CN107194881B - A kind of reflective method of removal image based on photometric stereo - Google Patents
A kind of reflective method of removal image based on photometric stereo Download PDFInfo
- Publication number
- CN107194881B CN107194881B CN201710177279.4A CN201710177279A CN107194881B CN 107194881 B CN107194881 B CN 107194881B CN 201710177279 A CN201710177279 A CN 201710177279A CN 107194881 B CN107194881 B CN 107194881B
- Authority
- CN
- China
- Prior art keywords
- reflective
- image
- photometric stereo
- light source
- gray value
- 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 45
- 239000000203 mixture Substances 0.000 claims abstract description 10
- 230000001678 irradiating effect Effects 0.000 claims abstract description 8
- 238000001914 filtration Methods 0.000 claims abstract description 7
- 238000007781 pre-processing Methods 0.000 claims abstract description 5
- 238000005286 illumination Methods 0.000 claims description 20
- 241001566735 Archon Species 0.000 claims description 16
- 230000003287 optical effect Effects 0.000 claims description 16
- 230000000903 blocking effect Effects 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000003384 imaging method Methods 0.000 claims description 2
- 238000009434 installation Methods 0.000 claims description 2
- 230000003760 hair shine Effects 0.000 claims 2
- 230000000149 penetrating effect Effects 0.000 claims 1
- 238000012545 processing Methods 0.000 abstract description 4
- 230000000694 effects Effects 0.000 description 11
- 230000011514 reflex Effects 0.000 description 5
- 238000001514 detection method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000005684 electric field Effects 0.000 description 3
- 239000004615 ingredient Substances 0.000 description 3
- 238000013461 design Methods 0.000 description 2
- 240000007594 Oryza sativa Species 0.000 description 1
- 235000007164 Oryza sativa Nutrition 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002146 bilateral effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 210000001747 pupil Anatomy 0.000 description 1
- 238000012372 quality testing Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 235000009566 rice Nutrition 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 239000000126 substance Substances 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
Abstract
The reflective method of removal image based on photometric stereo that the invention discloses a kind of, belongs to machine vision and technical field of video image processing.The present invention installs four consistent light sources of light intensity in place first, and the irradiating angle of each light source is determined by Source calibration；Four light sources respectively shoot a width picture under conditions of lighting respectively, remove white noise to resulting four width image preprocessing with gaussian filtering, the gray scale adjusting parameter of each pixel is then sought, according to parameter to there may be the gray values of reflective point to be adjusted；Then treated four width images by photometric stereo algorithm composition overdetermined equation are sought into albedo figure, removes reflective image to obtain.It is reflective to remove that one aspect of the present invention with four width images does photometric stereo, has very strong robustness to the shade to be formed and complex pattern is blocked；On the other hand realizing can obtain image and light source uncoupling the reflective image of more accurate removal.
Description
Technical field
The present invention relates to machine vision and technical field of video image processing, are based on luminosity more specifically to one kind
Three-dimensional removal image reflex reflector and method.
Background technique
Machine vision is in the quality testing fields of industrial products using more prevalent, however many product surfaces are easily reflective
The some regions of image are covered by bloom after causing camera to adopt figure, therefore surface information such as texture, character or labeling etc. are also lost
It loses, the effect of image detection algorithm is allowed to have a greatly reduced quality.
Industrially removing reflective method is broadly divided into two classes at present: more image inputs and single image input.Based on more figures
As the complementarity of highlight area is reflective to remove image, such as image mosaic method, the average image method, the above method or to highlight area
Judgement lack enough accuracy, or processing it is excessively coarse, effect image is inaccurate.Based on gradient field solve Poisson's equation again by
Gradient image reconstructs method [Rogerio Feris, ramesh Raskar, the Kar-Han Tan, Matthew of image
Turk.Specular Reflection Reduction with Multi-Flash Imageing.] it is reflective to remove image,
Precision with higher and treatment effect, but this method long operational time, are not able to satisfy the requirement of industrial real-time, and for height
The overlapping input picture in light region lacks enough robustness.
