CN112884689B - Method for removing high light of strong reflection surface image - Google Patents
Method for removing high light of strong reflection surface image Download PDFInfo
- Publication number
- CN112884689B CN112884689B CN202110214047.8A CN202110214047A CN112884689B CN 112884689 B CN112884689 B CN 112884689B CN 202110214047 A CN202110214047 A CN 202110214047A CN 112884689 B CN112884689 B CN 112884689B
- Authority
- CN
- China
- Prior art keywords
- light source
- image
- gray
- images
- adjusting
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 230000004927 fusion Effects 0.000 claims abstract description 18
- 238000001914 filtration Methods 0.000 claims abstract description 17
- 238000012545 processing Methods 0.000 claims abstract description 8
- 238000007499 fusion processing Methods 0.000 claims abstract description 4
- 238000003384 imaging method Methods 0.000 claims description 8
- 230000000694 effects Effects 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000012546 transfer Methods 0.000 claims description 5
- 101100027969 Caenorhabditis elegans old-1 gene Proteins 0.000 claims description 3
- 230000000007 visual effect Effects 0.000 claims description 3
- 239000000919 ceramic Substances 0.000 description 16
- 239000000463 material Substances 0.000 description 11
- 239000011521 glass Substances 0.000 description 7
- 238000005286 illumination Methods 0.000 description 4
- 238000004806 packaging method and process Methods 0.000 description 4
- 230000009466 transformation Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 239000002184 metal Substances 0.000 description 3
- 229910052751 metal Inorganic materials 0.000 description 3
- 150000002739 metals Chemical class 0.000 description 3
- 229920003023 plastic Polymers 0.000 description 3
- 239000004033 plastic Substances 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000007429 general method Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration by the use of more than one image, e.g. averaging, subtraction
-
- G06T5/94—
-
- 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
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
Abstract
The invention belongs to the technical field of image processing, and relates to a method for removing high light of an image on a strong reflection surface, which comprises the following steps: 1) Image acquisition is carried out on an object to be shot in a mode of fixing a camera view angle and a multi-angle light source, images of a plurality of light source angles under the same view angle are obtained, and the position of a central point of the object to be shot is determined; the center point position is a reference position; 2) Acquiring three images with different light source angles of an object to be shot according to the obtained reference position in the step 1); 3) Carrying out fusion processing on the images obtained in the step 2) by adopting a fusion algorithm to obtain fused images; 4) And (3) carrying out homomorphic filtering treatment on the fused image obtained in the step (3), adjusting the overall brightness of the fused image, and improving the contrast of the fused image. The invention provides a high-efficiency and universal method for removing high-light of a strong reflection surface image.
Description
Technical Field
The invention belongs to the technical field of image processing, relates to a method for removing high light of an image, and particularly relates to a method for removing high light of an image with a strong reflection surface.
Background
At present, when products made of materials such as ceramics, metals and plastics are shot by using a camera, the problems of larger area, stronger specular reflection, local texture loss and the like are caused, and meanwhile, the uneven illumination of the products during imaging is caused because some products have more complex curved surface shapes, so that the imaging quality of objects is seriously influenced.
In order to solve the problem, many experts and scholars propose different solutions, wherein one class is a processing method based on a single image, but the method has relatively harsh requirements on the color, the area of a high-gloss area and the saturation of the image, and is not suitable for materials with larger area and strong specular reflection; the other type is a highlight processing method based on a plurality of images, which is divided into a multi-angle light source method and a multi-angle camera view angle method, wherein in the multi-angle camera view angle method, for materials such as ceramics, metals, plastics and the like, the characteristic matching effect is extremely poor due to the condition of no texture or even weak texture, even the highlight areas in images with different view angles are used as textures to perform false matching, and a corresponding transformation matrix among the plurality of images cannot be found; in the multi-angle light source method, although the problems of matching and matrix transformation of the texture features of weak textures and no textures can be skipped, the algorithm of the next highlight region detection and image fusion steps is extremely complex, and the adaptability is poor.
Therefore, how to design a set of more general image highlight removing system can not only aim at the object made of common materials, but also adapt to the materials with high light reflection and weak textures such as ceramics, metals, plastics and the like, can effectively remove almost all highlight problems, and greatly relieve the problem of uneven light irradiation steps.
