CN109709103A - A kind of the citrus early stage rotten fruit identifying system and method for ring-shaped stripe polishing imaging - Google Patents
A kind of the citrus early stage rotten fruit identifying system and method for ring-shaped stripe polishing imaging Download PDFInfo
- Publication number
- CN109709103A CN109709103A CN201910044898.5A CN201910044898A CN109709103A CN 109709103 A CN109709103 A CN 109709103A CN 201910044898 A CN201910044898 A CN 201910044898A CN 109709103 A CN109709103 A CN 109709103A
- Authority
- CN
- China
- Prior art keywords
- image
- citrus
- compounent
- ring
- phase
- 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.)
- Granted
Links
Landscapes
- Investigating Or Analysing Materials By Optical Means (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The present invention relates to optics and agricultural technology field, disclose the citrus early stage rotten fruit identifying system and method for a kind of ring-shaped stripe polishing imaging, it include: Digital light projector, visible near-infrared light source, computer, accurate moving stage and near infrared camera, wherein, phase image is demodulated by computer, to obtain DC component image and AC compounent image, based on DC component picture construction binaryzation template and remove the background of AC compounent image, carrying out picture breakdown and image reconstruction to the AC compounent image after going background using two-dimensional empirical mode decomposition enhances, to obtain reconstructed image, it is split based on reconstructed image and in conjunction with partitioning algorithm come the incipient decay region to citrus to be detected.Citrus early stage rotten fruit identifying system has the advantages that the position that recognition efficiency is high and can be accurately judged to rotten region.
Description
Technical field
The present invention relates to optics and agricultural technology field, more particularly to a kind of citrus early stage of ring-shaped stripe polishing imaging
Rotten fruit identifying system and method.
Background technique
China's citrus annual output surpasses 37,000,000 tons, occupies the first in the world.Rot caused by fungal infection be after citrus is adopted most
Easily occur, most serious, is but most difficult to the defect of detection.Presently relevant detection technique mainly has RGB computer vision technique, close red
External spectrum technology, multi-optical spectrum imaging technology, high light spectrum image-forming technology and Imaging-PAM etc..For RGB computer vision
Technology, the incipient decay fruit due to caused by fungal infection, infected zone coat color and normal fruit colour are about the same, lead
It causes extremely difficult using this defect of color camera detection.For near-infrared spectrum technique, although this technology can be to not
Same type tissue carries out model differentiation different characteristic information that is specific, effectively expressing, and establish according to spectral information, Jin Ershi
Now classify.However, near-infrared spectrum technique belongs to Single point detection techniques, only table can be carried out to measurement point area information
Reach, be difficult to make full use of in the application of the incipient decay defects detection caused by citrus fungal infection its fruit surface spatial information into
Row comprehensive descision is accurately classified with realizing.For more/high light spectrum image-forming technology, due to data information amount is big, system cost is high,
Real-time online detects the problems such as difficulty, and current this technology is caused to be also difficult to promote and apply in practice.With these technology phases
Than although fluorescent technique shows certain rotten fruit detection potentiality, not all rotten region all emits in practice
There is stability difference to make fluorescent technique in practical applications in fluorescence.
Summary of the invention
(1) technical problems to be solved
The object of the present invention is to provide the citrus early stage rotten fruit identifying system and method for a kind of ring-shaped stripe polishing imaging, with
Solve the technical problem that citrus fungal infection early stage rotten fruit in the prior art can not be effectively recognized and accuracy of identification is low.
(2) technical solution
In order to solve the above-mentioned technical problem, according to the first aspect of the invention, a kind of ring-shaped stripe polishing imaging is provided
Citrus early stage rotten fruit identifying system, comprising: Digital light projector；Visible near-infrared light source is thrown for being supplied to the digital light
Shadow instrument near infrared light；Computer, for generating two-dimensional annular stripe pattern and being transmitted to the Digital light projector, thus
So that the Digital light projector is launched fringe light and is radiated on citrus to be detected；Accurate moving stage, for accepting
Citrus to be detected simultaneously can drive the citrus to be detected to carry out side-to-side movement in the horizontal direction；And near infrared camera, for adopting
Collect the phase image of the citrus to be detected, wherein demodulate by the computer to the phase image, to obtain
DC component image and AC compounent image based on the DC component picture construction binaryzation template and remove AC compounent figure
The background of picture carries out picture breakdown to the AC compounent image after going background using two-dimensional empirical mode decomposition and image reconstruction increases
By force, to obtain reconstructed image, based on reconstructed image and in conjunction with partitioning algorithm come the incipient decay region to citrus to be detected into
Row segmentation.
Wherein, the early stage corruption fruit identifying system further includes that the front of the transmitting terminal of the Digital light projector is arranged in
First polarizing film.
Wherein, the early stage corruption fruit identifying system further includes that the narrowband filter of the camera lens front end of the near infrared camera is arranged in
Second polarizing film of wave plate and the front end that the narrow band filter slice is set.
Wherein, the accurate moving stage includes and the precision electric motor for calculating mechatronics and the accurate electricity
The rotation axis and be set in the periphery of the rotation axis and can be carried out along the axial direction of the rotation axis that the output end of machine is connected
The objective table of reciprocating motion.