Method removal image based on single image filtering is reflective, such as quick bilateral filtering, bandpass filtering, due to depending on unduly
Meet certain in image itself and assume that pattern such as is in certain regularity, the application surface of algorithm is more narrow.Based on block layer decomposition
Method [Yu Li, Michael S.Brown.Single Image Layer Separation using Relative
Smoothness] removal image is reflective, it is assumed that and short-tail distribution is presented in the histogram of gradients of illumination pattern, and long-tail point is presented in reflectogram
Cloth, but when two figure layer gray scale smoothnesses are not much different, this method will be helpless.In short, single image input is substantially
Underdetermined equation is solved, therefore robustness does not all reach requirement, existing single image input method, both real-time and accuracy are difficult
To meet the requirements simultaneously.
It is reflective about removal image through retrieving, also with the presence of many publications.Such as number of patent application
201210378478.9, the applying date is on October 8th, 2012, invention and created name are as follows: eliminates video image and comes from smooth object
The lateral reflective system and method in surface；The system of this application includes video camera, polagizing filter；Video camera is to obtain view
Frequency image；Before polagizing filter is set to camera lens, to eliminate the reflection that smooth object surface generates in video image
Light.The polagizing filter includes electric field supply unit, is polarized unit；The electric field supply unit is polarized unit to supply
Required electric field, the unit that is polarized is to be changed into linearly polarized light for elliptically polarized light.This application passes through in video camera
Polagizing filter is set before camera lens, the reflection light that smooth object surface generates in video image can be eliminated.But this application is only
It can eliminate that surface is laterally reflective, simultaneously because the use of polarizing film, the diffusing reflection ingredient of object itself is also weakened, and accuracy is not
Foot.
For another example number of patent application 201610338320.7, the applying date are on May 19th, 2016, invention and created name are as follows: one
Kind removes reflective filming apparatus and method；This application, which first passes through, identifies that content is identical in photo but the light intensity depth is different
Mutual dislocation the retroreflective regions that surround of two pictures, then the mistake for being influenced the photo by retroreflective regions by repairing nomography
Image is repaired, and removes retroreflective regions, but this application repairs retroreflective regions using the depth dislocation of image, be only applicable in and
It removes transparent substance between target and camera and generates reflective, versatility deficiency.
There are also the patent No. 201510564190.4, the applying date is on September 6th, 2015, invention and created name are as follows: recognition of face
Glasses removing method；This application is filtered out anti-by a reasonable reflective serious picture of threshold value automatic fitration eyeglass
Luminosity is lower, does not influence the picture further progress frame elimination of Pupil diameter.But this application first carries out multiple images
Then Face datection finds ocular by artificial preset coordinate, then with weighted average difference method to eye areas into
Row interpolation reparation, not only process is complicated, and time-consuming for algorithm, and has manually preset retroreflective regions, is not suitable for retroreflective regions not
Fixed scene, versatility very little.
Summary of the invention
1. technical problems to be solved by the inivention
The problem to be solved in the present invention is: machine vision increasingly increases in the application demand that industrial product quality detects, and
The reflective effect that can allow image detection algorithm is had a greatly reduced quality；The existing removal for carrying out high-accuracy by various complicated algorithms is reflective
Method operand it is larger, be unsatisfactory for industrial requirement of real-time；And existing it can quickly remove reflective method and be only capable of
Specific product is coped with, poor to the product removal reflecting effect for there are other complex patterns, i.e. existing method is difficult in summary
Accomplish the compatible problem of high real-time and high accuracy, provides a kind of removal image reflex reflector based on photometric stereo and side
Method；The present invention irradiates four width images of lower shooting using four light sources respectively, solves after gray scale pretreatment with photometric stereo
Albedo figure, the gray value of respective pixel is replaced with albedo, to remove the reflective of image, not only high-efficient, robustness is good,
And the position of obtained image and light source is fully decoupled, and the accuracy of image is very high after processing.
2. technical solution
In order to achieve the above objectives, technical solution provided by the invention are as follows:
A kind of reflective method of removal image based on photometric stereo of the invention, to same under successively being irradiated using four light sources
One target shoots four width images, solves to obtain albedo figure with photometric stereo after pretreatment, replaces respective pixel with albedo
Gray value, to remove the reflective of image.