Disclosure of Invention
In order to solve the technical problems in the background art, the invention provides an efficient and general method for removing the specular reflection surface image.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
a method for removing high light of a strong reflection surface image is characterized in that: the method for removing the high light of the image of the strong reflection surface comprises the following steps:
1) Image acquisition is carried out on an object to be shot in a mode of fixing a camera view angle and a multi-angle light source, images of a plurality of light source angles under the same view angle are obtained, and the position of a central point of the object to be shot is determined; the center point position is a reference position;
2) Acquiring three images with different light source angles of an object to be shot according to the obtained reference position in the step 1);
3) Carrying out fusion processing on the images obtained in the step 2) by adopting a fusion algorithm to obtain fused images;
4) And (3) carrying out homomorphic filtering treatment on the fused image obtained in the step (3), adjusting the overall brightness of the fused image, and improving the contrast of the fused image.
Preferably, the specific implementation manner of the step 1) adopted by the invention is as follows:
1.1 A darkroom image acquisition device is built; the darkroom image acquisition device comprises a first light source, a second light source, a third light source and a shooting camera with a fixed visual angle; the first light source, the second light source and the third light source are annularly arranged and are respectively provided with an inclination angle with the horizontal direction; the heights of the first light source, the second light source and the third light source are the same; the first light source, the second light source and the third light source are circumferentially arranged around the shooting camera and are respectively separated by 120 degrees;
1.2 The first light source is turned on, the intensity and the pose of the first light source are adjusted, so that the imaging brightness of an object to be shot is moderate, the brightness distribution is uniform, a first picture is acquired through a shooting camera, and the first picture is a reference image;
1.3 Turning off the first light source, turning on the second light source, pre-adjusting the intensity and the pose of the second light source through the imaging effect, and acquiring a second picture through a shooting camera; the second picture is a parameter adjusting image;
1.4 Aiming at the reference image obtained in the step 1.2) and the parameter adjusting image obtained in the step 1.3), firstly carrying out graying treatment on the reference image and the parameter adjusting image, then dividing a foreground object of the picture after the graying treatment from a black background by using a threshold value, then respectively calculating three indexes of gray average value, gray median, gray value mode and gray histogram statistical duty ratio of the foreground object and the background object, and adjusting the intensity of a second light source by taking the first three indexes of the reference image as a benchmark so that the three indexes of the gray average value, the gray median and the gray value mode of the picture shot by the camera are as close as possible to the reference image; then fixing the intensity of the second light source, and adjusting the inclination angle of the second light source, the distance between the second light source and the object and the pose, so that the gray histogram statistics of the photographed parameter-adjusting image is as close as possible to the gray histogram statistics of the reference image obtained in the step 1.2);
1.5 Turning off the second light source, turning on the third light source, repeating step 1.4) until the adjustment of the intensity and the pose of all the light sources is completed, recording the position of the center point of the photographed object, making marks, wherein the position of the center point is the reference position for placing the photographed object at the back.
Preferably, the specific implementation manner of adjusting the intensity of the second light source in the step 1.4) adopted by the invention is as follows:
automatic calculation of gray mean g of foreground gray images using python function package mean Gray median g median Gray value mode g most ;
Assuming that the gray average value corresponding to the reference image is the gray median and the gray value mode
Calculating the gray average value corresponding to the tone parameter image, wherein the gray median value and the gray value mode are
Index M for adjusting intensity of second light source 1 The method comprises the following steps:
preferably, in the step 1.4) adopted in the invention, the specific implementation modes of adjusting the inclination angle of the second light source, the distance between the second light source and the object and the pose are as follows:
the gray values 0 to 255 are divided into 8 bins, each interval being 256/8=32, and its interval being:
[[0,31],[32,63],[64,95],[96,127],[128,159],[160,191],[192,223],[224,255]]
counting the frequency of the gray values of the foreground images according to the above formula to obtain:
[n 0 .n 1 ,n 2 ,n 3 ,n 4 ,n 5 ,n 6 ,n 7 ]where n is the frequency at which pixels appear in each interval;
the duty ratio p of each interval frequency is calculated as follows:
finally, the duty ratio condition based on each interval is shown as the following formula:
[p 0 ,p 1 ,p 2 ,p 3 ,p 4 ,p 5 .p 6 ,p 7 ]
the total pixel interval duty ratio is ordered from large to small as follows:
[P 0 ,P 1 ,P 2 ,P 3 ,P 4 ,P 5 ,P 6 ,P 7 ]
accumulating the ordered results until the accumulated index sum is greater than 0.8, and recording the index position before the ordering; assume that the front index section corresponding to the reference image has the following duty ratio:the front index interval duty ratio corresponding to the parameter adjusting image is calculated as follows: />Index M for adjusting inclination angle of second light source, distance between second light source and object and pose 2 The method comprises the following steps: />
Preferably, the specific implementation manner of the step 3) adopted by the invention is as follows:
when the images with different light source angles obtained in the step 3) are fused, the image fusion formula is as follows:
wherein,
I old_1 ,I old_2 ,I old_3 the gray values of the three acquired images at the corresponding positions of the r, g and b channels are respectively;
r, g, b are three channels of the image respectively;
the minimum gray values of the corresponding positions of the r, g and b channels are the new images after the fusion.