Wherein, the computer includes motor control module, image capture module, projection control module and ring-shaped stripe
Image generation module, wherein the motor control module is electrically connected with the precision electric motor, for controlling the precision electric motor
Rotation and stopping；Described image acquisition module is electrically connected with the near infrared camera, for acquiring the phase of the citrus to be detected
Bit image；The projection control module is electrically connected with the Digital light projector, can be sent out for controlling the Digital light projector
Project ring light；The ring-shaped stripe image generation module is electrically connected with the projection control module, for generating two-dimensional annular
The two-dimensional annular stripe pattern is simultaneously loaded into the projection control module by stripe pattern.
Wherein, the first peace is configured between the center line of the Digital light projector and the center line of the near infrared camera
Clamping angle, the magnitude range of the first installation angle are more than or equal to 30 ° and to be less than or equal to 45 °.
According to a second aspect of the present application, a kind of citrus early stage rotten fruit identification side of ring-shaped stripe polishing imaging is also provided
Method, comprising: two-dimensional annular stripe pattern is generated using ring-shaped stripe image generation module；By the two-dimensional annular stripe pattern of generation
It is loaded into Digital light projector；The fringe light and accurate moving stage launched by Digital light projector, near-infrared
Camera acquires three phase images of citrus to be detected；Three phase images are demodulated to obtain DC component image respectively
With AC compounent image；Based on DC component picture construction binaryzation template and remove the background of AC compounent image；Using two
It ties up empirical mode decomposition and picture breakdown and image reconstruction enhancing is carried out to the AC compounent image after going background, to be reconstructed
Image；The morning that citrus to be detected is formed due to by fungal infection based on the reconstructed image of acquisition and in conjunction with conventional segmentation algorithm
Phase, rotten region was split.
Wherein, the conventional segmentation algorithm includes the one of which in watershed algorithm and global threshold method.
Wherein, the method also includes: by the rotation of the output end of computer control precise motor, thus described in driving
The rotation of rotation axis drives the objective table along the axial according to 2 π/3 of the rotation axis by the rotation of the rotation axis
Phase offset is moved.
Wherein, the generation formula of the two-dimensional annular stripe pattern is
Here, I indicates that two-dimensional annular stripe pattern, f are spatial frequency, x and y indicate circle in two-dimensional annular stripe pattern
Coordinate points on ring, DC indicate DC component image；AC indicates AC compounent image.
Wherein, when obtaining three phase images, the initial position of accurate moving stage is set first as phase 0, this
When, the first amplitude phase diagram picture is obtained, then, accurate moving stage travel(l)ing phase offset is 2 π/3, to obtain the second width phase
Image, travel(l)ing phase offset is 2 π/3 to accurate moving stage again, and to obtain third amplitude phase diagram picture, then, precision is mobile
Objective table returns to the initial position that phase is 0.
Wherein, the formula of the DC component image obtained after demodulation is
Obtain AC compounent image AC formula be
Wherein, P1, P2 and P3 respectively represent three
A phase image.
Wherein, the method for image reconstruction enhancing be AC compounent image by used after two-dimensional empirical mode decomposition formula for
AC1=(AC-IMF1+IMF2+MF3)/R carries out image reconstruction enhancing, wherein
AC1 is AC compounent image by decomposing and reconstructing enhanced image, and AC is AC compounent image, IMF1,
IMF2, IMF3 and R respectively represent AC compounent image the 1st, the 2nd and the 3rd caused by after two-dimensional empirical mode decomposition
A intrinsic mode function image and residual image.
(3) beneficial effect
The citrus early stage rotten fruit identifying system of ring-shaped stripe polishing imaging provided by the invention, compared with prior art, tool
It has the following advantages:
Launch fringe light by Digital light projector and be radiated on citrus to be detected, is existed by near infrared camera acquisition
Three phase images of the citrus to be detected under specific frequency, then, using demodulation techniques appropriate to the three of citrus to be detected
A phase image is demodulated to obtain crucial AC compounent image, which can highlight citrus to be detected
It is subcutaneous rot feature, further, using advanced two-dimensional empirical mode decomposition to the AC compounent image after going background into
It goes and decomposes and reconstruct enhancing, further to reach the normal region and the rotten region of early infection that enhance in AC compounent image
Contrast, the final identification for realizing the rotten region of the infection to citrus to be detected.
In addition, the citrus early stage rotten fruit identifying system of the application has the advantages that realize that relatively easy and recognition efficiency is high,
There is biggish application prospect in the detection of automation Quality Parameters in Orange.The application is to the synthesis for researching and developing high-end citrus fruit
Quality grading equipment, reduction citrus damaged and increased peasant income after adopting is of great significance.
Detailed description of the invention
Fig. 1 is the overall structure for the citrus early stage rotten fruit identifying system that the ring-shaped stripe polishing of embodiments herein is imaged
Schematic diagram；
Fig. 2 is the step process for the citrus early stage rotten fruit recognition methods that the ring-shaped stripe polishing of embodiments herein is imaged
Schematic diagram；
Fig. 3 is the two-dimensional annular stripe pattern generated using ring-shaped stripe image generation module；
Fig. 4 a is first phase image of near infrared camera sample collected；
Fig. 4 b is second phase image of near infrared camera sample collected；
Fig. 4 c is the third phase image of near infrared camera sample collected；
Fig. 5 a is the image of citrus to be detected；
Fig. 5 b is DC component image DC；
Fig. 5 c is AC compounent image AC；
Fig. 6 is the binary image obtained based on the DC component image DC in Fig. 5 b；
Fig. 7 is exemplary to AC compounent image AC progress picture breakdown and reconstruct enhancing based on two-dimensional empirical mode decomposition
Procedural image；
Fig. 8 a is for the first procedural image for reconstructing enhanced image progress watershed defect Segmentation；
Fig. 8 b is for the second procedural image for reconstructing enhanced image progress watershed defect Segmentation；
Fig. 8 c is for the third procedural image for reconstructing enhanced image progress watershed defect Segmentation；
Fig. 9 a is the image of example citrus with smaller infected area and when infected zone is in citrus edge to be detected；
Fig. 9 b is example citrus passes through image with smaller infected area and when infected zone is in citrus edge to be detected
Decompose and reconstruct enhanced AC1 image；
After Fig. 9 c is example citrus is divided with smaller infected area and when infected zone is in citrus edge to be detected
Infection rot area image.