More specifically, the present invention installs four consistent light of light intensity using camera shooting direction as optical axis, in the periphery of optical axis
Source, four light sources determine plane perpendicular to optical axis, meanwhile, light source direction of illumination is converted into a bit, and the point is located on optical axis,
The irradiating angle of each light source is determined by Source calibration；It is each to same target under conditions of four light sources are successively lighted respectively
A width picture is shot, white noise is removed to resulting four width image preprocessing with gaussian filtering, then seeks each pixel
Gray scale adjusting parameter, according to parameter to there may be the gray values of reflective point to be adjusted；It then will treated four width figures
As seeking albedo figure by photometric stereo algorithm composition overdetermined equation, it can be obtained and remove reflective image.
Further, detailed process is as follows for light source radiating angle calibration:
1) Archon is placed at the position of testee, adjusting light-source brightness makes Archon not overexposure, shoots a width figure
Picture；
2) the position O (x of the centre of sphere is found by the method for Hough circle fittingo, yo) and radius of a ball ro；
3) Archon is partitioned into according to the centre of sphere and the radius of a ball, the maximum point L (x of gray value is found in Archonl, yl)；
4) the horizontal angle of direction of illumination and camera imaging is denoted as slant, direction of illumination and camera shoot axis
Angle is denoted as tilt, then:
Slant=arctan ((yl-yo)/(xl-xo))
Tilt=2*arcsin (sqrt ((yl-yo)^2+(xl-xo)^2))
5) above-mentioned steps are carried out to each light source, seeks the tilt and slant of each light source.
Further, in the step 3) of light source radiating angle calibration, the maximum region of gray value is found in Archon, after
The continuous center of circle that its region is found with Hough circle fitting, treats as the maximum point of gray value for this center of circle.
Further, the process of each gray scale adjusting parameter is sought are as follows:
1) to any pixel (x, y), the gray value on kth width image is Ik, k=1,2,3,4, by each image gray scale
Value normalization:
2) by i1、i2、i3、i4It is ranked up, subtracts time maximum and time minimum number mean value with maximum number, it is assumed that i1> i2
> i3> i4, definition:
Y1=i1-0.5*(i2+i3)
Y2=0.5* (i2+i3)-i4
Adjusting parameter k1、k2Is defined as:
k1=sqrt (2*Y1-Y1*Y1)
k2=sqrt (2*Y2-Y2*Y2)。
Further, as follows to the process being adjusted there may be the gray value of reflective point according to adjusting parameter:
1) directionality, i are had the characteristics that according to mirror-reflection1、i2、i3、i4In the corresponding gray value of maximum number most have can
It can have occurred reflective, the corresponding gray value of the smallest number is possible to be blocked, other corresponding gray value mirror surfaces of two parameters are anti-
It is low to penetrate ingredient, it is approximately diffusing reflection that the probability that is blocked is small, therefore does not need to adjust, Ik′For adjustment after gray value, k=1,2,3,
4, it may be assumed that
I2′=I2
I3′=I3
2) for I4, gray scale adjusted is sought according to adjusting parameter are as follows:
I4′=(1-k1) * I4+0.5*k1*(I2+I3)
3) for I1, shade may be formd due to blocking, determination strategy takes threshold decision, gives when it is less than
When determining threshold value th, it need to be adjusted, gray scale after adjustment are as follows:
I1’=(1-k2) * I1+0.5*k2*(I2+I3)。
Further, treated four width images are sought into albedo figure by photometric stereo algorithm composition overdetermined equation
Detailed process are as follows:
1) mikey vector is converted by each light source radiating angle, conversion formula is as follows:
X=cos (slant) * sin (tilt)
Y=sin (slant) * sin (tilt)
Z=cos (tilt)
2) by the direction matrix M of the irradiation vector composition 4*3 of four light sources4*3, four width image groups of gray scale adjustment are pairs of
The gray vector I answered4*1；
3) gray vector and irradiation vector are formed into over-determined systems by photometric stereo algorithm:
I4*1=β * M4*3*N3*1
Wherein, β is albedo, N3*1For the surface unit normal vector of the point；
4) above formula the right and left is multiplied by M4*3Transposition, it may be assumed that
Due to N3*1For normalized vector, so β are as follows:
The β for finding out each point obtains albedo figure, as removes reflective image.
A kind of removal image reflex reflector based on photometric stereo of the invention, including a camera and four light sources, with
Camera shooting direction is optical axis, and four source symmetrics are distributed in the periphery of optical axis, the plane that four light sources determine perpendicular to optical axis,
Meanwhile four light source direction of illuminations are converted into a bit, and the point is located on optical axis.
