Remote sensing image haze removal method combining rolling deep learning and Retinex theory
A deep learning and remote sensing image technology, applied in the field of remote sensing image processing, can solve the problems of insignificant haze removal effect, inaccurate parameter estimation, poor haze removal effect, etc., and achieves good haze removal effect, good visual effect, Strong effect of removing haze
Pending Publication Date: 2021-07-23
XIAN UNIV OF POSTS & TELECOMM
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0006] For the first type of method in the background technology, when dealing with complex scene haze images, the effect is not ideal, and the problem of distortion is prone to occur. At the same time, the parameter estimation is not accurate, which will make the haze removal effect not obvious. The second type of method can effectively eliminate Haze, but it is prone to distortion problems, and the third method has poor haze removal effect due to inaccurate transmittance. The present invention proposes a remote sensing image haze removal method combining rolling deep learning and Retinex theory
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0075] The method provided by the present invention will be further described below from the aspects of basic realization principle, specific realization process and comparison of test results.
[0076] Basic realization principle
[0077] The present invention first uses the DeHazeNet algorithm based on deep neural network learning to scroll the input fog and haze remote sensing image. The purpose is to gradually remove the haze in the image. The color will gradually become darker. After scrolling to a certain number of times, there will be relatively less haze in the image. When scrolling continues, the effect cannot continue to improve.
[0078] Then extract relatively stable and clear images from the processed images to continue processing, and process images in two parallel directions. One is to enhance the saturation and brightness of the image color, to keep the image tone as much as possible, and to further remove the haze. The second is to use Retinex theory to cont...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a remote sensing image haze removal method combining rolling deep learning and a Retinex theory. The method mainly comprises the following steps: 1, inputting an original haze-containing image; 2, carrying out image haze removal rolling processing on the basis of a DeHazeNet algorithm; 3, obtaining a relatively stable clear image; 4, performing saturation and brightness equalization enhancement processing on the relatively stable clear image; 5, performing haze removal processing on the relatively stable and clear image by using a Retinex theory to obtain an image; and 6, performing fusion processing on the image obtained in the step 4 and the image obtained in the step 5 to obtain a final haze-filtered image. According to the method, the haze in the image can be effectively removed, more importantly, rich detail information and the high-contrast image can be effectively kept, the visual effect of the image is good, color distortion is little, and the method has good popularization and application value.
Description
technical field [0001] The invention belongs to the technical field of remote sensing image processing, and in particular relates to a remote sensing image haze removal method combining rolling deep learning and Retinex theory. Background technique [0002] The cause of haze weather is that the air contains a large number of aerosols and floating dust particles, which reduces the transmittance of light in the atmosphere and reduces the light reflected back from the scene area, resulting in a strong sense of blur in the image taken in the scene area, showing a blurred image. Gray feeling. This has a great impact on outdoor shooting, automatic driving, target tracking, navigation and optical remote sensing, so removing the effects of haze in optical images and videos is an important part of image preprocessing. There are many methods for removing haze in images, and these methods can be roughly divided into three categories. [0003] The first type of method is from the pers...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More IPC IPC(8): G06T5/00G06T5/40G06T5/50
CPCG06T5/40G06T5/50G06T2207/10032G06T2207/20081G06T2207/20084G06T2207/20221G06T2207/20024G06T5/73
Inventor黄世奇徐杰王文庆卢莹孙柯吝张茹程磊
OwnerXIAN UNIV OF POSTS & TELECOMM