Preferably, the homomorphic filtering processing in step 4) employed in the present invention has a transfer function of
Wherein:
h is the high frequency gain;
l is the low frequency gain;
D 0 is the cut-off frequency;
k is a sharpening coefficient;
d (x, y) is the frequency (x, y) to the center frequency (x 0 ,y 0 ) The specific calculation formula is as follows:
the invention has the advantages that:
the invention provides a method for removing high light of an image on a strong reflection surface, which is characterized in that: firstly, the invention uses a new multi-angle light source intensity and pose adjusting method, effectively solves the problem that the images shot by the multi-angle light source are influenced by objective hardware conditions, and can not ensure that the brightness difference among the images shot by the light sources with different angles is basically unchanged. Firstly, a picture with good effect is taken as a reference image after a light source is turned on to be adjusted to a proper angle and brightness, then the light source is turned off, the light source with the other angle is turned on, the second image is taken as a reference image after the angle and brightness are adjusted, the gray distribution of the two images can not be ensured to be close, at the moment, gray distribution diagrams of foreground targets of the reference image and the reference image are respectively obtained, average value, median and mode values of the reference image and the reference image are obtained, finally, the magnitude of the light source intensity is manually adjusted according to the magnitude of square difference among the average value, median and mode values of the reference image and the reference image, the magnitude of the light source intensity is manually adjusted to be minimum as possible, meanwhile, the same adjustment operation is carried out on the third light source and the image according to the magnitude of square difference between the gray distribution diagrams of the reference image and the reference image, the defect that the brightness difference and the distribution difference of the images are too large is caused by hardware conditions is overcome, and the subsequent fusion step is caused to generate larger luminance difference and discontinuous condition. Secondly, the method and the device effectively solve the problem of low overall brightness and contrast caused by the fusion algorithm in the previous step by using homomorphic filtering to process the fused image. After the image fusion, homomorphic filtering algorithm is used for processing, and according to different pixel sizesChanging the cut-off frequency D in the transfer function 0 The method can be used for adaptively solving the problem of low brightness and contrast generated by a fusion algorithm.
Drawings
FIG. 1 is a schematic diagram of a darkroom image acquisition device used in the present invention;
FIG. 2 is a flow chart of the adjustment of the brightness and pose of a multi-angle light source;
FIG. 3 is a view of multiple light source angles acquired using an acquisition device (ceramic disk back, ceramic disk front, cell phone case back, glass beaker and toilet paper bag, respectively);
FIG. 4 is an image after algorithm fusion (ceramic tray back, ceramic tray front, cell phone case back, glass beaker and toilet paper bag, respectively);
FIG. 5 shows the results of homomorphic filter treatment (back of ceramic tray, front of ceramic tray, back of cell phone case, glass beaker and toilet paper bag, respectively).
Detailed Description
The following detailed description of the embodiments of the invention refers to the accompanying drawings, which illustrate the steps of:
step one: as shown in FIG. 1, the camera is a fixed viewing angle, and the 3 annular light sources are respectively separated by approximately 120 degrees, form a certain inclination angle with the horizontal direction, and are placed at the same height position. The experimental device collects images under three angle light sources respectively to perform fusion and highlight removal, but because of the inclination angle of the light sources, the brightness control and the difference between the photographed object and the positions and angles of the light sources, the brightness and the distribution difference of the photographed images are overlarge under the light sources with three different angles, so that larger brightness difference and discontinuous results are generated in the image fusion step, and the visual effect and the image quality are seriously affected.
Step two: as shown in FIG. 2, the invention provides a flow chart for simply adjusting the light source intensity and the pose to dynamically adjust the light source intensity and the pose of the same object, so that the brightness between images shot at different light source angles is moderate, the difference is small, and the distribution is similar. The method comprises the following specific steps:
(1) Firstly, one light source in the darkroom device is turned on, the intensity and the pose of the light source are adjusted simultaneously by observing the picture on the computer, so that the imaging brightness of a shot object is moderate, the brightness distribution is uniform, and a first picture is acquired as a reference image.