3. beneficial effect
Using technical solution provided by the invention, compared with existing well-known technique, there is following remarkable result:
A kind of reflective method of removal image based on photometric stereo of the invention, directly to the gray scale of each location of pixels into
Row linear operation finds out the albedo parameter of characterization each point reflection characteristic, has very strong robustness while the algorithm time is short.
Main advantage is: 1) being put forward for the first time that seek albedo figure using photometric stereo algorithm reflective to remove image, with four light sources
The lower four width images shot of irradiation respectively, carry out photometric stereo operation after gray value adjusts, can effectively exclude due to
The influence of shade caused by blocking and various complex patterns further enhances the robustness of algorithm；2) pass through four light source elder generations
Four width images of the same target of the lower shooting of irradiation seek the design of object albedo image afterwards, realize and decouple image and light source
It closes, what image reflected is the reflection characteristic of each point of body surface, can obtain the reflective image of more accurate removal.
Detailed description of the invention
Fig. 1 is the flow diagram that removal image is reflective in the present invention；
Fig. 2 is light source scheme of installation in the present invention；
(a)~(c) in Fig. 3 is that light source direction of illumination demarcates schematic diagram in the present invention；
(a)~(d) in Fig. 4 is the image that four cameras are respectively shot in the present invention；
Fig. 5 is that Fig. 4 shooting image removes reflective effect picture；
(a)~(d) in Fig. 6 is the image that another product is respectively shot by four cameras in the present invention；
Fig. 7 is that Fig. 6 shooting image removes reflective effect picture.
Label declaration in schematic diagram:
1, light source；2, camera.
Specific embodiment
To further appreciate that the contents of the present invention, the present invention is described in detail in conjunction with the accompanying drawings and embodiments.
Referring to Fig. 1, a kind of reflective method of removal image based on photometric stereo of the invention is successively shone using four light sources
Penetrate it is lower to same target shoot four width images, solve to obtain albedo figure with photometric stereo after pretreatment, replaced with albedo
The gray value of respective pixel, so that the reflective of image is removed, specifically: it is consistent that four light intensity are installed in place first
Light source determines the irradiating angle of each light source by Source calibration；Four light sources respectively shoot a width picture under conditions of lighting respectively,
White noise is removed to resulting four width image preprocessing with gaussian filtering, then seeks the gray scale adjusting parameter of each pixel,
According to parameter to there may be the gray values of reflective point to be adjusted；Then will treated four width images by photometric stereo calculation
Method composition overdetermined equation seeks albedo figure, can be obtained and removes reflective image.The present invention can quickly and accurately remove figure
It mainly include that the calibration of light source direction of illumination, the adjustment of four gray value of images and photometric stereo solve albedo figure three parts as reflective,
The present invention is specifically introduced below in conjunction with embodiment.
Embodiment 1
Referring to Fig. 2, according to photometric stereo principle, it is ensured that the image grayscale of each camera shooting only irradiating angle this
Variable is different, and remaining variables include that the intensity of light source, light source distance etc. need to be consistent, while avoid the same point in multiple image
Upper appearance is reflective.The design scheme of the removal image reflex reflector of the present embodiment are as follows: using 2 shooting direction of camera as optical axis, four
Light source 1 is symmetrically distributed in the periphery of optical axis, the plane that four light sources 1 determine perpendicular to optical axis, meanwhile, 1 direction of illumination meeting of light source
Gather for a bit, and the point is located on optical axis.
