CN111539876A - Self-adaptive multi-scale Retinex dark scene visibility improving method based on polarization imaging - Google Patents
Self-adaptive multi-scale Retinex dark scene visibility improving method based on polarization imaging Download PDFInfo
- Publication number
- CN111539876A CN111539876A CN202010041827.2A CN202010041827A CN111539876A CN 111539876 A CN111539876 A CN 111539876A CN 202010041827 A CN202010041827 A CN 202010041827A CN 111539876 A CN111539876 A CN 111539876A
- Authority
- CN
- China
- Prior art keywords
- scale
- visibility
- image
- dark
- dark scene
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 68
- 230000010287 polarization Effects 0.000 title claims abstract description 53
- 238000003384 imaging method Methods 0.000 title claims abstract description 35
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 44
- 230000000694 effects Effects 0.000 claims abstract description 27
- 239000011159 matrix material Substances 0.000 claims abstract description 9
- 238000005259 measurement Methods 0.000 claims description 14
- 230000003044 adaptive effect Effects 0.000 claims description 13
- 230000000007 visual effect Effects 0.000 claims description 9
- 230000001965 increasing effect Effects 0.000 claims description 6
- 238000002834 transmittance Methods 0.000 claims description 6
- 230000002708 enhancing effect Effects 0.000 claims description 5
- 230000008033 biological extinction Effects 0.000 claims description 4
- 230000003247 decreasing effect Effects 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 230000000052 comparative effect Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Classifications
-
- G06T5/94—
-
- G06T5/73—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20004—Adaptive image processing
- G06T2207/20008—Globally adaptive
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20172—Image enhancement details
- G06T2207/20208—High dynamic range [HDR] image processing
Abstract
The invention discloses a polarization imaging-based self-adaptive multi-scale Retinex dark scene visibility improving method, which comprises the following steps of: s1: constructing a dark scene polarization imaging system, and acquiring a polarization image by using the dark scene polarization imaging system; s2: carrying out defogging operation in advance to obtain a defogging image and obtain a transmissivity matrix; s3: performing dark scene enhancement on the defogging image by using a multi-scale Retinex algorithm, and calculating the visibility of the defogging image before and after the dark scene enhancement by using the multi-scale Retinex algorithm; s4: and evaluating the lifting result of the algorithm, if the lifting effect reaches the standard, ending the operation, saving the picture, if the lifting effect does not reach the standard, adjusting the scale of the Retinex algorithm in the step S3, and returning to the step S3. The invention provides a self-adaptive multi-scale Retinex dark scene visibility improving method based on polarization imaging, which is strong in anti-interference capability, good in enhancement effect and self-adaptive.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a polarization imaging-based self-adaptive multi-scale Retinex dark scene visibility improving method.
Background
The computer vision technology is a technology of replacing a human vision physiological system with a system consisting of a camera and a computer, can identify, track and measure a target, and can further process an image. The image shot under the environment with weak light intensity can be enhanced to obtain the effect far superior to the observation of human eyes, and the expansion and extension of a visual system are realized.
Retinex image enhancement is a new tone mapping technique for high dynamic range images. The basic theory is that the color of the object is determined by the reflection ability of the object to long-wave (red), medium-wave (green) and short-wave (blue) light, and not by the absolute value of the intensity of the reflected light, and the color of the object is not affected by illumination nonuniformity and has uniformity, i.e., Retinex is based on color sense uniformity (color constancy). Unlike the traditional linear and nonlinear methods which can only enhance a certain feature of an image, Retinex can balance three aspects of dynamic range compression, edge enhancement and color constancy, so that various different types of images can be adaptively enhanced.
In the process of collecting dark scene images, on one hand, fog interference exists inevitably, so that color distortion and edge contour blurring of objects in the images are caused, and the enhancement effect is influenced; in addition, the single-scale Retinex algorithm has the defects that the enhancement effect of a bright area is not obvious, and the original image has the influence of noise on the image detail enhancement. For some smooth objects, even in a dark scene environment, the phenomenon of light reflection easily occurs, and the Retinex enhancement algorithm is difficult to enhance the highlight part. The scale selection of the multi-scale Retinex algorithm is generally fixed, and the enhancement effect of different dark scene original images is different.