In figure, 1: near infrared camera；2: camera lens；3: narrow band filter slice；4: the second polarizing films；5: computer；6: digital light
Projector；7: visible near-infrared light source；8: cut-off filter plate；9: the first polarizing films；20: accurate moving stage；10: accurate electricity
Machine；11: rotation axis；12: objective table；13: optical fiber；14: citrus to be detected.
Specific embodiment
With reference to the accompanying drawings and examples, specific embodiments of the present invention will be described in further detail.Following instance
For illustrating the present invention, but it is not intended to limit the scope of the invention.
In the description of the present invention, it should be noted that unless otherwise clearly defined and limited, term " installation ", " phase
Even ", " connection " shall be understood in a broad sense, for example, it may be being fixedly connected, may be a detachable connection, or be integrally connected；It can
To be mechanical connection, it is also possible to be electrically connected；It can be directly connected, can also can be indirectly connected through an intermediary
Connection inside two elements.For the ordinary skill in the art, above-mentioned term can be understood at this with concrete condition
Concrete meaning in invention.
As shown in Figure 1, schematically showing that citrus early stage rotten fruit identifying system includes near infrared camera 1, number in figure
Word light projector 6, visible near-infrared light source 7, computer 5, accurate moving stage 20 and citrus to be detected 14.
In embodiments herein, the visible near-infrared light source 7 is for being supplied to 6 near-infrared of Digital light projector
Light.
Computer 5 is for generating two-dimensional annular stripe pattern and being transmitted to the Digital light projector 6, so that should
Digital light projector 6 is launched fringe light and is radiated on citrus 14 to be detected.
Accurate moving stage 20 is for accepting citrus 14 to be detected and can drive the citrus 14 to be detected in the horizontal direction
Carry out side-to-side movement.In this way, by making the movement of accurate moving stage 20, so that the citrus 14 to be detected be driven to be transported
It is dynamic, by the movement of the citrus 14 to be detected, it can be convenient near infrared camera 1 and shoot citrus 14 to be detected and be in out of phase
Phase image.
Near infrared camera 1 is used to acquire the phase image of the citrus 14 to be detected, wherein by the computer 5 to the phase
Bit image is demodulated, to obtain DC component image DC and AC compounent image AC, is based on the DC component image DC structure
It builds binaryzation template and removes the background of AC compounent image AC, using two-dimensional empirical mode decomposition to the exchange after going background point
Spirogram carries out picture breakdown and image reconstruction enhancing as AC, to obtain reconstructed image, based on reconstructed image and segmentation is combined to calculate
Method is split the incipient decay region of citrus 14 to be detected.Specifically, fringe light is launched by Digital light projector 6
And be radiated on citrus 14 to be detected, pass through three phases of to be detected citrus 14 of the acquisition of near infrared camera 1 under specific frequency
Then bit image uses demodulation techniques appropriate to demodulate to obtain key three phase images of citrus 14 to be detected
AC compounent image AC, AC compounent image AC can highlight citrus 14 to be detected it is subcutaneous rot feature, further
Ground is decomposed and is reconstructed enhancing to the AC compounent image AC after going background using advanced two-dimensional empirical mode decomposition, with
Further reach the contrast of the normal region and the rotten region of early infection in enhancing AC compounent image AC, final realization pair
The identification in the rotten region of the infection of citrus 14 to be detected.
In addition, the citrus early stage rotten fruit identifying system of the application has the advantages that realize that relatively easy and recognition efficiency is high,
There is biggish application prospect in the detection of automation Quality Parameters in Orange.The application is to the synthesis for researching and developing high-end citrus fruit
Quality grading equipment, reduction citrus damaged and increased peasant income after adopting is of great significance.
As shown in Figure 1, early stage corruption fruit identifying system further includes setting in the comparison preferred embodiment of the application
Set the first polarizing film 9 in the front of the transmitting terminal of the Digital light projector 6.It should be noted that since " polarizing film " is one
Kind can make natural light become the optical element of polarised light, have the function of covering and penetrate to incident light, there is black and white and colour
Two classes can be divided into transmission, transflector again by application and counter transmit three classes.
As shown in Figure 1, to advanced optimize the early stage corruption fruit identifying system in above-mentioned technical proposal, in above-mentioned technical proposal
On the basis of, early stage corruption fruit identifying system further includes the narrow band filter slice 3 that 2 front end of camera lens of the near infrared camera 1 is arranged in
With the second polarizing film 4 of the front end that the narrow band filter slice 3 is arranged in.It should be noted that passing through the mirror in the near infrared camera 1
The narrow band filter slice 3 is added in head front end, so as to play the role of only the light in particular range of wavelengths being allowed to pass through.It is so-called
" specific wavelength " refer to wavelength of the wavelength size between 805 nanometers to 815 nanometers.
