CN-121981932-A - Non-uniform defogging method and system for remote sensing image based on wavelet guidance

CN121981932ACN 121981932 ACN121981932 ACN 121981932ACN-121981932-A

Abstract

The invention provides a non-uniform defogging method and system for a remote sensing image based on wavelet guidance. The remote sensing image defogging method comprises the steps of carrying out wavelet transformation on a remote sensing haze image based on a physical model of haze imaging, decomposing the image into a low-frequency illumination component and a high-frequency texture component, constructing an energy function comprising physical consistency constraint and multiple priori constraints by taking clear image and medium transmissivity as optimization variables, carrying out joint solution on the low-frequency illumination component in an alternating optimization mode, carrying out detail enhancement processing on the high-frequency texture component, and finally carrying out reconstruction fusion on the optimized low-frequency illumination component and the enhanced high-frequency texture component to obtain the defogged remote sensing image. Correspondingly, the invention also provides a remote sensing image non-uniform defogging system for realizing the method. Through the technical scheme, the invention can effectively improve the brightness distribution and detail expression of the remote sensing image under the non-uniform haze condition, and improve the stability and visual quality of the defogging result.

Inventors

LI XIULAI
LI XING
LIU XIAOZHANG
XU JIAJIN
WU XINZHE

Assignees

海南大学

Dates

Publication Date: 20260505
Application Date: 20251231

Claims (10)

1. The non-uniform defogging method for the remote sensing image based on wavelet guidance is characterized by comprising the following steps of: Processing an input remote sensing haze image, and establishing a degradation relation of the remote sensing image under a haze condition based on a physical model of haze imaging; Performing wavelet transformation on the remote sensing haze image, and decomposing the remote sensing haze image into a low-frequency illumination component and a high-frequency texture component; Constructing an energy function containing physical consistency constraint and multiple priori constraints by taking clear images and medium transmissivity as optimization variables, and performing variation optimization on the low-frequency illumination component; adopting an alternate optimization mode to carry out joint solution on the clear image and the medium transmittance; And carrying out detail enhancement processing on the high-frequency texture component, and carrying out reconstruction fusion on the optimized low-frequency illumination component and the enhanced high-frequency texture component to obtain a defogged remote sensing image.
2. The wavelet guide-based non-uniform defogging method for a remote sensing image according to claim 1, wherein the physical model based on haze imaging is an atmospheric scattering model for describing imaging relations among scene radiation, medium transmittance and atmospheric light in the remote sensing image.
3. The wavelet guide-based remote sensing image non-uniform defogging method according to claim 1, wherein the energy function comprises at least a data fidelity constraint term and a regularization constraint term, wherein the data fidelity constraint term is used for guaranteeing consistency between an optimization result and the haze imaging physical model, and the regularization constraint term is used for suppressing noise and maintaining image structure continuity.
4. The wavelet-guide-based non-uniform defogging method for remote sensing images according to claim 3, wherein the energy function further comprises a reflectivity constraint term based on Retinex theory for constraining the reasonability of the clear images obtained by restoration in illumination and reflection relation.
5. The wavelet guide-based remote sensing image non-uniform defogging method according to claim 4, wherein said energy function further comprises a color consistency constraint term for constraining consistency in luminance and chromaticity of defocused image and original remote sensing image in a color space.
6. The wavelet guide-based remote sensing image non-uniform defogging method according to claim 1, wherein, And introducing a multi-scale smoothing strategy in the process of carrying out joint solution on the clear image and the medium transmittance so as to enhance the stability of the medium transmittance on spatial distribution.
7. The wavelet guide-based remote sensing image non-uniform defogging method according to claim 1, wherein, The detail enhancement processing of the high-frequency texture component comprises adaptive thresholding and contrast-limited local enhancement processing, And in the reconstruction fusion stage, a fusion mode based on residual information is adopted, and the information of the original remote sensing image is introduced into a reconstruction result so as to improve the naturalness of the defogging result.
8. A remote sensing image non-uniform defogging system based on wavelet guidance, comprising: The image acquisition module is used for acquiring remote sensing haze images; the image decomposition module is used for carrying out wavelet transformation on the remote sensing haze image and decomposing the remote sensing haze image into a low-frequency illumination component and a high-frequency texture component; The energy model construction module is used for constructing an energy function containing physical consistency constraint and multiple priori constraints by taking clear images and medium transmittance as optimization variables; The optimization solving module is used for carrying out joint solving on the clear image and the medium transmittance in an alternating optimization mode; the reconstruction fusion module is used for carrying out detail enhancement processing on the high-frequency texture component, and carrying out reconstruction fusion on the optimized low-frequency illumination component and the enhanced high-frequency texture component to obtain a defogged remote sensing image; wherein the modules cooperate to implement the wavelet guide-based remote sensing image non-uniform defogging method of any of claims 1-7.
9. A computer readable storage medium having stored thereon a computer program, which when executed by a processor, implements the wavelet guided remote sensing image non-uniform defogging method of any of claims 1-7.
10. An electronic device comprising a processor and a memory, the memory having stored therein a computer program which, when executed by the processor, causes the processor to perform the wavelet guided remote sensing image non-uniform defogging method of any of claims 1-7.