Therefore, on one hand, the image is defogged before the dark scene Retinex algorithm is enhanced, so that the enhancement effect can be effectively improved; on the one hand, the polaroid is used in image acquisition, so that the influence of highlight on an enhancement algorithm can be inhibited. On the other hand, the image is enhanced by using the self-adaptive multi-scale Retinex algorithm, the advantages of the single-scale Retinex algorithm in three scales of high, medium and low are achieved, the problems that the enhancement effect of a bright area is not obvious in the single-scale Retinex algorithm, and the noise of an original image influences the image detail enhancement and subsequent research are solved, and meanwhile, the scale parameters are updated in real time, so that the optimal enhancement effect is achieved. The dark scene contrast improving method based on the Retinex principle further improves the enhancement effect and reduces the distortion of the enhancement result on the premise of ensuring the integrity of image information.
Chinese patent publication No. CN107730470A, published 2018, month 02 and 23, entitled an improved Retinex image enhancement method for explicitly expressing multi-scale and histogram truncation, which proposes an improved multi-scale (>3) Retinex enhancement method based on mathematical analysis and derivation for low visibility aerial images, and in addition, introduces a histogram truncation technique as a means of image post-processing to map the output of multi-scale Retinex into a dynamic range for display; this application does not carry out visibility to the picture of handling and judges, and this just makes can not guarantee to export the picture that visibility is high, can not use the polaroid to filter when gathering the picture moreover, and the visibility that causes the image of handling is not good.
Disclosure of Invention
The invention provides a self-adaptive multi-scale Retinex dark scene visibility improving method based on polarization imaging, which has strong anti-interference capability and good enhancement effect and has self-adaptation, in order to overcome the problems that the Retinex algorithm in the prior art has poor enhancement effect on a foggy environment and a highlight part and the optimal selection of scale parameters in the multi-scale Retinex algorithm is different from graph to graph.
In order to achieve the purpose, the invention adopts the following technical scheme:
the technical scheme adopted by the invention for solving the technical problems is as follows: a self-adaptive multi-scale Retinex dark scene visibility improving method based on polarization imaging comprises the following steps:
s1: constructing a dark scene polarization imaging system, and acquiring a polarization image by using the dark scene polarization imaging system;
s2: carrying out defogging operation in advance to obtain a defogging image and obtain a transmissivity matrix;
s3: carrying out dark scene enhancement on the defogging image by using a multi-scale Retinex algorithm;
s4: calculating the visibility of the defogging image before and after dark scene enhancement by using a multi-scale Retinex algorithm;
s5: and evaluating the lifting result of the algorithm, if the lifting effect reaches the standard, ending the operation, saving the picture, if the lifting effect does not reach the standard, adjusting the scale of the Retinex algorithm in the step S3, and returning to the step S3. The self-adaptive multi-scale Retinex visibility improving method provided by the invention can automatically adjust scale parameters, effectively improves the visibility improving effect on the premise of not damaging the simplicity and convenience of system operation, provides a dark scene image with low fog interference by using a dark primary color prior method aiming at the condition that the enhanced image is distorted due to the existence of fog in the original dark scene image, and obtains the best material image by adaptively adjusting the angle of the polaroid according to different images by using a self-adaptive polarization imaging method aiming at the condition that the enhanced effect of the dark scene image enhancing method on highlight parts is poor and the visibility is interfered in the evaluation of the dark scene image, thereby further improving the visibility improving effect.
Preferably, the specific process of step S1 is as follows: the dark scene polarization imaging system comprises a camera, a polarizing film, a motor and a controller, wherein the polarizing film of the camera lens is arranged in front of the camera lens, the motor is connected with the polarizing film, the control end of the controller is connected with the controlled end of the motor, the controller controls the rotating speed of the motor, the motor drives the polarizing film to change the polarization angle, a polarization angle threshold value is set, each time the polarization angle of the polarizing film changes one polarization angle threshold value, the camera shoots a picture, and a scene picture with the best intensity is selected from scene pictures shot by the camera to serve as a dark scene picture. .
Preferably, the specific process of step S2 is as follows: and establishing an atmospheric scattering model, defogging the image by using dark primary color prior, obtaining a statistical rule of the outdoor fog-free image database by using the dark primary color prior, and obtaining a transmissivity matrix t. For the case where the presence of fog in the original dark scene image results in distortion of the enhanced image, a dark scene image with low fog interference can be provided using a dark channel prior method.