The present embodiment installs the consistent light source 1 of four light intensity in place first, is determined by Source calibration each
The irradiating angle of light source 1；Four light sources 1 respectively shoot a width picture to same target under conditions of successively lighting, and filter with Gauss
Wave removes white noise to resulting four width image preprocessing, the gray scale adjusting parameter of each pixel is then sought, according to parameter pair
There may be the gray values of reflective point to be adjusted；Then by treated, four width images press photometric stereo algorithm composition overdetermination
Equation seeks albedo figure, that is, obtains and remove reflective image.It is specific as follows:
1, light source direction of illumination is demarcated:
Light measurement and image optics show that image grayscale and direction of illumination are related with the angle of body surface normal vector, because
This must be known by light source direction of illumination.Four light sources are installed firstly the need of in suitable position, are located to four source symmetrics shooting axis
Periphery, be adjusted to suitable irradiating angle (such as Fig. 2), no longer it changed in actually detected.Adjust the light of four light sources
It is consistent according to intensity, an Archon is placed in the position of examining object, adjusting light-source brightness makes Archon not overexposure, lights one every time
Light source simultaneously shoots piece image, and each image determines corresponding light source direction of illumination.Consider light source, Archon, camera this system,
The most bright part of image, Archon reflected light are just incident on camera, according to the relationship of incidence reflection, the line of most bright spot to the centre of sphere
Angle with shooting axis is the half of direction of illumination.Center of circle O (the x of Archon is found with Hough circle fitting firsto, yo) and radius ro,
It is partitioned into the Archon region (shown in (b) in such as Fig. 3), maximum gray value is then found in the area, if maximum value has
It is multiple, i.e., the maximum region of gray value is found in Archon, subtracts 2 as the direct binarization segmentation of threshold value using the maximum gradation value
Brightest area out finds the center of circle with Hough circle fitting, using this center of circle as most bright spot L (xl, yl).The horizontal line shot with camera
X axis is done, shooting axis does z-axis, finds out the spherical coordinates of light source direction of illumination unit vector.Direction of illumination and the angle tilt of z-axis can
To be acquired with the most plan range of bright spot L and the centre of sphere with the radius of a ball, angle slant most bright spot L and the ball of direction of illumination and x-axis
It acquires (shown in (c) in such as Fig. 3) relative position of the heart.Accordingly, the specific steps of light source direction calibration are as follows:
Center of circle O (the x of Archon is found in Hough circle fittingo, yo) and radius ro, found with Hough circle fitting after maximizing
It is most bright
Point L (xl, yl), seek irradiating angle tilt and slant:
Slant=arctan ((yl-yo)/(Xl-Xo))
Tilt=2*arcsin (sqrt ((yl-yo)^2+(Xl-Xo)^2))
Above-mentioned steps are carried out to each light source, seek the tilt and slant of each light source.
2, four gray value of images adjust:
For there are reflective location of pixels, generally occurring in excessively bright and other dark images feelings on certain width image
Condition, source symmetric distribution can occur on multiple image reflective to avoid the same point.For most bright pixel value, with secondary most bright
It is adjusted with secondary two most dark pixels；It, also only can be on certain piece image for being blocked and being formed the point of shade
Occur, corresponding gray value can be close to 0, and a threshold parameter is arranged in the present embodiment, which is set as detection zone without screening
The lower lowest gray value of gear, is considered to be blocked lower than the threshold value, and the pixel value that is blocked is with time most bright and most dark two secondary
Pixel is adjusted.The gap of most bright spot or shadow spots and intermediate two values is bigger, illustrates that non-irreflexive ingredient is more, abnormal
The reliability of pixel itself reduces, and this gap is characterized using adjusting parameter, bigger pixel of adjusting parameter weight itself is lower.