The central wavelength of the narrow band filter slice 3 is 810nm (nanometer), and wave width is 10nm (nanometer), in this way, filtering in the narrowband
The image of the light imaging by the narrow band filter slice 3 can be only obtained with the help of wave plate 3, near infrared camera 1, that is,
Acquisition central wavelength is the near-infrared image of 810nm (nanometer).Here, selecting central wavelength for the near-infrared of 810nm (nanometer)
Image, on the one hand, because near-infrared image is insensitive to the color of fruit surface, on the other hand, 810nm (nanometer) is citrus corruption
The sensitive wave length in rotten region.By being equipped with the second polarizing film 4 in the front of the narrow band filter slice 3, second polarizing film 4 is to close red
The light of exterior domain is effective, and near infrared camera 1 is connect with computer 5 by data line.
Meanwhile in citrus early stage rotten fruit identifying system shown in Fig. 1, installed in the front of the visible near-infrared light source 7
There is cut-off filter plate 8, which has cut-off effect to 780nm (nanometer) light below, (receives to guarantee to be greater than 780nm
Rice) light can be smoothly through optical fiber 13 enter Digital light projector 6 in, be only into the light in Digital light projector 6
Near infrared light is equipped with the first polarizing film 9 for matching with the second polarizing film 4 in the front end of the Digital light projector 6, this is second partially
Vibration piece 4 and the first polarizing film 9 collectively form polarizer group, pass through the cooperation of the second polarizing film 4 and the first polarizing film 9, Ke Yiyou
Eliminate the high speck (also referred to as mirror-reflection spot) that near infrared light is formed on similar spherical fruit surface, Digital light projector in effect ground
6 are connect with computer 5 by data line.
As shown in Figure 1, to advanced optimize the accurate moving stage 20 in above-mentioned technical proposal, in above-mentioned technical proposal
On the basis of, which includes the precision electric motor 10 being electrically connected with the computer 5 and the precision electric motor 10
The rotation axis 11 and be set in the periphery of the rotation axis 11 and can be carried out along the axial direction of the rotation axis 11 past that output end is connected
The objective table 12 moved again.It should be noted that the output end of the precision electric motor 10 is that coaxial line is connect with rotation axis 11, preferably
The secured connection of axle sleeve realization between the two can be used in ground.Specifically, by starting the precision electric motor 10 and making the precision electric motor
10 output end carries out axial rotation, is rotated by the axial direction of the output end, will drive the axial rotation of the rotation axis 11, lead to
The axial rotation for crossing the rotation axis 11, will make the objective table 12 carry the citrus 14 to be detected along the axis of the rotation axis 11
To moving back and forth, further, it is ensured that near infrared camera 1 can shoot the phase that citrus 14 to be detected is in out of phase
Bit image.
In a preferred embodiment, which includes motor control module, image capture module, projection control
Module and ring-shaped stripe image generation module, wherein the motor control module is electrically connected with the precision electric motor 10, for controlling
The rotation and stopping of the precision electric motor 10.
The image capture module is electrically connected with the near infrared camera 1, for acquiring the phase image of the citrus 14 to be detected.
The projection control module is electrically connected with the Digital light projector 6, can be launched for controlling the Digital light projector 6
Ring light.
The ring-shaped stripe image generation module is electrically connected with the projection control module, for generating two-dimensional annular stripe pattern
And the two-dimensional annular stripe pattern is loaded into the projection control module.
In another preferred embodiment, the center line of the center line of the Digital light projector 6 and the near infrared camera 1
Between be configured with the first installation angle, this first installation angle magnitude range be more than or equal to 30 ° and be less than or equal to 45 °.Its
In, which is preferably 33 °, 36 °, 39 ° or 42 °.
It should be noted that by controlling the magnitude range of the first installation angle between 30 ° to 45 °, so as to
So that the fringe light launched through the Digital light projector 6 can accurately be radiated at the outer surface of citrus 14 to be detected and obtain
Obtain optimal phase image.
As shown in Fig. 2,3,4a, 4b, 4c, 5a, 5b, 5c, 6,7,8a, 8b, 8c, 9a, 9b and 9c it is found that according to the application
Second aspect, a kind of citrus early stage rotten fruit recognition methods of ring-shaped stripe polishing imaging is also provided, this method comprises: step
S1 generates two-dimensional annular stripe pattern using ring-shaped stripe image generation module.
The two-dimensional annular stripe pattern of generation is loaded into Digital light projector 6 by step S2.
Step S3, the fringe light launched by Digital light projector 6 and accurate moving stage 20, near infrared camera
Three phase images of 1 acquisition citrus 14 to be detected.
Step S4 respectively demodulates to obtain DC component image DC and AC compounent image three phase images
AC。
Step S5 constructs binaryzation template based on DC component image DC and removes the background of AC compounent image AC.
Step S6 carries out picture breakdown and figure to the AC compounent image AC after going background using two-dimensional empirical mode decomposition
As reconstruct enhancing, to obtain reconstructed image.
Step S7, based on the reconstructed image of acquisition and in conjunction with conventional segmentation algorithm come to citrus 14 to be detected because by fungi sense
The incipient decay region for contaminating and being formed is split.