Preferably, the step S3 includes the steps of:
s31: carrying out visibility measurement on the defogging image, recording the result as V1, and setting a threshold value VthIf V1 > VthIf yes, ending the dark scene visibility improving algorithm; if V1 < VthStarting the next step;
s32: enhancing the dark scene image subjected to defogging by using a multi-scale Retinex algorithm, wherein the multi-scale Retinex algorithm is divided into three scales of high scale, medium scale and low scale;
s33: and carrying out visibility measurement on the enhanced image, solving the visibility of the enhanced image, and recording the result as V2.
Preferably, the specific process of the visibility measurement is as follows: visibility V is calculated using Koschmieder's law, and is expressed as:
wherein the visual contrast threshold is a physical quantity related to the visual characteristics of human eyes, and β is an extinction coefficient with a value of t (z) e-βd(z)It is determined that d (z) is a distance between the target and the image plane at the position z, and t (z) represents a transmittance at z.
Preferably, the step S4 includes the following specific steps: comparing V1 and V2, the visibility improvement ratio Rate is calculated as: and setting an experience threshold T when the Rate is (V2-V1)/V1, automatically adjusting the scale parameter if the Rate is less than T, returning to the step S3 after the adaptive adjustment of the scale, and ending and saving the picture if the Rate is more than T.
Preferably, the value range is 0.02-0.05.
Preferably, the process of automatically adjusting the scale parameter is as follows: and setting an adjustment threshold, increasing the high scale by one adjustment threshold, and decreasing the low scale by one adjustment threshold.
Preferably, if all the automatically adjusted scale parameters cannot enable the Rate to be greater than T, the picture with the maximum Rate is selected as the final picture.
Therefore, the invention has the following beneficial effects: (1) the method for improving visibility of the adaptive multi-scale Retinex can automatically adjust scale parameters, effectively improves the visibility improvement effect on the premise of not damaging the simplicity and convenience of system operation, provides a dark scene image with low fog interference by using a dark primary color prior method aiming at the condition that the enhanced image is distorted due to the existence of fog in an original dark scene image, and obtains the best material image by using an adaptive polarization imaging method aiming at the condition that the enhanced effect of the dark scene image enhancement method on a highlight part is poor and the visibility evaluation has interference so as to adaptively adjust the angle of a polaroid sheet according to different images, thereby further improving the visibility improvement effect;
drawings
FIG. 1 is a flow chart of the present invention
FIG. 2 is a comparative diagram of case 1 of the present invention
FIG. 3 is a comparative diagram of case 2 of the present invention
FIG. 4 is a comparative diagram of case 3 of the present invention
FIG. 5 is a schematic diagram of a local area selection location according to the present invention
FIG. 6 is a schematic diagram of a local area selection principle according to the present invention
FIG. 7 is a table of contrast analysis of case 1 of the present invention
FIG. 8 is a table of contrast analysis of case 2 of the present invention
FIG. 9 is a table of contrast analysis of case 3 of the present invention
Detailed Description
The invention is further described with reference to the following detailed description and accompanying drawings.
Example (b): a self-adaptive multi-scale Retinex dark scene visibility improving method based on polarization imaging is shown in FIG. 1: the method comprises the following steps:
s1: constructing a dark scene polarization imaging system, and acquiring a polarization image by using the dark scene polarization imaging system; the specific process is as follows: the dark scene polarization imaging system comprises a camera, a polarizing film, a motor and a controller, wherein the polarizing film of the camera lens is arranged in front of the camera lens, the motor is connected with the polarizing film, the control end of the controller is connected with the controlled end of the motor, the controller controls the rotating speed of the motor, the motor drives the polarizing film to change the polarization angle, a polarization angle threshold is set, the camera shoots a picture when the polarization angle of the polarizing film changes one polarization angle threshold, and a scene picture with the best intensity is selected from scene pictures shot by the camera to serve as a dark scene picture;
s2: carrying out defogging operation in advance to obtain a defogging image and obtain a transmissivity matrix; the specific process is as follows: establishing an atmospheric scattering model, defogging an image by using dark primary color prior, obtaining a transmittance matrix t according to a statistical rule of an outdoor fog-free image database by using the dark primary color prior;
s3: carrying out dark scene enhancement on the defogging image by using a multi-scale Retinex algorithm;
