperspectral image and laser radar point cloud fusion. For the difficulties in theland vegetation coverage classification and identification, a combined classification method for fusing the hyperspectral image and the laser radar point cloud data features is adopted, and the optimal selection and combination are performed on the space-spectrum feature influencing the classification precision, thereby improving the vegetation coverage classification and identification precision, and increasing the fineness of the vegetation coverage drafting.

The invention discloses a hyperspectral image vegetation classification method for spatial feature adaptive optimization. The method comprises the steps that first, normalization processing is performed on a hyperspectral image, PCA dimension reduction is performed on the hyperspectral image obtained after normalization processing, morphological filtering is performed on the hyperspectral image obtained after PCA dimension reduction, and a spatial information set is acquired; second, domain transformation standard convolution filtering is performed on the hyperspectral image obtained after normalization processing for spatial information extraction, and spatial information linear superposition fusion is performed on the acquired spatial information set; and finally the spatial informationset obtained after fusion is classified. Through the method, spatial texture information and relevancy information are effectively utilized, filtering parameters of a best classified structural type is further optimized, a good filtering effect is achieved, and classification precision of the hyperspectral image is effectively improved.

The invention discloses a type prediction method of vegetation in a high-altitude permafrost region, and relates to the field of geography. The method comprises the steps that the survey data of characteristics of the vegetation in the permafrost region of Qinghai-Tibet Plateau is obtained; bio-climatic parameters are obtained; according to an NDVI data set, NDVI parameters are obtained; according to a digital elevation model (DEM), the slope, the slope direction and the profile curvature at each grid pixel element point in the permafrost region of Qinghai-Tibet Plateau are obtained; the elevations, the slopes, the slope directions and the profile curvatures are used as topography parameters; the parameters with the correlation coefficients larger than 0.8 are selected in the bio-climatic parameters, the NDVI parameters and the topography parameters through a principal component analysis method, and vegetation classification parameters are obtained; according to the survey data of the characteristics of the vegetation, the vegetation classification parameters, climate scenarios data and a climatic system mode, the types of the vegetation in the permafrost region of Qinghai-Tibet Plateau are obtained through a decision tree classification method. By means of the method, the type distribution prediction, in the mode of ten types of ten categories of climatic systems and four types of climate scenarios in 2050 and 2070, of the vegetation in the permafrost region of Qinghai-Tibet Plateau can be achieved.

The invention relates to the field of ecological environment monitoring, and discloses an arbor biomass measuring and calculating method based on an unmanned aerial vehicle hyperspectrum and a machinelearning algorithm, which is used for better realizing biomass monitoring of arbor of a target area type. The method comprises the following steps: acquiring a hyperspectral image of terrestrial plants in a target area by using an unmanned aerial vehicle, modeling based on the hyperspectral image, and extracting elevation information of a digital surface model; extracting spectral information from the original image photos, and performing quantitative inversion model training by adopting a machine learning algorithm according to types of terrestrial plant ecological environment monitoring vegetation classifications in combination with high-level information, characteristic wave bands and vegetation indexes of various plants in the target area to obtain an inversion model; classifying thevegetation types of the target area by using an inversion model so as to extract arbor classification data; and finally, calculating to obtain the arbor biomass by utilizing the extracted arbor classification data and combining with an aboveground biomass formula. The invention is suitable for arbor biomass measurement and calculation.

The invention discloses a method for carrying out aquatic vegetation annual change statistics by utilizing an unmanned aerial vehicle and a multispectral satellite image. The method comprises the following steps: acquiring multispectral satellite images of an investigation area, preprocessing the images, processing pictures acquired by the unmanned aerial vehicle to obtain an orthoimage, and registering the orthoimage to the preprocessed multispectral satellite image; carrying out aquatic vegetation classification by using the multispectral image and the registered orthoimage; marking aquaticvegetation types in the investigation area on the multispectral satellite images; acquiring satellite images of the same technical specification in the same time period of the second year, and preprocessing the satellite images; registering the multispectral satellite images obtained in the second year with the multispectral satellite images obtained in the first year; comparing the difference between the samples obtained by each aquatic vegetation marking point, and updating a marking point sample library according to the difference; and repeatedly collecting every year to obtain a statistical result of the annual change conditions of the aquatic vegetation.

The invention discloses an unmanned aerial vehicle side slope vegetation classification method based on plant height. The method comprises the following steps: (1) enabling an unmanned aerial vehicleto fly along a plurality of fixed-height equally-divided interval airlines in a snake-shaped manner and suspend at sampling points on the fixed-height equally-divided interval airlines to acquire images; (2) importing the image into a Pix4DMapper and synthesizing the image into high-density point cloud data; obtaining a DSM and an ortho-image according to high-density point cloud data, importing the DSM and the ortho-image into an ArcGIS, manually selecting a ground point in the ArcGIS according to the ortho-image, extracting the elevation of the ground point from the DSM, and generating a DTMby using an inverse distance weight interpolation method, wherein nDSM=DSM-DTM, nDSM is the height of the plant, the DSM is a digital surface model and the DTM is a digital terrain model; (3) constructing a sample training manager; (4) applying an ecd file generated by the sample training manager based on the scheme management category and the sample library to the ortho-image in combination witha classification method to obtain a classification result of the slope plant image. According to the method, the classification precision reaches 90%.