In a preferred embodiment, the conventional segmentation algorithm includes in watershed algorithm and global threshold method
It is one of.It should be noted that due to watershed algorithm and global threshold method be it is well-known to those skilled in the art, be
For the sake of saving length, it is not detailed herein.
In a specific embodiment, which can be Matlab software.
In a specific example, orange is a kind of citrus fruit, has very important economic value in China.
Therefore, it is illustrated in the present embodiment using orange as object.
Firstly, prepare 100 orange samples, each 50 of the incipient decay fruit including being formed after normal fruit and fungal infection,
Wherein, fungal infection fruit is obtained using the method for artificial infection, method particularly includes: it falls ill from natually morbid orange fruit
Place cuts the pericarp tissue of infection fungi, using spore under aseptic water washing, spore is dissolved in sterile water and forms spores solution, so
Afterwards using disposable syringe to 50 normal fruit inoculating spores solution, being inoculated with depth is about subcutaneous 5mm (millimeter), Mei Gecheng
The amount of solution of sub- inoculating spores is 0.05ml (milliliter).Then, inoculated sample is placed on incubator (environment temperature 25-
27 degree, relative humidity be 96%~98%) in 2 days, is formed diameter for 5~15mm (millimeter) infected zone, at this point, infected area
The naked eyes in domain are relatively difficult to.
Then, all samples are carried out using citrus early stage rotten fruit identifying system (as shown in Figure 1) proposed by the invention
Image Acquisition.
When the sample image to citrus 14 to be detected is acquired, firstly, as shown in Figure 3 using Matlab Software Create
Two-dimensional annular stripe pattern, the two-dimensional annular stripe pattern generate formula be
Here, I indicates that two-dimensional annular stripe pattern, f are spatial frequency, x and y indicate circle in two-dimensional annular stripe pattern
Coordinate points on ring, DC indicate DC component image；AC indicates AC compounent image.Wherein, f is equal to 0.15mm-1。
Then, the two-dimensional annular stripe pattern of generation is loaded into Digital light projector 6, starts Digital light projector 6
Two-dimensional annular striped polishing can be thus achieved.
Then, citrus 14 to be detected is placed on the objective table 12 of accurate moving stage 20, is set accurate mobile
The initial position of objective table 20 is phase 0, at this point, the collected sample image of near infrared camera 1 is the first amplitude phase diagram picture (ginseng
As shown in Fig. 4 a).
Then, under the control of the precision electric motor 10 of accurate moving stage 20, objective table 12 along rotation axis 11 according to
Certain phase offset is moved, phase offset be 2 π/3, by near infrared camera 1 obtain the second amplitude phase diagram picture (referring to
Shown in Fig. 4 b).
Again, promote objective table 12 to deviate 2 π/3 along 11 travel(l)ing phase of rotation axis, obtain third amplitude phase diagram picture (referring to figure
Shown in 4c).
Then, accurate moving stage 20 returns to the initial position that phase is 0.
After obtaining three amplitude phase diagram pictures, three phase images of citrus 14 to be detected need to be demodulated, to obtain direct current
Component image DC and AC compounent image AC, wherein phase demodulating obtain DC component image DC formula be
Obtain AC compounent image AC formula be
Wherein P1, P2 and P3 respectively represent three phase images.Fig. 5 a indicates citrus 14 to be detected in the present embodiment,
In, encircled is that (original image is colour, and fungal infection region is slightly deep yellow, with orange peel without obvious right for infected zone
According to), very indiscernible rotten region is in AC compounent image AC (as shown in Figure 5 c) on citrus to be detected 14 in fig 5 a
It is presented.
In order to further increase the normal pericarp of citrus 14 to be detected and the control of infected zone, here, to AC compounent
Image AC is further processed.
Firstly, Fig. 5 b DC component image DC combination single threshold theoretical (threshold value can be 20) is used to obtain binaryzation template figure
Then picture by the binaryzation template image multiplied by AC compounent image AC, moves the background of AC compounent image AC
It removes.Based on the AC compounent image AC after background is removed, picture breakdown is carried out using two-dimensional empirical mode decomposition, preceding 3 are taken after decomposition
A intrinsic mode function image and residual image R, wherein the 1st intrinsic mode function image represents noise, and the 2nd and the 3rd
A intrinsic mode function image represents details, and residual image R represents 14 surface brightness information of citrus to be detected.
Then, use formula for AC1=(AC-IMF1+IMF2+MF3)/R,
Carry out image reconstruction enhancing, wherein AC1 is AC compounent image by decomposing and reconstructing enhanced image, AC
For AC compounent image, IMF1, IMF2, IMF3 and R respectively represent AC compounent image and are produced after two-dimensional empirical mode decomposition
The 1st, the 2nd and the 3rd raw intrinsic mode function image and residual image.
Fig. 7 shows that carrying out picture breakdown and reconstruct to AC compounent image AC based on two-dimensional empirical mode decomposition enhances,
AC1 is AC compounent image AC by decomposing and reconstructing enhanced image.Image AC1 and image AC are compared, held very much
The brightness of easily image its image entirety of the discovery after two-dimensional empirical mode decomposition and reconstruct is more uniform, and sample is normal
It is more obvious to presenting a note between region and rotten region.
