SAR image change detection method based on a deep capsule network

Abstract

The invention discloses an SAR image change detection method based on a deep capsule network structure. The SAR image change detection method mainly solves the problems that in the prior art, more useful feature information cannot be extracted from an image, and effective image information contained in an obtained training sample is insufficient. The method comprises the following implementation steps: acquiring two SAR experimental images; According to the pixel information of the experimental image, obtaining a similarity threshold between pixels and a difference value of the pixels; pre-classifying the experimental images by using a KI threshold method to obtain labels of the images; obtaining a training label and a training sample according to the similarity threshold, the difference value of the pixels and the label of the image, and training the training label and the training sample through a capsule network; layering and deepening the trained network ; and inputting the experiment diagram into the deep capsule network to obtain a change detection result diagram. According to the method, more useful feature information can be obtained from the training sample, the change detection precision is improved, and the method can be used for SAR image change detection of environment monitoring, agricultural investigation and disaster relief work.

Description

technical field [0001] The invention belongs to the technical field of image processing, in particular to a SAR image change detection method, which can be used in environmental monitoring, agricultural investigation and disaster relief. Background technique [0002] Synthetic aperture radar (SAR) has the characteristics of all-day, all-weather, high resolution, etc. Compared with visible light and infrared sensors, it has unique advantages. Change detection is the most important application in the field of remote sensing. It analyzes two images of the same area at different times, and obtains the required ground object change information according to the difference between the images. With the continuous development of remote sensing technology, change detection technology has also been developed rapidly and is widely used in agricultural production and scientific research. [0003] The process of SAR image change detection can be divided into image preprocessing and image...

AI Technical Summary

Problems solved by technology

Although the feature information processed by the capsule network can achieve good classification results, the network cannot more effectively classify the acquired high-level feature information, thereby reducing the detection accuracy and failing to obtain better change detection results.

Images