Unlock instant, AI-driven research and patent intelligence for your innovation.

Rapid image defogging method based on depth estimation prior and electronic equipment

A technology of depth estimation and depth map, which is applied in the field of image processing, can solve problems such as weak defogging, low production efficiency, and sky distortion, and achieve the effects of reducing time cost, reducing computing power consumption, and simplifying operations

Pending Publication Date: 2022-03-25
上海赛昉科技有限公司
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] The existing deep learning dehazing algorithm has high requirements on hardware and is difficult to marginalize; the prior-based dehazing algorithm may not be strong in dehazing, or produce distortion in local areas such as the sky, and it is difficult to make a good trade-off between the two relationship; the previous two algorithms consume more resources and have low production efficiency

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 more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Rapid image defogging method based on depth estimation prior and electronic equipment
  • Rapid image defogging method based on depth estimation prior and electronic equipment
  • Rapid image defogging method based on depth estimation prior and electronic equipment

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0060] refer to figure 1 As shown, this embodiment provides a fast image defogging method based on depth estimation prior, including the following steps:

[0061] S1 acquires the data set of the foggy image and its depth map pair after adding fog to the fog-free image, and converts it into a form that can be processed by digital image;

[0062] S2 uses multiple foggy images to construct a depth estimation model for dehazing, and generates an initial depth map corresponding to the input image;

[0063] S3 uses the atmospheric light estimation algorithm to estimate the three-channel atmospheric light value corresponding to the input image, and at the same time uses minimum value filtering, down-sampling and guided filtering to refine the depth map;

[0064] S4 brings the refined depth map and atmospheric light value into the image restoration formula, and dehazes the fogged image and outputs it.

[0065] This embodiment is constructed to use a large number of foggy images to s...

Embodiment 2

[0069] see figure 2 As shown, this embodiment provides a fast image defogging algorithm based on depth estimation prior, including the following steps:

[0070] Transfer the input image from the RGB color space to the HSV color space, and use the depth estimation prior to generate a scene depth map;

[0071] The input image is processed by a simplified quadtree algorithm to obtain the atmospheric light value of the RGB three-channel;

[0072] Perform minimum value filtering and smoothing on the scene depth map, and then use guided filtering to perform edge-preserving smoothing filtering, and refine the edges in the grayscale image of the input image to the depth map;

[0073] The atmospheric scattering model is used to restore the image, and the processed dehazed image is output.

[0074] Further as a preferred embodiment, the step of generating a scene depth map using a depth estimation prior includes the following steps:

[0075] By observing the correlation between the ...

Embodiment 3

[0089] In terms of specific implementation, on the basis of a fast image defogging algorithm based on depth estimation priors, this embodiment provides a simplified quadtree algorithm to process the input image to obtain the RGB three-channel atmospheric light value. Specific steps are as follows:

[0090] Convert the input image to a grayscale image;

[0091] Divide the grayscale image into four rectangular areas, and use the average value of the area as the score in each area;

[0092] Select the area with the highest score as a new area, repeat the previous step, and use the intermediate variable to store the original RGB image block corresponding to the area with the highest score at this time, and erase the RGB image block value stored in the previous iteration;

[0093] After repeating the previous step five times, the area of ​​the obtained image block is already smaller than 0.1% of the size of the original image, and the average value of the RGB three channels in the...

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

No PUM Login to View More

Abstract

The invention relates to the technical field of image processing, in particular to a depth estimation prior-based fast image defogging method and electronic equipment, and the method comprises the following steps: obtaining a foggy image after a fogless image is added with fog and a data set of a depth map pair of the foggy image, and converting the data set into a form capable of carrying out digital image processing; constructing a depth estimation model for defogging by using the plurality of fog-containing images, and generating an initial depth map corresponding to the input image; estimating a three-channel atmospheric light value corresponding to the input image by using an atmospheric light estimation algorithm, and refining the depth map by using minimum filtering, down-sampling and guided filtering; and substituting the refined depth map and the atmospheric light value into an image restoration formula, and carrying out defogging operation on the fog-containing image and outputting the fog-containing image. According to the method, the boundary between defogging and non-defogging can be better processed, up-down sampling is introduced in the algorithm process, the operation is optimized and simplified, and the computing power consumption and time consumption of the algorithm are reduced.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a fast image defogging method based on depth estimation prior and electronic equipment. Background technique [0002] In the field of computer vision, the sharpening of images in foggy scenes is an important issue, which is crucial for subsequent operations (such as visual interpretation and computer vision analysis). In the range of visible light imaging, due to the influence of fog, dust and other particles in the atmosphere, as the transmission distance increases, the light reflected by the object and reaching the photosensitive sheet of the camera is very weak, making the image blurred, especially in foggy weather. Visibility is very low, and the haze of images taken in foggy weather is even more serious. Therefore, the defogging technology has great practical significance. [0003] The existing deep learning dehazing algorithm has high requirements on hardware and...

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
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06T5/00G06T5/20G06N20/00G06N3/08G06K9/62G06V10/774
CPCG06T5/20G06N3/08G06N20/00G06T2207/20081G06T2207/30192G06F18/214G06T5/73
Inventor 詹志康雷雨范赐恩俞虎吴圆梅
Owner 上海赛昉科技有限公司