The early infection region that orange fungal infection is formed is divided using fractional spins based on image AC1
It cuts, segmentation result is as shown in Fig. 8 a, 8b and 8c, wherein Fig. 8 a is indicated using the watershed formed after fractional spins
Cut-off rule can clearly be seen that rotten region is effectively split from Fig. 8 a, and Fig. 8 b is in removal Fig. 8 a
Only retain the image of rotten region segmentation line behind orange boundary, Fig. 8 c is the rotten region of infection of the practical orange for being divided and.
In order to further verify the performance of the application, as special case, orange surface infection region is very small (less than 5 millis
Rice), also, the infected zone is located at fruit edge, as illustrated in fig. 9, wherein encircled is infected zone (since fruit is
It is spherical, cause surface uneven illumination even, usually intermediate bright, edge is dark, in this way, when infected zone is in fruit edge
Wait, defect Segmentation is most difficult), at this point, carry out image deflects segmentation detection be it is the most difficult, according to previous embodiment
Identical method, as shown in figs. 9 b and 9 c, Fig. 9 b is by picture breakdown and to reconstruct enhanced AC1 image to testing result, figure
9c is the rotten area image of infection obtained after being divided.
From testing result as can be seen that technology provided by the invention will not be infected the influence of region present position, i.e.,
So that infected zone is located at the most edge of orange in image, can also obtain satisfactory result.
Using technology proposed by the invention, to the identification knot of 100 samples (normal fruit and each 50 of fungal infection fruit)
Fruit is as shown in table 1, and the accuracy of identification is 100%, to further demonstrate effectiveness of the invention.
Table 1
In conclusion launching fringe light by Digital light projector 6 and being radiated on citrus 14 to be detected, by close red
Three phase images of to be detected citrus 14 of the outer acquisition of camera 1 under specific frequency, then, using demodulation techniques pair appropriate
Three phase images of citrus 14 to be detected are demodulated to obtain crucial AC compounent image AC, AC compounent image AC
The subcutaneous feature of rotting that citrus 14 to be detected can be highlighted, further, using advanced two-dimensional empirical mode decomposition to going
AC compounent image AC after background is decomposed and is reconstructed enhancing, further to reach in enhancing AC compounent image AC just
The contrast in normal region and the rotten region of early infection, the final identification for infecting region of rotting realized to citrus 14 to be detected.
In addition, the citrus early stage rotten fruit identifying system of the application has the advantages that realize that relatively easy and recognition efficiency is high,
There is biggish application prospect in the detection of automation Quality Parameters in Orange.The application is to the synthesis for researching and developing high-end citrus fruit
Quality grading equipment, reduction citrus damaged and increased peasant income after adopting is of great significance.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Within mind and principle, any modification, equivalent replacement, improvement and so on be should all be included in the protection scope of the present invention.
Claims (13)
1. a kind of citrus early stage rotten fruit identifying system of ring-shaped stripe polishing imaging characterized by comprising
Digital light projector；
Visible near-infrared light source, for being supplied to the Digital light projector near infrared light；
Computer, for generating two-dimensional annular stripe pattern and being transmitted to the Digital light projector, so that described
Digital light projector is launched fringe light and is radiated on citrus to be detected；
Accurate moving stage, for accepting citrus to be detected and the citrus to be detected can be driven to be controlled in the horizontal direction
Movement；And
Near infrared camera, for acquiring the phase image of the citrus to be detected, wherein by the computer to the phase
Image is demodulated, to obtain DC component image and AC compounent image, is based on the DC component picture construction two-value
Change template and remove the background of AC compounent image, using two-dimensional empirical mode decomposition to the AC compounent image after going background into
Row picture breakdown and image reconstruction enhancing, so that reconstructed image is obtained, based on reconstructed image and in conjunction with partitioning algorithm come to be checked
The incipient decay region for surveying citrus is split.
2. the citrus early stage rotten fruit identifying system of ring-shaped stripe polishing imaging according to claim 1, which is characterized in that institute
Stating early stage rotten fruit identifying system further includes the first polarizing film that the front of the transmitting terminal of the Digital light projector is arranged in.
3. the citrus early stage rotten fruit identifying system of ring-shaped stripe polishing imaging according to claim 1, which is characterized in that institute
Stating early stage rotten fruit identifying system further includes that the narrow band filter slice of the camera lens front end of the near infrared camera is arranged in and is arranged in institute
State the second polarizing film of the front end of narrow band filter slice.
4. the citrus early stage rotten fruit identifying system of ring-shaped stripe polishing imaging according to claim 1, which is characterized in that institute
Stating accurate moving stage includes being connected with the precision electric motor for calculating mechatronics, with the output end of the precision electric motor
Rotation axis and the objective table that is set in the periphery of the rotation axis and can move back and forth along the axial direction of the rotation axis.
5. the citrus early stage rotten fruit identifying system of ring-shaped stripe polishing imaging according to claim 4, which is characterized in that institute
Computer is stated to include motor control module, image capture module, project control module and ring-shaped stripe image generation module,
In, the motor control module is electrically connected with the precision electric motor, for controlling the rotation and stopping of the precision electric motor；
Described image acquisition module is electrically connected with the near infrared camera, for acquiring the phase image of the citrus to be detected；
The projection control module is electrically connected with the Digital light projector, can be launched for controlling the Digital light projector
Ring light；
The ring-shaped stripe image generation module is electrically connected with the projection control module, for generating two-dimensional annular stripe pattern
And the two-dimensional annular stripe pattern is loaded into the projection control module.