s31: carrying out visibility measurement on the defogging image, recording the result as V1, and setting a threshold value VthIf V1 > VthIf yes, ending the dark scene visibility improving algorithm; if V1 < VthStarting the next step;
s32: enhancing the dark scene image subjected to defogging by using a multi-scale Retinex algorithm, wherein the multi-scale Retinex algorithm is divided into three scales of high scale, medium scale and low scale;
s33: carrying out visibility measurement on the enhanced image, solving the visibility of the enhanced image, and recording the result as V2; the specific process of visibility measurement is as follows: visibility V is calculated using Koschmieder's law, and is expressed as:
in the formula: the visual contrast threshold is a physical quantity related to the visual characteristics of human eyes, and the value range is0.02-0.05, β is extinction coefficient, and t (z) is e-βd(z)Determining d (z) is the distance between the target object and the imaging surface at the position z, and t (z) represents the transmittance at the position z; carrying out visibility measurement on the collected original image of the polarized dark scene, recording the result as V1, and setting a threshold value VthIf V1 > VthClosing the dark scene visibility improving algorithm; if V1 < VthImplementing a self-adaptive multi-scale Retinex algorithm to obtain an enhanced image, solving the visibility of the enhanced image, and recording the result as V2;
s4: evaluating the algorithm lifting result, if the lifting effect reaches the standard, ending the operation, saving the picture, if the lifting effect does not reach the standard, adjusting the scale of the Retinex algorithm in the step S3, and returning to the step S3; the specific process is as follows: comparing V1 and V2, the visibility improvement ratio Rate is calculated as: setting an experience threshold T when the Rate is (V2-V1)/V1, automatically adjusting the scale parameter if the Rate is less than T, returning to the step S3 after the adaptive adjustment of the scale, ending if the Rate is more than T, and saving the picture, wherein the process of automatically adjusting the scale parameter is as follows: and setting an adjustment threshold, increasing the high scale by one adjustment threshold, decreasing the low scale by one adjustment threshold, and selecting the picture with the maximum Rate as the final picture if the Rate cannot be larger than T by all the automatically adjusted scale parameters.
The invention will be further illustrated with reference to the following specific embodiments and the accompanying drawings:
s1: constructing a dark scene polarization imaging system, and acquiring a polarization image by using the dark scene polarization imaging system; the specific process is as follows: the dark scene polarization imaging system comprises a camera, a polarizing film, a motor and a controller, wherein the polarizing film of the camera lens is arranged in front of the camera lens, the motor is connected with the polarizing film, the control end of the controller is connected with the controlled end of the motor, the controller controls the rotating speed of the motor, the motor drives the polarizing film to change the polarization angle, the threshold value of the polarization angle is set to be 5 degrees, when the polarization angle of the polarizing film is changed by 5 degrees, the camera shoots a picture, and a scene picture with the best intensity is selected from scene pictures shot by the camera to be used as a dark scene picture, such;
s2: carrying out defogging operation in advance to obtain a defogging image and obtain a transmissivity matrix; the specific process is as follows: establishing an atmospheric scattering model, defogging an image by using dark channel prior, obtaining a statistical rule of an outdoor fog-free image database by using the dark channel prior, and obtaining a transmissivity matrix t, wherein the atmospheric scattering model is as follows: g (z) ═ I (z) t (z) + a (1-t (z)), z is the coordinate position/spatial position, g is the intensity of the observed image, I is the original observed intensity (corresponding to the clear image to be defogged), a is the sky light, t (z) is the transmittance at z;
s3: carrying out dark scene enhancement on the defogging image by using a multi-scale Retinex algorithm;
s31: carrying out visibility measurement on the defogging image, recording the result as V1, and setting a threshold value VthIs 500, if V1 > VthIf yes, ending the dark scene visibility improving algorithm; if V1 < VthStarting the next step;
s32: enhancing the dark scene image subjected to defogging by using a multi-scale Retinex algorithm, wherein the multi-scale Retinex algorithm is divided into three scales of high scale, medium scale and low scale, and can be expressed as follows:
wherein, (x, y) is the coordinate of each pixel point of the image, i ∈ R, G, B represent three color channels, i equals 1 to process the gray image, i equals 3 to process the color image, W equals 3 to process the gray image, andnis represented by the formulan(x, y) corresponding to a weight, the sum of n weights being 1, namely:
Fn(x, y) is a surround function and has:
wherein
Is a normalization factor, cnIs a scale constant;
