Change detection method based on high-resolution remote sensing image

Abstract

The invention discloses a change detection method based on a high-resolution remote sensing image. The method is characterized by using a superpixel segmentation and synthesis algorithm to segment a multi-temporal remote sensing image after carrying out necessary ortho-rectification, image registration, histogram matching and other preprocessing, using a superpixel as a unit to carry out local characteristic calculating and sample selection, and realizing the automatic annotation of a changing area possessing an obvious tendency or an unchanged area in the image; and then, taking an annotationresult as a sample to train twinborn convolutional neural networks, classifying image changing conditions, carrying out noise reduction, morphological filtering and other postprocessing, and acquiring a final changing detection result. An experiment shows that in a high-score number-two satellite remote sensing image data set, each index in the method is better than each index in a traditional change detection algorithm; a Kappa coefficient is averagely increased by 0.3; and an average overall error rate is less than 3.5%. The detection result has a high practical value.

Description

technical field [0001] The invention belongs to the technical field of remote sensing image recognition and deep learning, and in particular relates to a change detection method based on high-resolution remote sensing images. Background technique [0002] In recent years, satellite technology has achieved rapid development, and the application fields of satellite remote sensing images have also been continuously expanded, playing an important role in meteorology, geology, surveying and mapping, agriculture, forestry, animal husbandry and fishery, military reconnaissance and other fields. Remote sensing image change detection refers to the use of remote sensing images in the same area at different times and related data such as atmosphere and sensors, after image correction and other preprocessing, and with the help of mathematical statistics or artificial intelligence related technologies, feature extraction and comparison of remote sensing images, and Analyze and judge its ...

AI Technical Summary

Problems solved by technology

[0005] Although high-resolution multispectral satellite remote sensing images contain more information, they also introduce numerous interference factors and technical challenges. How to fully and reasonably use the information contained in the images and effectively reduce the impact of various interference factors on The impact of analysis is an urgent problem to be solved in change detection; the introduction of deep learning theory and methods has proposed new ideas for the optimization of change detection algorithms

[0006] In the learning process of the neural network, the importance of data is irreplaceable; the amount of data of remote sensing satellite images is very large, and the change itself lacks an objective definition, and may change with the change of the application scene. Labeling is not only a lot of work but also very difficult

Images