6. the citrus early stage rotten fruit identifying system of ring-shaped stripe polishing imaging according to any one of claim 1 to 5,
It is characterized in that, the first installation folder is configured between the center line of the Digital light projector and the center line of the near infrared camera
Angle, the magnitude range of the first installation angle are more than or equal to 30 ° and to be less than or equal to 45 °.
7. a kind of citrus early stage rotten fruit recognition methods of ring-shaped stripe polishing imaging characterized by comprising use ring-shaped stripe
Image generation module generates two-dimensional annular stripe pattern；
The two-dimensional annular stripe pattern of generation is loaded into Digital light projector；
The fringe light and accurate moving stage, near infrared camera launched by Digital light projector acquire citrus to be detected
Three phase images；
Three phase images are demodulated to obtain DC component image and AC compounent image respectively；
Based on DC component picture construction binaryzation template and remove the background of AC compounent image；
Carrying out picture breakdown and image reconstruction to the AC compounent image after going background using two-dimensional empirical mode decomposition enhances, from
And obtain reconstructed image；
The morning that citrus to be detected is formed due to by fungal infection based on the reconstructed image of acquisition and in conjunction with conventional segmentation algorithm
Phase, rotten region was split.
8. the method according to the description of claim 7 is characterized in that the conventional segmentation algorithm includes watershed algorithm and the overall situation
One of which in threshold method.
9. the method according to the description of claim 7 is characterized in that the method also includes: pass through computer control precise electricity
The rotation of the output end of machine, by the rotation of the rotation axis, drives the objective table to drive the rotation of the rotation axis
Axial along the rotation axis is moved according to the phase offset of 2 π/3.
10. the method according to the description of claim 7 is characterized in that the generation formula of the two-dimensional annular stripe pattern is
Here, I indicates two-dimensional annular stripe pattern, f is spatial frequency, and x and y are indicated in two-dimensional annular stripe pattern on annulus
Coordinate points, DC indicate DC component image；AC indicates AC compounent image.
11. the method according to the description of claim 7 is characterized in that setting is accurate first moves when obtaining three phase images
The initial position of dynamic object stage is phase 0, at this point, obtaining the first amplitude phase diagram picture, then, accurate moving stage travel(l)ing phase
Offset is 2 π/3, and to obtain the second amplitude phase diagram picture, travel(l)ing phase offset is 2 π/3 to accurate moving stage again, to obtain the
Three amplitude phase diagram pictures, then, accurate moving stage return to the initial position that phase is 0.
12. the method according to the description of claim 7 is characterized in that the formula of the DC component image obtained after demodulation is
Obtain AC compounent image AC formula be
Wherein, P1, P2 and P3 respectively represent three phases
Bit image.
13. the method according to the description of claim 7 is characterized in that the method for image reconstruction enhancing is that AC compounent image is logical
Cross after two-dimensional empirical mode decomposition use formula for
AC1=(AC-IMF1+IMF2+MF3)/R
Carry out image reconstruction enhancing, wherein AC1 is AC compounent image by decomposition and reconstructs enhanced image, and AC is to hand over
Flow component image, IMF1, IMF2, IMF3 and R respectively represent AC compounent image caused by after two-dimensional empirical mode decomposition
1st, the 2nd and the 3rd intrinsic mode function image and residual image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910044898.5A CN109709103B (en) | 2019-01-17 | 2019-01-17 | Orange early-stage rotting identification system and method based on annular stripe polishing imaging |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910044898.5A CN109709103B (en) | 2019-01-17 | 2019-01-17 | Orange early-stage rotting identification system and method based on annular stripe polishing imaging |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109709103A true CN109709103A (en) | 2019-05-03 |
| CN109709103B CN109709103B (en) | 2020-08-11 |
Family
ID=66262414
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910044898.5A Active CN109709103B (en) | 2019-01-17 | 2019-01-17 | Orange early-stage rotting identification system and method based on annular stripe polishing imaging |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109709103B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113128249A (en) * | 2019-12-26 | 2021-07-16 | 合肥美亚光电技术股份有限公司 | Garlic scar identification method and system and sorting equipment |
| CN113256575A (en) * | 2021-05-13 | 2021-08-13 | 南京农业大学 | Fruit defect detection method based on structured hyperspectral system |
| WO2021258148A1 (en) * | 2020-06-24 | 2021-12-30 | Rubens IP Pty Ltd | Detecting plant product properties |
| CN115620299A (en) * | 2022-12-14 | 2023-01-17 | 深圳思谋信息科技有限公司 | Image recognition method and device, computer equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102854192A (en) * | 2012-08-22 | 2013-01-02 | 北京农业智能装备技术研究中心 | System and method for detecting apple surface defect |
| GB2498086A (en) * | 2011-12-23 | 2013-07-03 | Maf Agrobotic | Non-destructive detection of defects in fruits and vegetables |