s33: carrying out visibility measurement on the enhanced image, solving the visibility of the enhanced image, and recording the result as V2; the specific process of visibility measurement is as follows: energy calculation Using Koschmieder LawVisibility V, the expression for visibility V is:
wherein the visual contrast threshold is a physical quantity related to human visual features, and has a value range of 0.02-0.05, and β is an extinction coefficient represented by t (z) e-βd(z)Determining d (z) is the distance between the target object and the imaging surface at the position z, and t (z) represents the transmittance at the position z; carrying out visibility measurement on the collected original image of the polarized dark scene, recording the result as V1, and setting a threshold value VthIf V1 > VthClosing the dark scene visibility improving algorithm; if V1 < VthImplementing a self-adaptive multi-scale Retinex algorithm to obtain an enhanced image, solving the visibility of the enhanced image, and recording the result as V2; s4: evaluating the algorithm lifting result, if the lifting effect reaches the standard, ending the operation, saving the picture, if the lifting effect does not reach the standard, adjusting the scale of the Retinex algorithm in the step S3, and returning to the step S3; the specific process is as follows: comparing V1 and V2, the visibility improvement ratio Rate is calculated as: setting an experience threshold T to be 0.3 when the Rate is (V2-V1)/V1, automatically adjusting the scale parameter if the Rate is less than T, returning to the step S3 after the adaptive adjustment of the scale, ending if the Rate is more than T, and saving the picture, wherein the process of automatically adjusting the scale parameter is as follows: setting an adjusting threshold value as 5, increasing the high scale by 5, decreasing the low scale by 5, and if all the automatically adjusted scale parameters can not enable the Rate to be larger than T, selecting the picture with the maximum Rate as a final picture.
Selecting three places to shoot three groups of case photos, respectively using a traditional Retinex enhancement method and the method of the invention to process, and analyzing the contrast improvement condition, wherein the photo processing condition of case 1 is shown as figure 2, and in figure 2, (a) is a shooting original image, (b) is a picture obtained by using the traditional Retinex enhancement method, and (c) is a picture obtained by using the method of the invention; case 2 photo processing is shown in fig. 3, (a) is the original image taken, (b) is the image obtained by using the conventional Retinex enhancement method, and (c) is the image obtained by using the method of the present invention; case 3 photo processing is shown in fig. 4, where (a) in fig. 4 is a photograph of an original image, (b) is a photograph obtained by using a conventional Retinex enhancement method, and (c) is a photograph obtained by using the method of the present invention;
and analyzing the local part of the image, and adopting a local comparison calculation mode. As shown in fig. 5, the straight-line distance from the calibrated building to the shooting point in fig. 5(a) is 0.5km, the straight-line distance from the shooting point in fig. 5(b) is 1km, and the straight-line distance from the shooting point in fig. 5(c) is 3 km.
And (4) examining the contrast improvement by using the contrast of the local area where the target is located so as to replace the contrast of the global whole image. Since such local examination indexes are more scientific.
The formula for calculating the contrast is:
l1 represents the mean of the gradation values of the selected target area image, and L2 represents the mean of the gradation values of the selected target vicinity area image. As shown in fig. 6, L1 represents the mean value of the gradation values of the small frame selection region image, and L2 represents the mean value of the gradation values of the large frame selection region (non-small frame region) image.
The selection of the small frame selection area and the large frame selection area which represent the local area has the following principle:
selecting a more prominent object in the small frame selection area, wherein the object does not contain background components or only contains few backgrounds;
the large frame selection area selects an area extending outwards from the small frame selection area, wherein the area comprises the small frame selection area and also comprises a background component; the area of the large frame selection area is about 4 times of the area of the small frame selection area, and the small frame selection area is located in the center of the large frame selection area.
As shown in fig. 7, the contrast ratio of fig. 2(a) after being processed by the method of the present invention is improved by 32.8%, while the contrast ratio is improved by only 5.9% by using the conventional Retinex enhancement method.
As shown in fig. 8, the contrast ratio of fig. 3(a) after being processed by the method of the present invention is improved by 63.3%, while the contrast ratio is improved by only 60.4% by using the conventional Retinex enhancement method.
As shown in FIG. 9, the contrast ratio of FIG. 4(a) after the processing of the present invention is increased by 105.6%, while the contrast ratio is increased by-37.2% by using the conventional Retinex enhancement method.
