124 results about "Visual interpretation" patented technology

Completely automated tests that exploit capabilities of human vision to tell humans apart from automated entities are disclosed herein. Persistence of vision and simultaneous contrasts are some of the properties of human vision that can be used in these tests. A video of an image is generated in colors that are distinguishable to the human eye but are not easily distinguished numerically. The image includes text manipulated such that positive image data and negative whitespace data occur at equal rates along with a noise component included in each of the video frames. Thus, raw data is made ambiguous while qualities of human visual interpretation are relied upon for extracting relevant meaning from the video.

The invention relates to a high-spatial resolution remote sensing image-based road region building change extraction method and device. According to the method and device, an object-oriented image processing strategy is adopted, the spectral information and spatial information (including structural indexes and spatial relationships) of images are comprehensively utilized, and a single-class classification method is adopted. In order to avoid interference (spectral similarity) on extracted results caused by ground feature classes except road regions, it is required that existing road information such as an existing road vector diagram, is provided in advance; the existing road vector diagram is adopted to extract a road region range; and newly increased buildings are extracted within the road region range. With the high-spatial resolution remote sensing image-based road area building change extraction method and device provided by the invention, time for a traditional method to obtain newly increased buildings such as illegal buildings by using image visual interpretation can be greatly decreased, efficiency can be improved, and human resources can be saved. The method and device can be used for road maintenance and monitoring business operation systems.

The invention discloses a plastic film mulching farmland remote sensing monitoring method based on spectrum and texture features, and the method comprises the steps: S1, carrying out preprocessing of remote sensing image data of a research area; S2, building a plastic film mulching agricultural remote sensing monitoring classification system; S3, employing a Google Earth image the same as the remote sensing image data of the research area, collecting irregular polygon samples of different ground features in the classification system, and drawing the outline of a regular polygon sample of an element at a preset size through visual interpretation; S4, carrying out the separable analysis, so as to select a usable band; S5, extracting a plurality of types of texture features based on the multiband data of a remote sensing image, and extracting the texture features at three steps in four directions; S6, carrying out dimension reduction of the extracted texture feature parameters, and selecting the texture features; S7, building an input feature parameter set based on the combination of spectrum and texture features; S8, carrying out the classification of ground features of the classification system through employing a classifier. The invention proposes a new method for monitoring a plastic film mulching farmland.

The present invention relates to a winter wheat plantation area calculation method based on a winter wheat area index. The comprises: performing preprocessing of multi-date remote sensing image, and obtaining the NDVI data in a winter wheat growth period; dividing the multi-date remote sensing image into regular nets, setting weight sample central points on the regular girds, and employing a visual interpretation mode to determine the types of the samples in the girds as winter wheat and non-winter wheat; obtaining the NDVI values corresponding to the weight sample central points in the multi-date gird data, respectively calculating the NDVI mean values of the winter wheat and non-winter wheat, comparing the size of the two and setting the weight; performing weight operation of the NDVI mean values according to the weight setting result to obtaining the winter wheat area index; and employing a self-adaptive mode to extract the winter wheat area index optimal threshold value, and obtaining the plantation area of the winter wheat. The winter wheat plantation area calculation method based on the winter wheat area index can realize the high-precision calculation of the winter wheat area of the area scale of the multi-date high resolution images, the degree automation of the method is high, and the consistency of the identification result is high.

The invention discloses a remote sensing image mangrove forest extraction method and system based on a deep convolutional neural network, and the method comprises the steps: firstly carrying out the preprocessing of a high-resolution remote sensing image, including the atmospheric correction and research region cutting of the remote sensing image, and carrying out the waveband operation of each waveband after processing, so as to extract the priori feature information; achieving multi-band and feature information fusion by applying multi-source data fusion, and constructing a data set; training and verifying a semantic classification model ME-net built by a convolutional neural network; calling an ME-net model to realize automatic classification of the mangrove forest, and outputting a mask file in a png format, namely a classification and extraction result; and performing fine adjustment on the classification result through a long-distance conditional random field. In application of the classification model, the classification precision can reach 92.3% by expanding a data set, artificial visual interpretation can be completely replaced, and auxiliary technical support is providedfor updating of a high-precision image map and protection of a coastal region ecosystem.

