Remote sensing image enhancement method based on scene prior

Abstract

The invention belongs to the field of image processing, and relates to a remote sensing image enhancement method based on scene prior, which mainly comprises the following steps: 1, manually selectinghigh-quality remote sensing images of various scenes from remote sensing image data, and counting multi-scale features of the images; 2, performing scene classification training on the remote sensingimage data by using a convolutional neural network to obtain a stable scene classifier, and obtaining a scene of a to-be-enhanced image by using the scene classifier; and 3, enhancing the to-be-enhanced image by taking the multi-scale features of the high-quality remote sensing image in the same scene as constraint conditions so as to improve the image quality. Remote sensing images in differentscenes can be remarkably enhanced, and the defect that an existing image enhancement method is unstable in enhancement effect in different scenes is avoided.

Description

technical field [0001] The invention belongs to the field of image processing and relates to a remote sensing image enhancement method. Background technique [0002] During the acquisition process of remote sensing images, due to weather effects and equipment performance degradation, image contrast decreases, which affects the visual effects of remote sensing images and subsequent applications such as object classification, target recognition, and transformation detection. Image enhancement technology can increase the contrast of the image while maintaining the original information of the image, highlighting the overall and detailed information of the image. [0003] At present, prior-based image enhancement methods mainly focus on the following two aspects: [0004] 1) Image enhancement methods based on the information of the image itself. Most of these methods take the edge, texture, and sparsity of the image as prior constraints, and can obtain better image enhancement e...

AI Technical Summary

Problems solved by technology

However, the statistical information of remote sensing images in different scenes is often different, and it is difficult to use statistical information for image enhancement regardless of the scene to meet the actual application requirements.

Images