(2) The front light source is turned off, the second light source is turned on, the brightness and the pose of the light source are pre-adjusted through the imaging effect, and the second picture is collected as a parameter adjusting image after adjustment.
(3) Aiming at a reference image and a parameter adjusting image, firstly, carrying out graying treatment on the two images, then dividing a foreground object in the images from a black background by using a threshold value, then respectively calculating three indexes of gray average value, gray median and gray value mode and gray histogram statistical duty ratio of the foreground object in the images, and taking the first three indexes of the reference image as a benchmark, adjusting the intensity of a second light source to enable the intensity of the second light source to be close to the gray average value, the gray median and the gray value mode of the images as much as possible; then fixing the intensity of the light source, and adjusting the inclination angle of the light source and the conditions of the distance and the pose of the object, so that the gray histogram statistics of the picture shot by the light source is as close as possible to the first picture.
The method for the reference index of the foreground image comprises the following steps:
1) And adjusting the index of the light source intensity.
Automatically calculating gray average value of foreground gray image by using python function package, wherein gray median and gray value mode are g respectively mean 、g median 、g most . Assuming that the gray average value corresponding to the reference image is the gray median and the gray value modeCalculating the gray average value corresponding to the tone parameter image, wherein the gray median value and the gray value mode are
The index for adjusting the intensity of the light source at this time is:
2) And (3) adjusting the indexes such as the inclination angle of the light source, the distance between the light source and the object, the pose and the like.
The gray values 0 to 255 are divided into 8 bins, each interval being 256/8=32, and its interval being:
[[0,31],[32,63],[64,95],[96,127],[128,159],[160,191],[192,223],[224,255]]
the gray values of the foreground images are counted according to the gray distribution intervals, and the result is shown in the following graph:
[n 0 .n 1 ,n 2 ,n 3 ,n 4 ,n 5 ,n 6 ,n 7 ]
the duty ratio of each interval frequency is calculated as follows:
finally, the calculated duty ratio condition based on the frequency of each interval is shown as the following formula:
[p 0 ,p 1 ,p 2 ,p 3 ,p 4 ,p 5 .p 6 ,p 7 ]
the total pixel interval duty ratio is ordered from large to small as follows:
[P 0 ,P 1 ,P 2 ,P 3 ,P 4 ,P 5 ,P 6 ,P 7 ]
and accumulating the ordered results until the accumulated index sum is greater than 0.8, and recording the index before corresponding to the index position before ordering. Assume that the front index section corresponding to the first picture has the following duty ratio:the front index interval duty ratio corresponding to the second picture is calculated as follows: />The indexes of adjusting the inclination angle of the light source, the distance between the light source and an object, the pose and the like are as follows: />
Turning off the second light source, turning on the third light source, at this time, the first image is still a reference image, the acquired third image is a parameter-adjusting image, the above steps are repeated once, the adjustment of the intensity and the pose of the three light sources is completed, the brightness between pictures is ensured to be close, the brightness distribution is relatively similar and uniform, at this time, the center point position of a shot object is recorded, marks are made, and the function of a reference position for placing the object is played when the same kind of objects are shot.
Step three: as shown in fig. 3, the reference positions corresponding to different objects are found by using the image acquisition method and the image acquisition step in the last step, and then three images of light sources with different angles are acquired for each article, wherein the images comprise a ceramic disc, a toilet paper packaging bag, a glass beaker and the back of a mobile phone shell.
Step four: as shown in fig. 4, the fusion algorithm is used to fuse the images of the ceramic disc, the toilet paper packaging bag, the glass beaker and the back of the mobile phone shell, wherein the method for taking the minimum gray value at the corresponding position in the three images is used, and the formula is as follows:
wherein:
I old_1 ,I old_2 ,I old_3 gray values of the corresponding positions of the r, g and b channels are respectively acquired for the three acquired images;
r, g, b are three channels of the image respectively;
for the fused image at r respectivelyThe minimum of gray values at the corresponding positions of the g and b channels is the new image.
The acquired images can be fused according to the fusion algorithm formula, and the highlight region can be well removed by obtaining the fused images.
