Global perception high-resolution remote sensing image multi-scale residual road extraction method

Abstract

The invention relates to a global perception high-resolution remote sensing image multi-scale residual error road extraction method. According to the method, the capability of capturing the long-distance dependency relationship of the model is improved from two aspects: on the one hand, the long-distance dependency relationship is captured from two dimensions of space and channel by adopting global perception operation between residual modules; on the other hand, in the residual module, we use multi-scale residual convolution learning to extract multi-scale features and increase a network receptive field at the same time, thereby providing richer information for subsequent global perception operation. According to the method, the problem of scattered road recognition caused by the fact that the receptive field of an existing deep learning model is limited and is not matched with the long-span connection attribute of the road can be solved to a certain extent, and the road recognition precision can be remarkably improved.

Description

technical field [0001] The present invention is based on the technical processing field of remote sensing images, and in particular relates to a method for extracting multi-scale residual roads from globally perceived high-resolution remote sensing images. Background technique [0002] As an important infrastructure construction of the country, roads are an important ground object in high-resolution remote sensing images, and have a very important impact on human life. For a country, the more developed the transportation system, the more conducive to the country's economic and social construction. The traditional road information acquisition method requires professionals to survey maps on the spot, vectorize editing, and finally obtain road data. This process involves a lot of manual participation, many external factors interfere, and the update speed is slow, which can no longer meet the needs of rapid economic and social development. On the high-resolution remote sensing ...

AI Technical Summary

Problems solved by technology

[0006] The proposed multi-scale residual road extraction method for global sensing high-resolution remote sensing images increases the ability of the model to capture long-distance dependencies from two aspects. Two dimensions to capture long-distance dependencies; on the other hand, in the residual module, we use multi-scale residual convolution learning to extract multi-scale features while increasing the network receptive field, providing more information for the subsequent global perception operation. Abundant information; the present invention can solve the problem of fragmented road recognition caused by the limited receptive field of the existing deep learning model and the mismatch of road long-span connection attributes

Images