Claims (9)
1. A self-adaptive multi-scale Retinex dark scene visibility improving method based on polarization imaging is characterized by comprising the following steps:
s1: constructing a dark scene polarization imaging system, and acquiring a polarization image by using the dark scene polarization imaging system;
s2: carrying out defogging operation in advance to obtain a defogging image and obtain a transmissivity matrix;
s3: performing dark scene enhancement on the defogging image by using a multi-scale Retinex algorithm, and calculating the visibility of the defogging image before and after the dark scene enhancement by using the multi-scale Retinex algorithm;
s4: and evaluating the lifting result of the algorithm, if the lifting effect reaches the standard, ending the operation, saving the picture, if the lifting effect does not reach the standard, adjusting the scale of the Retinex algorithm in the step S3, and returning to the step S3.
2. The method for improving visibility in dark scenes with adaptive multi-scale Retinex based on polarization imaging as claimed in claim 1, wherein the specific process of step S1 is as follows: the dark scene polarization imaging system comprises a camera, a polarizing film, a motor and a controller, wherein the polarizing film of the camera lens is arranged in front of the camera lens, the motor is connected with the polarizing film, the control end of the controller is connected with the controlled end of the motor, the controller controls the rotating speed of the motor, the motor drives the polarizing film to change the polarization angle, a polarization angle threshold value is set, each time the polarization angle of the polarizing film changes one polarization angle threshold value, the camera shoots a picture, and a scene picture with the best intensity is selected from scene pictures shot by the camera to serve as a dark scene picture.
3. The method for improving visibility in dark scenes with adaptive multi-scale Retinex based on polarization imaging as claimed in claim 1, wherein the specific process of step S2 is as follows: and establishing an atmospheric scattering model, defogging the image by using dark primary color prior, obtaining a statistical rule of the outdoor fog-free image database by using the dark primary color prior, and obtaining a transmissivity matrix t.
4. The method for improving visibility in dark scenes with adaptive multi-scale Retinex based on polarization imaging as claimed in claim 1, wherein the step S3 comprises the following steps:
s31: carrying out visibility measurement on the defogging image, recording the result as V1, and setting a threshold value VthIf V1 > VthIf yes, ending the dark scene visibility improving algorithm; if V1 < VthStarting the next step;
s32: enhancing the dark scene image subjected to defogging by using a multi-scale Retinex algorithm, wherein the multi-scale Retinex algorithm is divided into three scales of high scale, medium scale and low scale;
s33: and carrying out visibility measurement on the enhanced image, solving the visibility of the enhanced image, and recording the result as V2.
5. The method for improving the visibility of the adaptive multi-scale Retinex dark scene based on the polarization imaging as claimed in claim 4, wherein the visibility measurement comprises the following specific processes: visibility V is calculated using Koschmieder's law, and is expressed as:
wherein the visual contrast threshold is a physical quantity related to the visual characteristics of human eyes, and β is an extinction coefficient with a value of t (z) e-βd(z)It is determined that d (z) is a distance between the target and the image plane at the position z, and t (z) represents a transmittance at z.
6. The method for improving visibility in dark scenes with adaptive multi-scale Retinex based on polarization imaging according to claim 1 or 4, wherein the step S4 comprises the following steps: comparing V1 and V2, the visibility improvement ratio Rate is calculated as: and setting an experience threshold T when the Rate is (V2-V1)/V1, automatically adjusting the scale parameter if the Rate is less than T, returning to the step S3 after the adaptive adjustment of the scale, and ending and saving the picture if the Rate is more than T.
7. The method for improving the visibility of the adaptive multi-scale Retinex dark scene based on the polarization imaging as claimed in claim 5, wherein the value range is 0.02-0.05.
8. The method for improving visibility in dark scenes with adaptive multi-scale Retinex based on polarization imaging as claimed in claim 6, wherein the process of automatically adjusting scale parameters is as follows: and setting an adjustment threshold, increasing the high scale by one adjustment threshold, and decreasing the low scale by one adjustment threshold.