The invention discloses a vegetation spatial distribution pattern investigation method based on an unmanned aerial vehicle technology. The method comprises the following steps: S1, obtaining an orthographic spliced image of a research area; S2, performing supervised classification and result verification on the orthographic spliced image in S1; S3, making a species classification extraction distribution diagram; and S4, performing spatial distribution pattern analysis on the species. The method has the beneficial effects that vegetation species classification is carried out on the orthographicspliced images by utilizing a supervised classification method, so that specific ecological statistical analysis is carried out on individuals of each species; manpower and material resources are saved, meanwhile, a pattern analysis quadrat can be expanded by tens of times, and a spatial pattern analysis conclusion can be obtained more comprehensively and accurately; multi-angle spatial distribution pattern analysis can be more conveniently carried out on the species classification extraction distribution diagram, such as a sample method, a point pattern analysis method and the like, so thatpattern analysis results are more comprehensive and specific.

The invention discloses a hyperspectral image vegetation classification method based on spatial auto-correlation information. The method comprises the following steps that firstly a hyperspectral dataset is normalized so that a hyperspectral image data set R of which the information content is redistributed is obtained; then PCA dimension reduction is performed on the hyperspectral images and a new data set K is formed; then spatial information is extracted from all the spectral information R by using domain transformation linear interpolation convolutional filtering; then the spatial information is extracted from the data set K by using domain transformation linear interpolation convolutional filtering; then the extracted spatial information is linearly fused so as to form a new data setH; then a training set Hs is randomly extracted from the spatial information data set H, and the rest part acts as a test set Ht; then the training set HS is trained by using the SVM supported by theradial basis function so as to acquire a training model; and finally the test set Ht is classified by using the SVM supported by the radial basis function so as to obtain the classification result.

The invention discloses an unmanned aerial vehicle slope vegetation classification method based on an equal-height air route. The method comprises the following steps that (1) an unmanned aerial vehicle flies along a plurality of equal-height air routes in a snakelike mode and hovers on sampling points on the equal-height air routes to collect images, the distances between the equal-height air routes and the ground of a side slope are equal, the equal-height air routes are evenly arranged side by side in the inclination direction of the side slope at intervals, and the air routes are evenly provided with the sampling points; (2) the image is synthesized into high-density point cloud data by using Pix4DMapper, and an orthographic image is obtained from the high-density point cloud data; (3)a sample manager is constructed, which comprises the steps of firstly, adding the artificially distinguished plant species into ArcGIS to generate a scheme management category; then, according to thescheme management category, circling a to-be-classified object category in the near-earth orthographic image to obtain a sample library; (4) an ecd file generated by the sample training manager basedon the scheme management category and the sample library is combined with a classification method to act on the orthoimage, and a classification result of the slope plant image is obtained; the method improves the classification precision.

The invention discloses a marsh vegetation stack ensemble learning classification method integrating hyperspectral and multi-band full-polarization SAR images, and the method comprises the steps: integrating a hyperspectral image and different-band full-polarization SAR images, carrying out the variable optimization through multi-scale segmentation, high-correlation variable elimination and a Boruta algorithm, constructing a multi-dimensional variable data set, and carrying out the classification of the marsh vegetation stack. And carrying out stack integration on the classification models after different parameter optimization by using a Stacking algorithm, constructing a marsh vegetation identification and classification model, finally classifying data to be classified by using the model to obtain a marsh wetland vegetation classification result, and carrying out quantitative evaluation on the classification result by using an evaluation index. According to the method, the rich spectral information advantage of the hyperspectral image and the advantage that the polarimetric SAR image can penetrate through the vegetation canopies are integrated to realize high-precision recognition and classification of the marsh vegetation.

A method of categorising vegetation is disclosed, the method comprising: inputting and storing in a memory information related to a ground observation; determining the identity of the vegetation of said observations corresponding to an existing memory entry; determining a characteristic for said vegetation in dependence on said ground observation; and updating said memory with said characteristic.

The invention discloses a high-spectral image unmixing method based on the differential search. According to the technical scheme of the invention, bound terms corresponding to the characteristic structure that the abundance of a high-spectral image is nonnegative and the abundance sum of the high-spectral image is 1, are combined with the mutual information to serve as an objective function. Based on the differential search method, the objective function is optimized and solved to realize the unmixing purpose of the high-spectral image. The method plays an important role in the fields of ecological environment monitoring, disaster prediction, resource exploration, crop yield estimation, vegetation classification, military satellite reconnaissance, military target detection, and the like.