Step five: as shown in fig. 5, although the highlight region is well removed from the image obtained by the previous fusion processing, other negative effects are brought about, for example, the overall brightness of the image is low and the contrast is low due to the fusion algorithm, so that the image quality is reduced. Aiming at the problem, homomorphic filtering is used for processing, and the homomorphic filtering algorithm is an algorithm in a frequency domain, combines the advantages of a gray level change and linear proportional conversion method, can realize gray level adjustment of an image, and can enhance the contrast of the image by adopting a frequency filtering method. The brightness of the image after homomorphic filtering is adjusted to a proper range, and the contrast of the image is greatly improved, and the image quality is remarkably improved. In fig. 3, fig. 4 and fig. 5, in order to prove the versatility of the present invention, the image acquisition system is used to acquire the images of the back surface of the ceramic disc, the front surface of the ceramic disc, the back surface of the mobile phone case, the glass beaker and the toilet paper packaging bag, each material comprises 3 light sources with different angles, after the images are fused, each three images are fused into one image with the highlight area removed, and finally the final image is obtained through homomorphic filtering. The example adopts 4 materials and 5 groups of images, each group is three, and the images are shot by light sources with different angles, so that the effectiveness and the universality of the invention are proved, and the comparison test is carried out.
The homomorphic filtering comprises the following specific steps:
an image f (x, y) can be expressed as the product of its illumination component i (x, y) and reflection component r (x, y) according to an illumination-reflection model, which is:
f(x,y)=i(x,y)*r(x,y) (1)
wherein f (x, y) represents an original image; i (x, y) is the illumination component; r (x, y) represents the reflection component.
It cannot be used directly here to process the frequency components of the illumination because in the fourier transform mathematical expression the transformation of the product is not identical to the product of the transformation. Taking the logarithm of both sides of them, and changing them into a sum form, it is possible to obtain:
f(x,y)=In[i(x,y)]+In[r(x,y)] (2)
the resulting sum form is then fourier transformed and converted into the frequency domain:
F(x,y)=I(x,y)+R(x,y) (3)
f (x, y) is processed with a frequency domain enhancement transfer function H (x, y) to obtain:
F(x,y)*H(x,y)=I(x,y)*H(x,y)+R(x,y)*H(x,y) (4)
where H (x, y) is the homomorphic filter transfer function, where the high frequency enhancement filter is very close to a Gaussian high pass filter, the formula is:
wherein H is high frequency gain, L is low frequency gain, D 0 For the cut-off frequency, k is the sharpening factor, in terms of points (x 0 ,y 0 ) The frequency domain distance calculation formula for the filter center is as follows:
D(x,y)=((x-x 0 ) 2 +(y-y 0 ) 2 ) 1/2 (6)
transforming the frequency domain enhanced image into a space domain to obtain:
f * (x,y)=i * (x,y)*r * (x,y) (7)
the indexes are taken from the two sides to be reduced to obtain:
exp[f * (x,y)]=exp[i * (x,y)]*exp[r * (x,y)] (8)
let f ** (x,y)=exp[f * (x,y)],i ** (x,y)=exp[i * (x,y)]
r ** (x,y)=exp[r * (x,y)]It is possible to obtain:
f ** (x,y)=i ** (x,y)*r ** (x,y) (9)
in homomorphic filter coefficients, h=2.5, l=0.5 and k=1 are selected, and after the original image f (x, y) is homomorphic filtered in the above steps, a high-quality enhanced image f with balanced brightness and higher contrast is obtained ** (x,y)。
The cut-off frequency cannot be set to a fixed value, and because the same cut-off frequency is fixed in the homomorphic filtering process of the fused images with different materials and pixel sizes, the processed effect is uneven, so that a more general cut-off frequency needs to be found so as to process the fused images with different materials and pixel sizes.
Because the ceramic disk is an almost texture-free image and the similarity between the images is large, the fixed cut-off frequency is D when the homomorphic filtering processing of the ceramic disk image is performed 0 For other 3 different materials and shapes, the shape and texture are greatly different, and the cutoff frequency D of different materials and shapes 0 A fixed value cannot be selected, so the method used in literature for averaging D (u, v), i.e
Wherein M and N are the width and height of the image, and D (u, v) is the distance from the center of the frequency, so that the self-adaptive cut-off frequency can be obtained according to different shapes of objects.
And finally, homomorphic filtering treatment is carried out on the ceramic disc, the toilet paper packaging bag, the glass beaker and the back of the mobile phone shell according to the given cut-off frequency respectively by the homomorphic filtering method, and the obtained treated image is shown in figure 5.