The invention discloses a remote-sensing image multi-scale object-oriented classification method based on joint sparsity representation. The method includes the steps of firstly, mining a spatial characteristic of a remote-sensing image and constructing an augmentation characteristic through the combination of the spatial characteristic and a spectral characteristic, secondly, constructing a redundant dictionary by means of a training pixel sample and the augmentation characteristic, carrying out joint sparsity representation on initially segmented patches by the adoption of the redundant dictionary, thirdly, carrying out homogeneity analysis and reconstruction effect analysis on the patches on the basis of the joint sparsity representation, and finally, judging whether patch segmentation is reasonable according to analysis results of the homogeneity analysis and the reconstruction effect analysis, and carrying out classification identification on the patches meeting the requirements of the homogeneity level and the reconstruction effect. The method achieves organic combination of the segmentation process and the classification process so as to obtain proper ground object patches for classification, achieves classification identification of the remote-sensing image on the object level, can obtain a classification result meeting the requirement of visual interpretation, greatly improves interpretation precision of the remote-sensing image, and has great application value.

The invention discloses a method for detecting subcompartment forestry resources, comprising the following steps: step 1, checking the extraction of subcompartment; 2, collecting and processing forestry data; 3, proces that remote sensing image; 4, detect that change of the remote sensing image in the early and late period; 5, image visual interpretation; 6, supplemental investigation; 7, update that data model; 8, forestry resource information management; 9, update that output; In the first step, random sampling or typical sampling method can be used to check the extraction of subcompartments; When adopting typical sample selection, the appraised objects should be classified according to forest species, woodland type, tree species type and age group. The invention is highly effective andfeasible, and takes 3S technology and mathematical model as main means, and a forestry resource renewal method based on investigation subcompartment data, and puts forward a complete solution of annual output number of forestry resource space and attribute data, and the data detection result is objective and accurate.

The invention provides a night scene denoising and enhancing processing method, relates to the field of remote sensing image processing, and solves the problems existing in an existing night scene image denoising method that visual interpretation is inconvenient, and in a process of performing enhancing processing on an image, an enhancing effect is poor and consumed time is long. Denoising processing is performed on a received original image, thereby obtaining a denoised image; Bell interpolation is performed on the denoised image, thereby obtaining a color image of three wavebands of RGB; the obtained color image of the three wavebands of RGB is converted from RGB into a YUV color domain, thereby obtaining image brightness, according to the image brightness, an image gray scale stretching coefficient is calculated, and a ratio coefficient of image brightness before and after stretching is utilized to adjust three channels of R, G and B in the same proportion, thereby obtaining an enhanced image. The binarized image generated by combination of two methods can effectively separate a foreground from a background, noise is removed, high-frequency details of the image is kept, an algorithm is simple and practicable, and thus the method is easy to directly apply in engineering treatment.

The invention discloses a touch screen mobile terminal image quick comparison interaction method. The method comprises the steps that a first layer is viewed; a second layer is added above the first layer for overlapping, then registration is performed according to projection coordinates, the first layer and the second layer are completely overlapped, and an overlapping layer is formed; when a user presses a screen for time exceeding preset hidden pressing time, a mobile terminal system modifies the transparency of the second layer till the second layer has a hidden effect, and the first layeris viewed through the second layer; and when the mobile terminal system judges that a finger makes contact with the mobile terminal screen, the overlapping layer can be dragged to move. Through the touch screen mobile terminal image quick comparison interaction method, more clearer content can be displayed on the screen, the operable range is enlarged, more convenience is brought to comparison for visual interpretation, screen operation cost caused by pressing and dragging is reduced, and discomfort caused by friction between the finger and the screen is relieved.

The present invention provides a mountain vegetation vertical zone investigation monitoring method based on a UAV. The method comprises the steps of (S1) determining an investigation time, an investigation mode and investigation precision, (S2) planning a route and carrying out automatic monitoring according to a preset route if the investigation mode is a programmed control operation mode, and manually controlling an aircraft to carrying out image capturing and obtaining monitoring data if the investigation mode is a radio remote control operation mode, (S3) carrying out space correcting processing on each image in the monitoring data, realizing the space connection of images, carrying out visual interpretation or computer automatic classification by using the spectrum and texture structure differences of different vegetation zones in a remote sensing image, identifying a vegetation vertical zone boundary, and completing the statistical measurement analysis of a vegetation vertical zone structure and result output through a geographic information system. According to the mountain vegetation vertical zone investigation monitoring method, time and labor are saved, climbing a mountain to carry out sampling positioning by investigation and monitoring personnel is not needed, an investigation result is processed by using the image analysis technology, the vertical vegetation zone identification precision is high, and the positioning is more accurate.