The calculating of adjusting parameter is the opposite drift value for calculating most bright spot first, then opposite drift value is mapped to a convex curves
On；Adjustment parameter acquiring method for doubtful shadow spots is also in this way, only needing first to judge whether it is shadow spots, specific step
Suddenly are as follows:
1) to any pixel (x, y), the gray value on kth width image is Ik(k=1,2,3,4), by each image
The gray value of corresponding pixel points (x, y) normalizes:
2) by i1、i2、i3、i4It is ranked up, subtracts time maximum and time minimum number mean value with maximum number, it is assumed that i1>
i2> i3> i4, definition:
Y1=i1-0.5*(i2+i3)
Y2=0.5* (i2+i3)-i4
Adjusting parameter k1、k2Is defined as:
k1=sqrt (2*Y1-Y1*Y1)
k2=sqrt (2*Y2-Y2*Y2)
3) gray value is adjusted:
A, directionality, i are had the characteristics that according to mirror-reflection1、i2、i3、i4In the corresponding gray value of maximum number most have can
It can have occurred reflective, the corresponding gray value of the smallest number, which is likely to occur, to be blocked, other corresponding gray value mirrors of two parameters
Face reflex components are low, and the probability that is blocked is small, are approximately diffusing reflection, therefore do not need to adjust, Ik′For adjustment after gray value (k=1,
2,3,4), it may be assumed that
I2′=I2
I3′=I3
B, for I4, gray scale adjusted is sought according to adjusting parameter are as follows:
I4′=(1-k1) * I4+0.5*k1*(I2+I3)
C, for I1, shade may be formd due to blocking, determination strategy takes threshold decision here, when its is small
When given threshold value th (such as th=3), it need to be adjusted, gray scale after adjustment are as follows:
I1′=(1-k2) * I1+0.5*k2*(I2+I3)
3, photometric stereo solves albedo figure:
1) mikey vector is converted by each light source radiating angle, conversion formula is as follows:
X=cos (slant) * sin (tilt)
Y=sin (slant) * sin (tilt)
Z=cos (tilt)
2) the albedo figure for solving removal iridescent image pixel-by-pixel, first by the mikey moment of a vector of four light source directions
Array forms the direction matrix M of 4*34*3, four width images of gray scale adjustment are formed into corresponding gray vector I4*1.It is vertical by luminosity
Gray vector and irradiation vector are formed over-determined systems by body algorithm:
I4*1=β * M4*3*N3*1
Wherein, β is albedo, N3*1For the surface unit normal vector of the point.Both sides are multiplied by M4*3Transposition, it may be assumed that
Due to N3*1For normalized vector, so β are as follows:
The albedo for calculating each point obtains albedo figure, thus obtains removing reflective image.
Wherein (a) in Fig. 4, Fig. 6, (b), (c), (d) are respectively the image that four cameras are respectively shot, and Fig. 5 and Fig. 7 are
The reflective effect picture of corresponding removal, image sources to be processed are detected in rice coding.As can be seen that the present embodiment removal is reflective
Effect it is obvious, for it is complicated it is reflective have stronger robustness, remove reflective image and illumination and reached uncoupling
Effect, effect picture are very identical with actual object.
Schematically the present invention and embodiments thereof are described above, description is not limiting, institute in attached drawing
What is shown is also one of embodiments of the present invention, and actual structure is not limited to this.So if the common skill of this field
Art personnel are enlightened by it, without departing from the spirit of the invention, are not inventively designed and the technical solution
Similar frame mode and embodiment, are within the scope of protection of the invention.
Claims (6)
1. a kind of reflective method of removal image based on photometric stereo, it is characterised in that: successively irradiate lower pair using four light sources
Same target shoots four width images, solves to obtain albedo figure with photometric stereo after pretreatment, replaces corresponding picture with albedo
The gray value of element, to remove the reflective of image；Specifically: using camera shooting direction as optical axis, in the periphery installation four of optical axis
A consistent light source of light intensity, four light sources determine plane perpendicular to optical axis, meanwhile, light source direction of illumination is converted into a bit, and
The point is located on optical axis, and the irradiating angle of each light source is determined by Source calibration；In the item that four light sources are successively lighted respectively
A width picture is respectively shot to same target under part, white noise is removed to resulting four width image preprocessing with gaussian filtering, so
The gray scale adjusting parameter for seeking each pixel afterwards, according to parameter to there may be the gray values of reflective point to be adjusted；Then
Treated four width images by photometric stereo algorithm composition overdetermined equation are sought into albedo figure, can be obtained and remove reflective figure
Picture.
2. a kind of reflective method of removal image based on photometric stereo according to claim 1, it is characterised in that: light source shines
Penetrating angle calibration, detailed process is as follows:
1) Archon is placed at the position of testee, adjusting light-source brightness makes Archon not overexposure, shoots piece image；
2) the position O (x of the centre of sphere is found by the method for Hough circle fittingo, yo) and radius of a ball ro；
3) Archon is partitioned into according to the centre of sphere and the radius of a ball, the maximum point L (x of gray value is found in Archonl, yl)；
4) the horizontal angle of direction of illumination and camera imaging is denoted as slant, the angle of direction of illumination and camera shooting axis
Be denoted as tilt, then:
Slant=arctan ((yl-yo)/(xl-xo))
Tilt=2*arcsin (sqrt ((yl-yo)^2+(Xl-Xo)^2))
5) above-mentioned steps are carried out to each light source, seeks the tilt and slant of each light source.