| CN103698009A (en) * | 2013-12-25 | 2014-04-02 | 北京农业智能装备技术研究中心 | Multispectral image acquiring method and system on basis of line scanning hyperspectral imaging |
| CN105427274A (en) * | 2015-10-28 | 2016-03-23 | 北京农业智能装备技术研究中心 | Image detection method for rotting oranges caused by penicillium infection |
| CN106442561A (en) * | 2016-08-29 | 2017-02-22 | 北京农业智能装备技术研究中心 | Online image acquisition system for detecting citrus skin defects and method |
| JP2018059775A (en) * | 2016-10-04 | 2018-04-12 | 三井金属計測機工株式会社 | Fruit and vegetable inspection device |
-
2019
- 2019-01-17 CN CN201910044898.5A patent/CN109709103B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2498086A (en) * | 2011-12-23 | 2013-07-03 | Maf Agrobotic | Non-destructive detection of defects in fruits and vegetables |
| CN102854192A (en) * | 2012-08-22 | 2013-01-02 | 北京农业智能装备技术研究中心 | System and method for detecting apple surface defect |
| CN103698009A (en) * | 2013-12-25 | 2014-04-02 | 北京农业智能装备技术研究中心 | Multispectral image acquiring method and system on basis of line scanning hyperspectral imaging |
| CN105427274A (en) * | 2015-10-28 | 2016-03-23 | 北京农业智能装备技术研究中心 | Image detection method for rotting oranges caused by penicillium infection |
| CN106442561A (en) * | 2016-08-29 | 2017-02-22 | 北京农业智能装备技术研究中心 | Online image acquisition system for detecting citrus skin defects and method |
| JP2018059775A (en) * | 2016-10-04 | 2018-04-12 | 三井金属計測機工株式会社 | Fruit and vegetable inspection device |
Non-Patent Citations (3)
| Title |
|---|
| Y. LU等: "DEVELOPMENT OF A MULTISPECTRAL STRUCTURED ILLUMINATION REFLECTANCE IMAGING (SIRI) SYSTEM AND ITS APPLICATION TO BRUISE DETECTION OF APPLES", 《TRANSACTIONS OF THE ASABE》 * |
| YUZHEN LU等: "Detection of Fresh Bruises in Apples by Structured-Illumination Reflectance Imaging", 《PROC. SPIE》 * |
| YUZHEN LU等: "Fast Bi-dimensional empirical mode decomposition as an image enhancement technique for fruit defect detection", 《COMPUTERS AND ELECTRONICS IN AGRICULTURE》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113128249A (en) * | 2019-12-26 | 2021-07-16 | 合肥美亚光电技术股份有限公司 | Garlic scar identification method and system and sorting equipment |
| CN113128249B (en) * | 2019-12-26 | 2022-08-23 | 合肥美亚光电技术股份有限公司 | Garlic scar identification method and system and sorting equipment |
| WO2021258148A1 (en) * | 2020-06-24 | 2021-12-30 | Rubens IP Pty Ltd | Detecting plant product properties |
| CN113256575A (en) * | 2021-05-13 | 2021-08-13 | 南京农业大学 | Fruit defect detection method based on structured hyperspectral system |
| CN113256575B (en) * | 2021-05-13 | 2023-10-27 | 南京农业大学 | Fruit defect detection method based on structured hyperspectral system |
| CN115620299A (en) * | 2022-12-14 | 2023-01-17 | 深圳思谋信息科技有限公司 | Image recognition method and device, computer equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109709103B (en) | 2020-08-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109709103A (en) | A kind of the citrus early stage rotten fruit identifying system and method for ring-shaped stripe polishing imaging | |
| CN111062880B (en) | Underwater image real-time enhancement method based on condition generation countermeasure network | |
| CN103534583B (en) | The defect inspection method of tire | |
| EP2485639B1 (en) | Method and system for carrying out photoplethysmography | |
| CN108765402B (en) | Non-woven fabric defect detection and classification method | |
| CN101079149A (en) | Noise-possessing movement fuzzy image restoration method based on radial basis nerve network | |
| CN106296670B (en) | A kind of Edge detection of infrared image based on the watershed Retinex--Canny operator | |
| CN108961251A (en) | A kind of raw cotton fault and defects inspecting and recognition methods and system | |
| JP6318755B2 (en) | Filamentous bacteria detection apparatus and filamentous bacteria detection method | |
| CN1874528A (en) | Method and apparatus for improving quality of composite video signal and method and apparatus for decoding composite video signal | |
| CN107295236A (en) | A kind of snapshot Difference Imaging method based on time-of-flight sensor | |
| CN106778640B (en) | The generation method of vegetation cover model under a kind of three-dimensional visualization environment | |
| CN109345524A (en) | A kind of bearing open defect detection system of view-based access control model | |
| Al-Kadi | A gabor filter texture analysis approach for histopathological brain tumor subtype discrimination | |
| CN104268893B (en) | Method for segmenting and denoising lung parenchyma through lateral scanning and four-corner rotary scanning | |
| US8090183B2 (en) | Pattern noise correction for pseudo projections | |
| Huang et al. | An underwater image enhancement method for simultaneous localization and mapping of autonomous underwater vehicle | |
| Ketenci et al. | Design of Gaussian star filter for reduction of periodic noise and quasi-periodic noise in gray level images | |
| CN103679726B (en) | A kind of method improving rock debris image image quality | |
| CN102760295A (en) | Fire disaster image detection system for edge detection-based operator | |
| CN102509271B (en) | Image restoration method based on multi-dimensional decomposition, iteration enhancement and correction | |
| CN102385745B (en) | Pseudo-color matching method of safety inspection data of single-source pseudo-dual-energy detector | |
| CN113433080A (en) | Method for measuring chlorophyll content of rhododendron leaf | |
| KR102558937B1 (en) | Systems and method for vision inspection with multiple types of light | |
| Feng et al. | Underwater salient object detection based on red channel 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 |