The invention discloses a spectral characteristic based remote sensing monitoring method of plastic film mulched farmland. The spectral characteristic based remote sensing monitoring method of the plastic film mulched farmland comprises a step S1 of carrying out preprocessing on a remote sensing image, which comprises S1, radiation correction, atmospheric correction and the carrying out of inlaying and cutting processing on an image to obtain a research-area image, on the remote sensing image; a step S2 of establishing a plastic film mulched farmland remote sensing monitoring classification system, so as to distinguish the plastic film mulched farmland and other ground objects; a step S3 of collecting irregular polygonal samples of different ground object types in the classification system through visually interpreting a Google Earth image with a time phase same with that of the research-area image, and then delineating regular polygonal samples of a predetermined-size picture element anew in irregular polygons through visual interpretation; a step S4 of carrying out analysis on the separability of the different ground objects in the research-area image, so as to select a separable wave band; a step S5 of classifying the research-area image by utilizing the regular polygonal samples and using a classifier. The invention provides a new method for monitoring the plastic film mulched farmland.

The invention provides an expert system for remote sensing image processing. The expert system for remote sensing image processing comprises a user interface, a system interface, an inference machine, knowledge bases, a management system of the knowledge bases, a database, a database management system, a knowledge obtaining mechanism and an interpretation mechanism. The expert system earnestly and scientifically conclude and summarize and establish the knowledge bases by simulating visual interpretation experiences of a human expert, then quotes the knowledge bases to classify remote sensing data, utilizes a mode recognition method to obtain multiple characteristics of ground feature, provides evidence for remote sensing image interpretation, meanwhile applies an artificial intelligence technology, applies experiences and methods of the remote sensing image interpretation expert, simulates specific thinking process of remote sensing image visual interpretation, and performs remote sensing image interpretation to enable the computer interpretation level to reach the expert level.

The invention discloses a remote sensing earthquake damage information extracting and digging method based on pixels. The method comprises the following steps of: firstly, automatically finding a change area by a spectrum variance and main component integration method according to the remote sensing images of different time phases before and after earthquake; then, segmenting the change image by using a dual-threshold area growth algorithm to obtain change information binary images; finally, carrying out artificial visual interpretation on the segmented binary images of the change area, and extracting the earthquake damage information. The invention solves the problems that the workload of artificial interpretation on the earthquake change information is large and the survey period is long, enhances the efficiency, and simultaneously guarantees the sufficient precision.

The invention belongs to the technical field of remote sensing image processing, and particularly discloses a high-resolution remote sensing image fusion method based on linear Bayesian estimation. According to the high-resolution remote sensing image fusion method, an observation model between a high-resolution multi-spectrum image and a low-resolution multi-spectrum image and an observation model between the high-resolution multi-spectrum image and a full-color image are introduced, and the two observation models are connected to form a Bayesian linear model. Estimation on the high-resolution multi-spectrum image in the linear minimum mean square error sense is obtained by applying the Bayes Gauss-Markov theorem. According to the high-resolution remote sensing image fusion method, space details can be strengthened, the spectral characteristics are well kept, the performance of the high-resolution remote sensing image fusion method is better than the performance of a traditional HIS method, a traditional PCA method, a traditional wavelet transformation method, an existing Nishii method based on statistic parameter estimation and an existing Hardie method based on statistic parameter estimation, and new and effective technical supports are provided for improving the visual interpretation accuracy of a remote sensing image, the information definition and the reliability.

The invention discloses a method and platform for dynamic monitoring of typical ground features in mining based on multi-source remote sensing data fusion and a deep neural network. The method includes: 1) acquiring time-series multi-source remote sensing data and basic mine data; 2) processing the remote sensing data and performing fusion of multi-source heterogeneous data to obtain enhanced remote sensing images of high-resolution multi-spectral images; 3) constructing a deep neural network model, and performing intelligent extraction on typical ground features by a method combining the deepneural network and artificial visual interpretation; 4) performing online dynamic monitoring and data analysis management on the extracted typical features. This method can improve the spatial resolution of remote sensing data images and it is easier to identify typical features after enhancing the image. At the same time, the machine interpretation manner adopted can efficiently and accurately identify the typical features, so that the typical features of mines can be online monitored in real time.