Description

Non-uniform defogging method and system for remote sensing image based on wavelet guidance Technical Field The invention relates to the technical field of remote sensing image processing, in particular to a non-uniform defogging method and system for a remote sensing image based on wavelet guidance. Background The remote sensing image is an important data form obtained by imaging the ground surface scene through aviation or aerospace imaging equipment, and is widely applied to the fields of environment monitoring, resource investigation, disaster assessment, target identification and the like. In the actual imaging process, the remote sensing image is often affected by atmospheric conditions, wherein haze, which is a typical atmospheric phenomenon, can have a significant influence on imaging quality, and causes image contrast reduction, brightness imbalance and detail information loss, so that the subsequent remote sensing image interpretation and application effects are affected. The existing remote sensing image defogging technology mainly comprises a physical model-based method and a priori hypothesis-based method. The method based on the prior assumption can achieve a certain effect in part of scenes, but often depends on a single prior condition, and problems such as brightness distortion, color deviation or excessive detail enhancement are easy to occur under the complicated haze condition. Therefore, the problem that the defogging effect is unstable under the non-uniform haze condition and the luminance balance, the structure maintainability and the color consistency are difficult to be simultaneously considered still exists in the prior art, and a remote sensing image defogging technical scheme capable of fusing various prior information and adapting to the haze space change characteristic under the constraint of a physical model is needed. Disclosure of Invention In order to solve the problems in the prior art, the embodiment of the invention provides a non-uniform defogging method and a non-uniform defogging system for a remote sensing image based on wavelet guidance. The technical scheme is as follows: in one aspect, a method for non-uniformly defogging a remote sensing image based on wavelet guidance is provided, which comprises the following steps: Processing an input remote sensing haze image, and establishing a degradation relation of the remote sensing image under a haze condition based on a physical model of haze imaging; Performing wavelet transformation on the remote sensing haze image, and decomposing the remote sensing haze image into a low-frequency illumination component and a high-frequency texture component; Constructing an energy function containing physical consistency constraint and multiple priori constraints by taking clear images and medium transmissivity as optimization variables, and performing variation optimization on the low-frequency illumination component; adopting an alternate optimization mode to carry out joint solution on the clear image and the medium transmittance; And carrying out detail enhancement processing on the high-frequency texture component, and carrying out reconstruction fusion on the optimized low-frequency illumination component and the enhanced high-frequency texture component to obtain a defogged remote sensing image. Further, the physical model based on haze imaging is an atmospheric scattering model, and is used for describing imaging relations among scene radiation, medium transmittance and atmospheric light in a remote sensing image. Further, the energy function at least comprises a data fidelity constraint term and a regularization constraint term, wherein the data fidelity constraint term is used for guaranteeing consistency between an optimization result and the haze imaging physical model, and the regularization constraint term is used for suppressing noise and keeping continuity of an image structure. Further, the energy function further comprises a reflectivity constraint term based on the Retinex theory, and the reflectivity constraint term is used for constraining the reasonability of the clear image obtained through restoration in the illuminance and reflection relation. Further, the energy function further includes a color consistency constraint term for constraining consistency in luminance and chromaticity of the defogged image and the original remote sensing image in a color space. Further, in the process of carrying out joint solution on the clear image and the medium transmittance, a multi-scale smoothing strategy is introduced so as to enhance the stability of the medium transmittance on spatial distribution. Further, the detail enhancement processing of the high-frequency texture component comprises adaptive thresholding and contrast-limited local enhancement processing, And in the reconstruction fusion stage, a fusion mode based on residual information is adopted, and the information of the original remote sensing image is int