METHOD OF GEODETIC GEOINFORMATION MONITORING OF NATURAL AND MAN-MADE OBJECTS WITH THE APPLICATION OF THE AUTOMATIC INTERPRETATION METHOD OF MULTISPECTRAL DIGITAL AEROSPACE PHOTOGRAPHS Russian patent published in 2018 - IPC G01C11/12 

FIELD: image processing means.

SUBSTANCE: invention relates to methods for processing multispectral digital aerospace photographs and can be used for geodetic geoinformation monitoring of natural and man-made objects. Summary: on the monitored area an aerospace digital photography is performed with the help of a single survey aerospace system with reference to the plan-altitude coordinate system (AD). Results of aerospace digital photography of a given territory are obtained in the form of initial multispectral digital aerospace photographs in a given period of time. Obtained results are transferred to a PC and their automated interpretation is performed with the use of computer program by a linear spectral transformation method using the Tasseled Cap (TC) coefficients. Results of automated interpretation are obtained in the form of initial multispectral digital images of a given terrain site. Geodetic geoinformation monitoring of objects on the controlled site is performed in a given period of time. In this process, in the process of automated interpretation, additionally in the same program, rationing of the original multispectral digital images by bringing the same type of objects to the same brightness. Results are obtained in the form of normalized multispectral digital images. There are selected with the use of a cartographic material and saved separately test sections of the same size, that contain the same type of objects in the measurement space. Filtering parameters re determined for removing in the automatic mode from the matrix of each test section of the normalized multispectral digital images of the elements, not related to the objects in the measurement space required for interpretation. Eigen vectors for each test sections are calculated using the method of increasing the stability of the MCD covariance matrix (minimal determinant of the covariance matrix). First, an initial covariance matrix is calculated for all elements of the matrix of the normalized multispectral digital image for each test section. Then the percentage of points, taken in the processing, are selected, the number of points that should be discarded is determined. Further, in the same program, from calculated for each point Mahalanobis distance the points with the its maximum distance are discarded. Covariance matrix for each test section are calculated for the remaining points after the discard. After this, for each covariance matrix, the eigenvectors are calculated, on the basis of which the TC transformation coefficients, calculated for sections with a typical set of objects. Transformation of the TC for the normalized multispectral digital images of aerospace images is performed using the calculated TC coefficients. Results of the transformation are obtained in the form of transformed normalized multispectral digital images with the largest spectral distance between objects of different types. From the transformed images using the cartographic materials, the standards for each object are selected. Standards are used as a training sample in the automatic classification and vectorization of the transformed from the calculated TC coefficients of normalized multispectral digital images. Graphical thematic digital map with simultaneous automatic calculation of the area of the given terrain site with reference to the AD coordinate system in a given period of time. Said monitoring is performed for natural and man-made objects by means of automated comparison of graphic thematic digital maps obtained at different time periods to the same territory according to the results of automated interpretation by the linear spectral transformation method using the TC coefficients.

EFFECT: technical result: increasing the information content and accuracy of the results obtained, increasing the processing speed of multispectral digital aerospace photographs.

1 cl, 1 dwg