The invention discloses a method for detecting a marine moving target of a thermal infrared remote sensing image based on multi-feature fusion. The method comprises the following steps of carrying out sea-land segmentation on a thermal infrared image to obtain a sea area image; utilizing a priori segmentation parameter to carry out multi-scale segmentation on the sea area image to obtain an object; extracting brightness and geometrical characteristics of the object obtained by segmentation; and carrying out decision tree classification on a segmentation object through corresponding classification rules which are obtained by respectively carrying out contrast training on the obtained brightness and the geometrical characteristics of the segmentation object in order to realize detection of the marine moving target. The method utilizes the thermal infrared remote sensing image as a data source, so that the defects of failure of a visible/near-infrared multispectral image at night, difficulty in visual interpretation of an SAR image, complicated processing steps and the like in the existing method can be made up to be beneficial to detection of a detection result, so that the accuracy of information is further improved; the obtained temperature information is extracted based on the marine moving target obtained by detection to be beneficial to judgement of state of the marine moving target.

The invention discloses a deep learning satellite data cloud detection method supported by hyperspectral data. The method comprises the following steps: selecting enough cloud and clear sky pixels toconstruct a hyperspectral data sample library, and performing analog computation on the hyperspectral pixel sample library according to parameters such as a spectral response function and a waveband width of a to-be-detected sensor to obtain a cloud and clear sky surface pixel library of the to-be-detected sensor; based on a Keras deep learning framework, designing a deep BP neural network for cloud detection, inputting multispectral sample data obtained through simulation into the network for training and learning, and obtaining a multispectral sensor cloud detection rule based on spectral characteristics; based on a Markov random field model, optimizing a cloud detection result by utilizing an iterative condition mode algorithm, and removing a part of misclassification and misclassification errors of cloud detection. According to the method, various sensor data are selected and compared with a cloud coverage result of artificial visual interpretation for analysis, and the result shows that the algorithm achieves a better cloud detection effect and can meet the requirements of data application for cloud detection.

The invention discloses a double-path convolutional neural network remote sensing classification method taking a spatial neighborhood relationship into account. The double-path convolutional neural network remote sensing classification method comprises the steps of: data acquisition; data preprocessing; thematic map production; multi-scale segmentation; visual interpretation; first-order neighborhood acquisition; experimental sample set selection; patch size selection; training data generation; mat format conversion; convolutional neural network structure construction; subject single-path neural network training; first-order neighborhood single-path neural network training; complementarity analysis; and double-path convolutional neural network training taking the spatial neighborhood relationship into account. The invention aims to provide a method for regarding a subject and a first-order neighborhood of the subject as input image data on the same scale, realizing the enhancement of sample information through increasing neighborhood information of samples and reducing interference caused by other types of samples, so as to realize the improvement of classification precision and provide a reference for classification decision-making.

The invention discloses a surface form deviation measurement method of a flat optical element. Firstly, a camera collects a standard sample image data and sends the collected standard sample image data to DSP via a USB interface for digital signal processing; the processed image data trains a BP neural network to determine the adopted BP neural network; because the camera collects the interferencefringe image data of a measured flat optical element, the camera sends the collected image data to the DSP via the USB interface for digital signal processing, and the obtained measuring data is sentto a determined BP neural network for calculation; the BP neural network outputs a surface form deviation value and sends into a DSP cache region; and the DSP sends the surface form deviation value to a display device for displaying. Because a manipulator can conveniently observe the interference fringe image and adjust the light path, detection cost is lowered; the measuring result is not affected by a person of visual interpretation, thus effectively improving measurement precision.

The invention relates to a landslide recognition knowledge base establishment method and device, electronic equipment and a storage medium. The method comprises the following steps: carrying out the analysis of an SAR image data set of each target region through employing an InSAR technology, and determining a time sequence deformation diagram and an accumulated deformation diagram of each target region; based on the slope map and the slope direction map of the target area, deformation points in the time sequence deformation map of the target area are analyzed by jointly applying a hotspot clustering technology and a spatial clustering analysis technology, and the landslide boundary of the target area is determined; and based on a preset interpretation mark, performing remote sensing visual interpretation on the optical image data set, the time sequence deformation diagram, the accumulated deformation quantity diagram and the landslide boundary of the target area, and determining landslide attribute information of the target area. According to the method, the influence of the slope direction and the slope on landslide recognition is considered, the accuracy of the obtained landslide attribute information is improved, the data precision of a landslide recognition knowledge base is further improved, and a good foundation is laid for subsequent AI automatic recognition.