9. The method for improving visibility in a dark scene based on polarization imaging according to claim 6, wherein if all the automatically adjusted scale parameters cannot make the Rate > T, the picture with the maximum Rate is selected as the final picture.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010041827.2A CN111539876A (en) | 2020-01-15 | 2020-01-15 | Self-adaptive multi-scale Retinex dark scene visibility improving method based on polarization imaging |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010041827.2A CN111539876A (en) | 2020-01-15 | 2020-01-15 | Self-adaptive multi-scale Retinex dark scene visibility improving method based on polarization imaging |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111539876A true CN111539876A (en) | 2020-08-14 |
Family
ID=71978521
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010041827.2A Pending CN111539876A (en) | 2020-01-15 | 2020-01-15 | Self-adaptive multi-scale Retinex dark scene visibility improving method based on polarization imaging |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111539876A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102539383A (en) * | 2012-01-09 | 2012-07-04 | 北京大学 | Method for identifying fog and haze in scattering type visibility meter, and method for measuring visibility |
| CN104851144A (en) * | 2015-05-29 | 2015-08-19 | 山东鲁能智能技术有限公司 | Inspection robot image processing system and method with self-adaptive polarizer adjusting function |
| CN106548463A (en) * | 2016-10-28 | 2017-03-29 | 大连理工大学 | Based on dark and the sea fog image automatic defogging method and system of Retinex |
| CN108871576A (en) * | 2017-05-12 | 2018-11-23 | 长春理工大学 | A kind of polarization imaging device based on calculating optical |
-
2020
- 2020-01-15 CN CN202010041827.2A patent/CN111539876A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102539383A (en) * | 2012-01-09 | 2012-07-04 | 北京大学 | Method for identifying fog and haze in scattering type visibility meter, and method for measuring visibility |
| CN104851144A (en) * | 2015-05-29 | 2015-08-19 | 山东鲁能智能技术有限公司 | Inspection robot image processing system and method with self-adaptive polarizer adjusting function |
| CN106548463A (en) * | 2016-10-28 | 2017-03-29 | 大连理工大学 | Based on dark and the sea fog image automatic defogging method and system of Retinex |
| CN108871576A (en) * | 2017-05-12 | 2018-11-23 | 长春理工大学 | A kind of polarization imaging device based on calculating optical |
Non-Patent Citations (3)
| Title |
|---|
| 丁虎: "低能见度图像的清晰化处理研究" * |
| 周悦;: "基于Retinex理论的雾霭天气图像增强" * |
| 徐敏 等: "基于场景深度的雾天图像能见度检测算法" * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Adaptive image enhancement method for correcting low-illumination images | |
| Wang et al. | Simple low-light image enhancement based on Weber–Fechner law in logarithmic space | |
| CN107424133B (en) | Image defogging method and device, computer storage medium and mobile terminal | |
| CN111292258B (en) | Image defogging method based on dark channel prior and bright channel prior | |
| CN109754377A (en) | A kind of more exposure image fusion methods | |
| CN110570360B (en) | Retinex-based robust and comprehensive low-quality illumination image enhancement method | |
| CN106971153B (en) | Illumination compensation method for face image | |
| CN105447825B (en) | Image defogging method and its system | |
| CN111968065B (en) | Self-adaptive enhancement method for image with uneven brightness | |
| CN110163807B (en) | Low-illumination image enhancement method based on expected bright channel | |
| CN116309559B (en) | Intelligent identification method for production flaws of medium borosilicate glass | |
| CN108133462B (en) | Single image restoration method based on gradient field region segmentation | |
| CN110852956A (en) | Method for enhancing high dynamic range image | |
| CN115578297A (en) | Generalized attenuation image enhancement method for self-adaptive color compensation and detail optimization | |
| CN113313702A (en) | Aerial image defogging method based on boundary constraint and color correction | |
| CN109325918A (en) | Image processing method and device and computer storage medium | |
| CN115883755A (en) | Multi-exposure image fusion method under multi-type scene | |
| CN111311502A (en) | Method for processing foggy day image by using bidirectional weighted fusion | |
| CN110852971B (en) | Video defogging method based on dark channel prior and Retinex and storage medium | |
| CN109949239B (en) | Self-adaptive sharpening method suitable for multi-concentration multi-scene haze image | |
| CN111539876A (en) | Self-adaptive multi-scale Retinex dark scene visibility improving method based on polarization imaging | |
| Xie et al. | Image defogging method combining light field depth estimation and dark channel | |
| CN110852977B (en) | Image enhancement method for fusing edge gray level histogram and human eye visual perception characteristics | |
| Xi et al. | Enhancement of unmanned aerial vehicle image with shadow removal based on optimized retinex algorithm | |
| Chen et al. | Low-light image enhancement based on Double component processing of Retinex model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200814 |
|
| RJ01 | Rejection of invention patent application after publication |