Claims (5)
1. A method for removing high light of a strong reflection surface image is characterized in that: the method for removing the high light of the image of the strong reflection surface comprises the following steps:
1) Image acquisition is carried out on an object to be shot in a mode of fixing a camera view angle and a multi-angle light source, images of a plurality of light source angles under the same view angle are obtained, and the position of a central point of the object to be shot is determined; the center point position is a reference position;
2) Acquiring three images with different light source angles of an object to be shot according to the obtained reference position in the step 1);
3) Carrying out fusion processing on the images obtained in the step 2) by adopting a fusion algorithm to obtain fused images;
4) Homomorphic filtering processing is carried out on the fused image obtained in the step 3), the overall brightness of the fused image is adjusted, and the contrast ratio of the fused image is improved;
the specific implementation mode of the step 1) is as follows:
1.1 A darkroom image acquisition device is built; the darkroom image acquisition device comprises a first light source, a second light source, a third light source and a shooting camera with a fixed visual angle; the first light source, the second light source and the third light source are annularly arranged and are respectively provided with an inclination angle with the horizontal direction; the heights of the first light source, the second light source and the third light source are the same; the first light source, the second light source and the third light source are circumferentially arranged around the shooting camera and are respectively separated by 120 degrees;
1.2 The first light source is turned on, the intensity and the pose of the first light source are adjusted, so that the imaging brightness of an object to be shot is moderate, the brightness distribution is uniform, a first picture is acquired through a shooting camera, and the first picture is a reference image;
1.3 Turning off the first light source, turning on the second light source, pre-adjusting the intensity and the pose of the second light source through the imaging effect, and acquiring a second picture through a shooting camera; the second picture is a parameter adjusting image;
1.4 Aiming at the reference image obtained in the step 1.2) and the parameter adjusting image obtained in the step 1.3), firstly carrying out graying treatment on the reference image and the parameter adjusting image, then dividing a foreground object of the picture subjected to the graying treatment from a black background by using a threshold value, then respectively calculating three indexes of gray average value, gray median, gray value mode and gray histogram statistical duty ratio of the foreground object and the background object, and adjusting the intensity of a second light source by taking the first three indexes of the reference image as a benchmark so that the three indexes of the gray average value, the gray median and the gray value mode of the picture shot by the camera are as close as possible to the reference image; then fixing the intensity of the second light source, and adjusting the inclination angle of the second light source, the distance between the second light source and the object and the pose, so that the gray histogram statistics of the photographed parameter-adjusting image is as close as possible to the gray histogram statistics of the reference image obtained in the step 1.2);
1.5 Turning off the second light source, turning on the third light source, repeating step 1.4) until the adjustment of the intensity and the pose of all the light sources is completed, recording the position of the center point of the photographed object, making marks, wherein the position of the center point is the reference position for placing the photographed object at the back.
2. The method for highlighting the image of a strongly reflecting surface as recited in claim 1, wherein: the specific implementation manner of adjusting the intensity of the second light source in the step 1.4) is as follows:
automatic calculation of gray mean g of foreground gray images using python function package mean Gray median g median Gray value mode g most ;
Assuming that the gray average value corresponding to the reference image is the gray median and the gray value mode
Calculating the gray average value corresponding to the tone parameter image, wherein the gray median value and the gray value mode are
Index M for adjusting intensity of second light source 1 The method comprises the following steps:
3. the method for highlighting the image of a strongly reflecting surface as recited in claim 2, wherein: the specific implementation modes of adjusting the inclination angle of the second light source, the distance between the second light source and the object and the pose in the step 1.4) are as follows:
the gray values 0 to 255 are divided into 8 bins, each interval being 256/8=32, and its interval being:
[[0,31],[32,63],[64,95],[96,127],[128,159],[160,191],[192,223],[224,255]]
counting the frequency of the gray values of the foreground images according to the above formula to obtain:
[n 0 .n 1 ,n 2 ,n 3 ,n 4 ,n 5 ,n 6 ,n 7 ]where n is the frequency at which pixels appear in each interval;
the duty ratio p of each interval frequency is calculated as follows:
finally, the duty ratio condition based on each interval is shown as the following formula:
[p 0 ,p 1 ,p 2 ,p 3 ,p 4 ,p 5 .p 6 ,p 7 ]
the total pixel interval duty ratio is ordered from large to small as follows:
[P 0 ,P 1 ,P 2 ,P 3 ,P 4 ,P 5 ,P 6 ,P 7 ]
accumulating the ordered results until the accumulated index sum is greater than 0.8, and recording the index position before the ordering; assume that the front index section corresponding to the reference image has the following duty ratio:the front index interval duty ratio corresponding to the parameter adjusting image is calculated as follows: />Index M for adjusting inclination angle of second light source, distance between second light source and object and pose 2 The method comprises the following steps: />
4. A method of specular image specular removal according to claim 3, wherein: the specific implementation manner of the step 3) is as follows:
when the images with different light source angles obtained in the step 2) are fused, an image fusion formula is as follows:
wherein,
I old_1 ,I old_2 ,I old_3 the gray values of the three acquired images at the corresponding positions of the r, g and b channels are respectively;
r, g, b are three channels of the image respectively;
the minimum gray values of the corresponding positions of the r, g and b channels are the new images after the fusion.