3. a kind of reflective method of removal image based on photometric stereo according to claim 2, it is characterised in that: light source shines
It penetrates in the step 3) of angle calibration, the maximum region of gray value is found in Archon, continue to find the region with Hough circle fitting
The center of circle, by this center of circle treat as the maximum point of gray value.
4. a kind of reflective method of removal image based on photometric stereo according to claim 2, it is characterised in that: seek every
The process of a gray scale adjusting parameter are as follows:
1) to any pixel (x, y), the gray value on kth width image is Ik, k=1,2,3,4, each image gray value is returned
One changes:
2) by i1、i2、i3、i4It is ranked up, subtracts time maximum and time minimum number mean value with maximum number, it is assumed that i1> i2> i3>
i4, definition:
Y1=i1-0.5*(i2+i3)
Y2=0.5* (i2+i3)-i4
Adjusting parameter k1、k2Is defined as:
k1=sqrt (2*Y1-Y1*Y1)
k2=sqrt (2*Y2-Y2*Y2)。
5. a kind of reflective method of removal image based on photometric stereo according to claim 4, it is characterised in that: according to tune
Whole parameter is as follows to the process being adjusted there may be the gray value of reflective point:
1) directionality, i are had the characteristics that according to mirror-reflection1、i2、i3、i4In the corresponding gray value of maximum number most possibly send out
Given birth to reflective, the corresponding gray value of the smallest number is possible to be blocked, other corresponding gray value mirror-reflections of two parameters at
Point low, it is approximately diffusing reflection that the probability that is blocked is small, therefore does not need to adjust, Ik′For adjustment after gray value, k=1,2,3,4, it may be assumed that
I2′=I2
I3′=I3
2) for I4, gray scale adjusted is sought according to adjusting parameter are as follows:
I4′=(1-k1)*I4+0.5*k1*(I2+I3)
3) for I1, shade may be formd due to blocking, determination strategy takes threshold decision, when it is less than given threshold
When value th, it need to be adjusted, gray scale after adjustment are as follows:
I1′=(1-k2)*I1+0.5*k2*(I2+I3)。
6. a kind of reflective method of removal image based on photometric stereo according to claim 5, it is characterised in that: will handle
Four width images afterwards are sought the detailed process of albedo figure by photometric stereo algorithm composition overdetermined equation are as follows:
1) mikey vector is converted by each light source radiating angle, conversion formula is as follows:
X=cos (slant) * sin (tilt)
Y=sin (slant) * sin (tilt)
Z=cos (tilt)
2) by the direction matrix M of the irradiation vector composition 4*3 of four light sources4*3, the four width images composition of gray scale adjustment is corresponding
Gray vector I4*1；
3) gray vector and irradiation vector are formed into over-determined systems by photometric stereo algorithm:
I4*1=β * M4*3*N3*1
Wherein, β is albedo, N3*1For the surface unit normal vector of the point；
4) above formula the right and left is multiplied by M4*3Transposition, it may be assumed that
Due to N3*1For normalized vector, so β are as follows:
The β for finding out each point obtains albedo figure, as removes reflective image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710177279.4A CN107194881B (en) | 2017-03-23 | 2017-03-23 | A kind of reflective method of removal image based on photometric stereo |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710177279.4A CN107194881B (en) | 2017-03-23 | 2017-03-23 | A kind of reflective method of removal image based on photometric stereo |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107194881A CN107194881A (en) | 2017-09-22 |
| CN107194881B true CN107194881B (en) | 2019-12-03 |
Family
ID=59870866
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710177279.4A Active CN107194881B (en) | 2017-03-23 | 2017-03-23 | A kind of reflective method of removal image based on photometric stereo |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107194881B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108154491B (en) * | 2018-01-26 | 2021-04-23 | 上海觉感视觉科技有限公司 | Image reflection eliminating method |
| CN108489418B (en) * | 2018-03-01 | 2020-02-14 | 北京农业智能装备技术研究中心 | Seedling leaf three-dimensional form measuring method and device |
| CN108647554A (en) * | 2018-05-15 | 2018-10-12 | 佛山市南海区广工大数控装备协同创新研究院 | A kind of sport footwear air cushion stereoscopic two-dimensional code identification and detection device and method |
