Land utilization category determination method fusing in streetscape images

Abstract

The invention discloses a land utilization category determination method fusing in streetscape images. The method comprises the following steps of firstly, extracting the fine ground object category information in the street view image sampling points through a deep learning convolutional neural network mode; meanwhile, preprocessing the remote sensing image, and obtaining a land cover map by using a supervised classification method; secondly, through the spectrum, the texture, the shape and the geographic distribution information of the pixels where the streetscape sampling points are located, deducing the category condition of the adjacent pixels; and finally, fusing the pixel classification information and the land cover map to obtain a fine multi-class land utilization result. According to the method, by starting from the remote sensing based on the pixel classification, the fine streetscape information is combined, and the precision of the classification result is high.

Description

technical field [0001] The invention relates to a method for determining a land use category and belongs to the field of geospatial information. Background technique [0002] Land use refers to all human activities that purposefully develop and utilize land resources. Land use information plays a vital role in urban planning, urban environmental monitoring, and urban transportation (Liu X, He J, YaoY, et al. al.). How to extract more accurate land use information is an important issue and challenge. As a large-scale, non-contact data source, remote sensing image information is an important means of land use information extraction (Zhao Yingshi). Using high-resolution remote sensing images acquired by resource satellites, combined with professional remote sensing image processing software, and applying scientific information extraction methods, the corresponding resource information can be obtained quickly and accurately. [0003] However, due to the influence of the accur...

AI Technical Summary

Problems solved by technology

However, these methods are limited to the block scale, and do not use single pixel information to classify land use, and for some areas with sparse street view sample points, or areas with uneven distribution of street view samples, using kernel density regression will produce big error

Images