5. The method for highlighting the image of a strongly reflecting surface as recited in claim 4, wherein: the homomorphic filtering processing in the step 4) has the transfer function of
Wherein:
h is the high frequency gain;
l is the low frequency gain;
D 0 is the cut-off frequency;
k is a sharpening coefficient;
d (x, y) is the frequency (x, y) to the center frequency (x 0 ,y 0 ) The specific calculation formula is as follows:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110214047.8A CN112884689B (en) | 2021-02-25 | 2021-02-25 | Method for removing high light of strong reflection surface image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110214047.8A CN112884689B (en) | 2021-02-25 | 2021-02-25 | Method for removing high light of strong reflection surface image |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112884689A CN112884689A (en) | 2021-06-01 |
CN112884689B true CN112884689B (en) | 2023-11-17 |
Family
ID=76054546
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110214047.8A Active CN112884689B (en) | 2021-02-25 | 2021-02-25 | Method for removing high light of strong reflection surface image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112884689B (en) |
Citations (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR19980014140A (en) * | 1996-08-08 | 1998-05-15 | 김태은 | Method of restoring shape by estimating surface reflection property of object |
WO1999038121A1 (en) * | 1998-01-27 | 1999-07-29 | Sensar, Inc. | Method and apparatus for removal of bright or dark spots by the fusion of multiple images |
CN101190121A (en) * | 2006-11-29 | 2008-06-04 | 佳能株式会社 | Focusing unit and ophthalmic photographing apparatus |
JP2008298703A (en) * | 2007-06-04 | 2008-12-11 | Toyohashi Univ Of Technology | Device for automatic inspection of surface defect and inspection method |
JP2013101109A (en) * | 2011-10-12 | 2013-05-23 | Shiseido Co Ltd | Lighting system and image acquisition device |
JP2014106328A (en) * | 2012-11-27 | 2014-06-09 | Fujitsu Telecom Networks Ltd | Image analysis device and optical fiber fusion connection system |
WO2015167009A1 (en) * | 2014-05-02 | 2015-11-05 | オリンパス株式会社 | Image processing device and image pickup device |
CN107194881A (en) * | 2017-03-23 | 2017-09-22 | 南京汇川图像视觉技术有限公司 | A kind of removal image reflex reflector and method based on photometric stereo |
CN107424118A (en) * | 2017-03-28 | 2017-12-01 | 天津大学 | Based on the spherical panorama mosaic method for improving Lens Distortion Correction |
CN108154491A (en) * | 2018-01-26 | 2018-06-12 | 上海觉感视觉科技有限公司 | A kind of reflective removing method of image |
CN108445007A (en) * | 2018-01-09 | 2018-08-24 | 深圳市华汉伟业科技有限公司 | A kind of detection method and its detection device based on image co-registration |
CN110648301A (en) * | 2019-09-30 | 2020-01-03 | 四川大学 | Device and method for eliminating imaging reflection |
CN110807751A (en) * | 2019-10-31 | 2020-02-18 | 栗智 | Method for eliminating illumination influence in industrial environment imaging |
CN111292256A (en) * | 2020-01-14 | 2020-06-16 | 哈尔滨工业大学 | Texture enhancement algorithm based on microscopic hyperspectral imaging |
CN111583123A (en) * | 2019-02-17 | 2020-08-25 | 郑州大学 | Wavelet transform-based image enhancement algorithm for fusing high-frequency and low-frequency information |
CN111741214A (en) * | 2020-05-13 | 2020-10-02 | 北京迈格威科技有限公司 | Image processing method and device and electronic equipment |
-
2021
- 2021-02-25 CN CN202110214047.8A patent/CN112884689B/en active Active
Patent Citations (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR19980014140A (en) * | 1996-08-08 | 1998-05-15 | 김태은 | Method of restoring shape by estimating surface reflection property of object |
WO1999038121A1 (en) * | 1998-01-27 | 1999-07-29 | Sensar, Inc. | Method and apparatus for removal of bright or dark spots by the fusion of multiple images |
CN101190121A (en) * | 2006-11-29 | 2008-06-04 | 佳能株式会社 | Focusing unit and ophthalmic photographing apparatus |
JP2008298703A (en) * | 2007-06-04 | 2008-12-11 | Toyohashi Univ Of Technology | Device for automatic inspection of surface defect and inspection method |
JP2013101109A (en) * | 2011-10-12 | 2013-05-23 | Shiseido Co Ltd | Lighting system and image acquisition device |
