Method for identifying impermeable surface of remote sensing shadow measurement repaired image

Abstract

The invention designs a set of high-resolution remote sensing image non-permeable surface identification method around measurement restoration of a shadow area on a high-resolution remote sensing image, land type interpretation and refined urban non-permeable surface identification: 1, aiming at the problem that urban non-permeable surface identification has uncertainty due to shadows generated by tall buildings and crowns on the image, the method is provided for the purpose of identifying the urban non-permeable surface; a high-resolution remote sensing image shadow measurement and restoration method oriented to a non-permeable surface is provided, PCA transformation and HIS transformation are introduced to obtain multiple spectral features, recognition measurement of a shadow area is achieved, and restoration of the city shadow area is achieved based on blue band component suppression and HSI space restoration; and secondly, in order to better obtain a city impermeable surface recognition result and analyze the influence of setting of different segmentation factors and parameters on impermeable surface recognition, a multi-decision tree combined classifier is adopted, object-oriented impermeable surface refined recognition is realized, the impermeable surface recognition accuracy is 96.38%, and a powerful guarantee is provided for research and application of the impermeable surface.

Description

technical field [0001] The application relates to a method for identifying an impermeable surface of a remote sensing shadow image, in particular to a method for identifying an impermeable surface of a remote sensing shadow metering restoration image, which belongs to the technical field of identifying an impermeable surface of a remote sensing image. Background technique [0002] The impermeable surface is the area where water cannot penetrate. The impermeable surface is mainly constructed artificially, including roads, buildings, parking lots, etc. The permeable surface opposite the impermeable surface includes green land, lakes, rivers, bare soil, etc. , with the current rapid development of urbanization, large areas of permeable surfaces are replaced by impermeable surfaces. The increase of the non-permeable surface in the city will make it difficult for precipitation to infiltrate into the ground, resulting in the reduction of infiltration water and soil moisture in the...

AI Technical Summary

Problems solved by technology

[0011] First, the identification of non-permeable surfaces was initially based on manual statistical surveys. The data source was limited by various aspects and could only be used in a small area. The work was repetitive, slow and heavy, and the real-time performance was poor. It is difficult to update, so there is an urgent need to improve the level of automation; and the use of low- and medium-resolution remote sensing images is suitable for the identification of impermeable surfaces in large areas, and the recognition accuracy is low. Recognition and acquisition requirements, but high-resolution images have problems such as the lack of information in shadow areas that cannot be avoided. This problem will have a greater negative impact on the recognition accuracy of regional non-permeable surfaces. The existing technology cannot completely remove the shadow on the image. There is no way to improve the lack of shadow information, and there is no recognized method for effectively repairing shadow areas.

It cannot improve the visual effect of the shadowed area, nor can it guarantee that the spectral information in the non-shaded area remains unchanged, and cannot meet the needs of accurate identification of non-permeable surfaces in remote sensing images;

[0012] Second, due to the complex types of ground objects contained in the impermeable surface, the use of low- and medium-resolution images for the identification of the impervious surface can easily be confused with soil, water bodies, and shadow areas. Although the sub-pixel-level impervious surface identification method It can eliminate the problem of assimilating pixels, but this method depends on the ratio of the impermeable surface occupied by each pixel, which may easily cause the problem of overestimation or underestimation of the impermeable surface. Use high-resolution remote sensing images to obtain regional impermeable surface information Higher-precision area impermeable surface information can be obtained. However, the use of high-resolution remote sensing images will inevitably encounter the problem of missing shadow area information. The existing technology cannot accurately identify and measure shadows in remote sensing images, and cannot effectively identify and measure shadows. Repair and improve the problem of lack of shadow information in high-resolution images, and then it is impossible to obtain accurate results of urban impermeable surfaces through automated classification and recognition methods, resulting in low accuracy of impermeable surface recognition accuracy, and even loss of practical use value;

[0013] Third, due to the shadows caused by undulating terrain, tall buildings, and tree crowns, it is difficult to detect shadow areas on high-resolution remote sensing images and to distinguish land types in shadow areas. The relationship between shadows and water bodies, shadows and plants, shadows and dark features The confusion between the two areas is serious, and the information in the shaded area is missing, which seriously affects the subsequent refined identification of urban impermeable surfaces, and brings great difficulties to the fine identification of urban impervious surfaces. There is no reasonable solution in the existing technology, which leads to subsequent Obstacles in the interpretation of impermeable surface classification, shadows have a great negative impact on the classification of impermeable surfaces, and ultimately lead to low recognition accuracy of impermeable surfaces, poor usability and reliability, and cannot be extended to many important fields such as the determination of impermeable surfaces. Extremely limited, poor accuracy leads to almost no practical value in actual use;

[0014] Fourth, even if the shadowed area is repaired, there is still a certain spectral difference from the non-shaded area, so it is necessary to classify the shadowed and non-shaded areas separately, but the accuracy and stability of the non-permeable surface identification and classification methods of the existing technology Unable to meet the requirements, the generalization error of the classifier cannot be converged, the overfitting problem cannot be avoided, the processing ability of remote sensing image datasets with missing features is poor, and various image features cannot be selected based on their importance, human intervention Many; the anti-noise and shadow processing capabilities are poor, the existing technology cannot be processed in parallel, and the calculation efficiency is low, which ultimately makes the remote sensing image impermeable surface identification method not feasible, and the accuracy is not guaranteed

Images