| CN109059800B (en) * | 2018-05-24 | 2020-04-24 | 东南大学 | Light source position calibration method of three-dimensional reconstruction device |
| CN111983874A (en) * | 2020-08-19 | 2020-11-24 | 四川迈视崔数字科技有限公司 | Method for obtaining object surface texture based on photometric stereo method and photostudio |
| CN112884689B (en) * | 2021-02-25 | 2023-11-17 | 景德镇陶瓷大学 | Method for removing high light of strong reflection surface image |
| CN113096057B (en) * | 2021-04-13 | 2022-08-19 | 南京大学 | Highlight elimination method based on four-light-source photometric stereo method |
| CN114345841B (en) * | 2021-12-31 | 2022-10-04 | 苏州德威尔卡光电技术有限公司 | Control method in laser cleaning process and laser cleaning control system |
| CN115484408A (en) * | 2022-09-07 | 2022-12-16 | 北京拙河科技有限公司 | Snow surface reflection coefficient generation method and device based on high-precision camera shooting |
| CN116071807B (en) * | 2023-03-06 | 2023-06-16 | 深圳市网联天下科技有限公司 | Campus card intelligent early warning method and system based on video monitoring |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104778749A (en) * | 2015-04-07 | 2015-07-15 | 浙江大学 | Group sparsity based photometric stereo method for realizing non-Lambert object reconstruction |
| CN105184765A (en) * | 2014-06-09 | 2015-12-23 | 株式会社其恩斯 | Inspection Apparatus, Inspection Method, And Program |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101087180B1 (en) * | 2011-06-22 | 2011-11-28 | 동국대학교 경주캠퍼스 산학협력단 | Reliable extraction scheme of 3 dimensional shape of metallic surface |
| JP6424020B2 (en) * | 2014-06-09 | 2018-11-14 | 株式会社キーエンス | Image inspection apparatus, image inspection method, image inspection program, computer-readable recording medium, and recorded apparatus |
-
2017
- 2017-03-23 CN CN201710177279.4A patent/CN107194881B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105184765A (en) * | 2014-06-09 | 2015-12-23 | 株式会社其恩斯 | Inspection Apparatus, Inspection Method, And Program |
| CN104778749A (en) * | 2015-04-07 | 2015-07-15 | 浙江大学 | Group sparsity based photometric stereo method for realizing non-Lambert object reconstruction |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107194881A (en) | 2017-09-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107194881B (en) | A kind of reflective method of removal image based on photometric stereo | |
| TWI451079B (en) | Method of measuring measurement target | |
| US11238303B2 (en) | Image scanning method for metallic surface and image scanning system thereof | |
| CN109490311B (en) | Backlight panel defect detection system and method based on multi-angle shooting | |
| CN109523541A (en) | A kind of metal surface fine defects detection method of view-based access control model | |
| JP5145555B2 (en) | Pupil detection method | |
| US20040047491A1 (en) | Image capturing device with reflex reduction | |
| CN107615049B (en) | Inspection illumination device and inspection system | |
| CN108445007A (en) | A kind of detection method and its detection device based on image co-registration | |
| CN110441323B (en) | Product surface polishing method and system | |
| CN103370598B (en) | For in optically scanning surfaces region or on the method at edge | |
| JPH0782539B2 (en) | Pupil imager | |
| CN109668904A (en) | A kind of optical element flaw inspection device and method | |
| CN110262092A (en) | A kind of liquid crystal display detection method based on NI Vision Builder for Automated Inspection | |
| CN108490001A (en) | The inspection of haptic lens in plastic casing | |
| CN113686879A (en) | Optical film defect visual detection system and method | |
| CN100340840C (en) | Method and device for optical form measurement and/or estimation | |
| CN212301365U (en) | Light source device and defect detection system | |
| KR101987922B1 (en) | Automatic system for capturing image by controlling light cell actively | |
| CN111999313A (en) | Light source device, defect detection method and defect detection system | |
| US20230243758A1 (en) | Tilt and focusing adjustment for consistent gem imaging | |
| CN208013080U (en) | One kind detecting Plastic foreign fiber device based on polarization state detection technique | |
| CN111344553A (en) | Defect detection method and system for curved surface object | |
| US11657495B2 (en) | Non-lambertian surface inspection system using images scanned in different directions | |
| US20040114035A1 (en) | Focusing panel illumination method and apparatus |
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 |