The invention relates to an early warning method and system based on satellite data fusion mountain fire monitoring and fire behavior deduction. The method comprises the following steps: firstly, acquiring a fire point position in a target area by utilizing processed and analyzed meteorological hot spot data, outputting a forest fire monitoring and early warning file according to a vegetation classification condition at the target position, issuing early warning information, and forming a forest fire risk distribution thermodynamic diagram according to forest fire historical data; meanwhile, inputting the monitored longitude and latitude of the fire point into the fire behavior deduction module, judging the fire behavior development speed and direction comprehensively according to the weather, landform and land cover data of the fire point position, and outputting and publishing the fire behavior development range within the set time. The forest fire condition of the target area can bemonitored in a large-scale and all-weather mode through the satellite remote sensing technology, the neutral period existing in manual on-site observation can be made up, early warning is conducted in the early stage of forest fire development, the forest fire spreading range is deduced and predicted, and first-hand data information is provided for forest fire protection and related departments to take measures.

The invention discloses a vegetation classification and biomass inversion method based on remote sensing data, and belongs to the field of mechanical learning algorithms. On the basis of multispectral data of a Sentinel-2B satellite, a vector boundary and a normalized vegetation index research area are used for extraction, then earth surface vegetation of a chonggift area is classified through a K-means clustering algorithm, and a classification result can be combined with a biomass inversion model to calculate the total biomass of an investigation area. A regional biomass inversion equation is constructed through multispectral remote sensing data and topographic data, an optimal biomass model of shrubby, broad-leaved forest and coniferous forest is obtained through multiple linear regression, the overall precision can reach 90% or above, and the survey precision requirement can be met.

The invention discloses a large-scale intertidal zone vegetation classification method based on a synthetic aperture radar. The method comprises the following steps: actually measuring the basic situation of an intertidal zone; carrying out intertidal zone partitioning, and carrying out sub-region partitioning on the intertidal zone based on the intertidal zone basic condition investigation result; generating a characteristic spectrum data set, calculating a new spectrum segment based on the original synthetic aperture radar data, and synthesizing the data set; inputting a training sample, andrandomly and uniformly selecting a plurality of samples of different ground objects; carrying out threshold segmentation, extracting a characteristic wave spectrum data set based on a sample, and selecting an appropriate threshold in the frequency distribution diagram to distinguish different regions; generating a decision tree, and constructing a set of large-scale intertidal zone vegetation classification system based on the threshold segmentation result of each region; inputting a to-be-classified region, and selecting a corresponding decision tree model according to the input region; generating a classification result; performing classification post-processing; area statistics and mapping. Compared with a traditional small-area classification method, the invention has the advantages of being small in workload, easy to operate, high in efficiency, high in robustness and the like.

The invention discloses a hyperspectral image unmixing method based on differential search. The present invention constructs corresponding constraint items according to the characteristics of non-negative abundance and abundance sum of hyperspectral image, combines with mutual information as an objective function, and uses differential search method to optimize and solve the objective function to realize hyperspectral image solution mix. The invention plays an important role in the fields of monitoring ecological environment, disaster prediction, resource exploration, crop yield estimation, vegetation classification, satellite reconnaissance in military affairs, target detection and the like.

The invention discloses a carbon sequestration and sink increase effect monitoring and accounting system based on multi-region ecological restoration, relates to the technical field of monitoring and measurement, and solves the technical problem that the monitoring capability of the carbon sequestration and sink increase effect is lagged. According to the scheme, the system comprises a regional vegetation ecological data information module, an ecological data fusion module, a vegetation classification model, a fusion output module, a measurement analysis module, a big data algorithm model, a wireless communication module, a carbon sequestration and sink increase monitoring module and a remote ecological data monitoring center. The output end of the ecological data fusion module is connected with the fusion output module through the vegetation classification model, the output end of the fusion output module is connected with the input end of the measurement analysis module, and the output end of the measurement analysis module is connected with the input end of the ecological area. The output end of the ecological area is connected with the carbon sequestration and sink increase monitoring module through the wireless communication module. According to the method, the carbon contents of different geological conditions are respectively calculated, so that the carbon sequestration and sink increase of the geological conditions are effectively evaluated.

The invention discloses a remote sensing image vegetation classification method based on a gradient scale interval change rule operator. According to the present invention, the gradient scale intervalchange rule operator is established and can describe the characteristics of different vegetations in a remote sensing image from a non-directional specific brightness interval and various scales, andthen the characteristics are learned by using a support vector machine model to obtain a classification model, and the classification result of the remote sensing image can be obtained by using the gradient scale interval change rule operator and the classification model. By utilizing the method, the influence of a specific scale and a specific direction on classification can be avoided, the characteristics of the vegetation on the remote sensing image are displayed according to the interval characteristics of the specific brightness pixel of the vegetation on the image and the change characteristics of the scale, and further the precision of the remote sensing image vegetation classification is improved by utilizing the characteristics.