JP2014106328A (en) * | 2012-11-27 | 2014-06-09 | Fujitsu Telecom Networks Ltd | Image analysis device and optical fiber fusion connection system |
WO2015167009A1 (en) * | 2014-05-02 | 2015-11-05 | オリンパス株式会社 | Image processing device and image pickup device |
CN107194881A (en) * | 2017-03-23 | 2017-09-22 | 南京汇川图像视觉技术有限公司 | A kind of removal image reflex reflector and method based on photometric stereo |
CN107424118A (en) * | 2017-03-28 | 2017-12-01 | 天津大学 | Based on the spherical panorama mosaic method for improving Lens Distortion Correction |
CN108445007A (en) * | 2018-01-09 | 2018-08-24 | 深圳市华汉伟业科技有限公司 | A kind of detection method and its detection device based on image co-registration |
CN108154491A (en) * | 2018-01-26 | 2018-06-12 | 上海觉感视觉科技有限公司 | A kind of reflective removing method of image |
CN111583123A (en) * | 2019-02-17 | 2020-08-25 | 郑州大学 | Wavelet transform-based image enhancement algorithm for fusing high-frequency and low-frequency information |
CN110648301A (en) * | 2019-09-30 | 2020-01-03 | 四川大学 | Device and method for eliminating imaging reflection |
CN110807751A (en) * | 2019-10-31 | 2020-02-18 | 栗智 | Method for eliminating illumination influence in industrial environment imaging |
CN111292256A (en) * | 2020-01-14 | 2020-06-16 | 哈尔滨工业大学 | Texture enhancement algorithm based on microscopic hyperspectral imaging |
CN111741214A (en) * | 2020-05-13 | 2020-10-02 | 北京迈格威科技有限公司 | Image processing method and device and electronic equipment |
Non-Patent Citations (3)
Title |
---|
"Application of Generating Adversarial Networks in High-light Removal of Wheel Hub Surface";Zhuangzhuang Shen等;《IEEE》;全文 * |
"基于多特征融合的水下图像清晰化";田鑫;《中国优秀硕士学位论文全文数据库 信息科技辑》;全文 * |
"多尺度特征融合的陶瓷盘缺陷检测算法的研究";林刚等;《陶瓷学报》;全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN112884689A (en) | 2021-06-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11257194B2 (en) | Method for image dehazing based on adaptively improved linear global atmospheric light of dark channel | |
CN110175964B (en) | Retinex image enhancement method based on Laplacian pyramid | |
WO2013099772A1 (en) | Cell contour forming device and method, storage medium storing computer-processable cell contour forming program | |
US9916644B1 (en) | Ghost artifact removal system and method | |
CN108288258B (en) | Low-quality image enhancement method under severe weather condition | |
CN110163807B (en) | Low-illumination image enhancement method based on expected bright channel | |
CN112837233B (en) | Polarization image defogging method for acquiring transmissivity based on differential polarization | |
CN107403435B (en) | Color image edge extraction method based on RGB color space | |
CN109274948B (en) | Image color correction method, device, storage medium and computer equipment | |
CN109639994B (en) | Dynamic adjusting method for exposure time of embedded vehicle-mounted camera | |
CN115937216B (en) | Magnetic rotor appearance quality detection method for new energy automobile | |
CN109738069B (en) | Method for correcting multispectral imaging illumination space nonuniformity | |
CN109584181A (en) | It is a kind of improved based on Retinex infrared image detail enhancing method | |
CN112884689B (en) | Method for removing high light of strong reflection surface image | |
CN117058161B (en) | Electronic cigarette atomized smoke image segmentation method based on image processing | |
Han et al. | Automatic illumination and color compensation using mean shift and sigma filter | |
US8693798B2 (en) | Image sharpness processing apparatus and image sharpness processing method | |
CN113947535A (en) | Low-illumination image enhancement method based on illumination component optimization | |
CN106355556B (en) | The method and apparatus of image defogging | |
CN111127350A (en) | Image enhancement method | |
CN112950484B (en) | Method for removing color pollution of photographic image | |
Terai et al. | Color image contrast enhancement by retinex model | |
CN113129270B (en) | Method for determining weld line | |
CN109345501A (en) | Spiral winded type material defect inspection method based on frequency domain image enhancement | |
CN114972102A (en) | Underwater image enhancement method based on global variable contrast enhancement and local correction |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |