CN103514452A - Method and device for detecting shape of fruit - Google Patents
Method and device for detecting shape of fruit Download PDFInfo
- Publication number
- CN103514452A CN103514452A CN201310301746.1A CN201310301746A CN103514452A CN 103514452 A CN103514452 A CN 103514452A CN 201310301746 A CN201310301746 A CN 201310301746A CN 103514452 A CN103514452 A CN 103514452A
- Authority
- CN
- China
- Prior art keywords
- fruit
- shape
- image
- gray level
- roller
- 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
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a method and device for detecting the shape of a fruit. The method comprises the steps of collecting a plurality of color images of the fruit with different poses, after an binary image is obtained, obtaining symmetric indexes and dispersion of the symmetric indexes on the basis of fruit area calculation by the minimum external connection moment of the fruit at four regions separated by two center lines of the fruit, and setting a threshold value by the symmetric indexes and the dispersion of the fruit with the correct shape to conduct comparison and judgment on whether the shape of the fruit is correct or not. The device comprises a chained conveying device, three kilomega network color cameras and two rows of LED light sources, wherein the chained conveying device is arranged at the bottom inside an illumination box, the three kilomega network color cameras are arranged at the upper portion of the illumination box, and the two rows of LED light sources are arranged between the kilomega network color cameras. Rollers are arranged on the chained conveying device, the detected fruit is placed on the rollers, and the LED light sources evenly irradiate the surface of the detected fruit through a scattering plate. According to the method and device, due to the fruit images collected through multiple stations, the influences, caused by the changeable poses, on the judgment on the shape of the fruit are avoided, whether the shape of the moving fruit with the changeable poses is correct or not is rapidly and accurately detected, and the consistency of the judgment of the shape of the fruit with the changeable poses on a production line is ensured.
Description
Technical field
The present invention relates to a kind of detection method and device of surface profile, relate in particular to a kind of fruit shape detection method and device.
Background technology
Fruit shape is one of key character of evaluating fruit quality.Yet because natural form difference is large, the accurate description of fruit shape is very difficult.Shape description method exists and carries out complexity at present, strict to fruit Gesture, detects the problems such as not comprehensive, can not meet the description of fruit shape in the changeable situation of fruit attitude in high speed classification process.
Common shape description method: the method for Fourier descriptor, the harmonic component of the Fourier transform of employing object boundary, as Shape Indexes, according to the value scope of each shape class of harmonic component, or is classified in conjunction with neural network classifier.（Mebatsion,H.K.,Paliwal,J.,Jayas,D.S.A novel,invariant elliptic Fourier coefficient based classification of cereal grains.Biosystems Engineering.2012,111(4):422-428.）
The describing method of various squares.Zernike square method, utilization standard square is normalized fruit image, make the image after normalization there is translation and yardstick unchangeability, then in the image from normalization, extract the Zernike moment characteristics with rotational invariance, method (Ying Yibin fruit shape being classified in conjunction with principal component analysis (PCA) and support vector base, Gui Jiangsheng, Rao Xiuqin. the fruit shape classification based on Zernike square. Jiangsu University's journal (natural science edition) .2007,28 (1): 1-4.).The method of Hu square, Hu(1962) 7 not bending moment expression formula (Hu with translation, rotation, yardstick unchangeability of plane geometric figure have been proposed, m.Visual-pattern recognition by moment invariants.Ire Transactions on Information Theory.1962,8 (2): 179-187.).Adopt Hu square can to the small pudding of fresh-cut carry out shape recognition (Zhang Shuifa, Wang Kaiyi ，Wang Shu peak, Liu Zhongqiang, hair fine jade. the online classification technique of fresh-cut dish based on optimizing square invariant features. Transactions of the Chinese Society of Agricultural Engineering .2011,27 (10): 354-358.).The method of wavelet moment, thus the transforming function transformation function that adopts wavelet function to replace in standard square obtains wavelet moment characteristic quantity.According to the variation judgement fruit shape of wavelet moment feature value.
Typical case's geometric figure shape facility amount describing method.Brewer(2006) the shape features amount description indexes that has proposed tomato as: fruit shape is eccentric, fruit shape circle, oval, heart-shaped, (Brewer, M.T., the Lang such as rectangle, L.X., Fujimura, K., Dujmovic, N., Gray, S., van der Knaap, E.Development of a controlled vocabulary and software application to analyze fruit shape variation in tomato and other plant species.Plant Physiology.2006,141 (1): 15-25.).Kondo(2009) adopt circularity, complexity to detect fruit vertical view (carpopodium/calyx, or the perspective view of top/bottom direction), adopt the centrifugal poor shape class (Kondo that detects fruit side projection figure, N.Robotization in fruit grading system.Sensing and Instrumentation for Food Quality and Safety.2009,3 (1): 81-87.).
In addition the shape description method that also has, Wavelet Descriptor.And other as: the curvature angle description of profile is that shape description refers to calibration method by the centre of form to the radius value between the sample point of profile; Multi-scale level set shape detecting method etc.; And 3D shape describing method.But because the data volume of three-dimensional shape model is larger than the data volume of X-Y scheme, calculated amount also increases a lot thereupon, on-line quick detection is difficulty comparatively.
On high speed fruit grading production line, the attitude of motion fruit constantly changes, and the detection of impact to fruit shape adopts current method cannot guarantee consistance and reliability that fruit shape is detected.
Summary of the invention
In order to realize the SHAPE DETECTION of the motion fruit that attitude is changeable, the object of the invention is to have proposed a kind of fruit shape detection method and device, and verified the reliability of the method and the consistance that fruit shape is detected by embodiment.
The technical solution adopted for the present invention to solve the technical problems is:
One, a kind of fruit shape detection method:
1), for each fruit of same kind fruit, gather the fruit coloured image of the different attitudes of 9-20 width;
2) every width fruit coloured image is carried out to background segment and target extraction, obtain fruit bianry image;
3) solve the external square of minimum of fruit in every width fruit bianry image;
4) two of minimum external square center lines are divided into Si Ge region by the fruit in every width fruit bianry image, solve respectively the fruit area in Si Ge region;
5) solve the symmetry index of fruit in every width fruit bianry image, at least 1 shape of usining is rectified and the intermediate value of the symmetry index dispersion of the same kind fruit that 1 shape is not rectified as the threshold value of this kind fruit, the threshold value of the symmetry index dispersion of any same kind fruit and this kind fruit is compared to judge that whether fruit shape proper.
Every width fruit coloured image carried out to the concrete steps that background segment and target extract be described step 2): extract the R in fruit coloured image, G, B component, by the arithmetical operation between component, obtain the fruit gray level image I (x after merging, y), the definition length of side is less than the rectangular configuration element S of the fruit gray level image length of side, by rectangular configuration element S and fruit gray level image I (x, y) carry out opening operation, the ground unrest of eliminating residual fruit gray level image obtains fruit gray level image I
1(x, y), then define the circular mask w that diameter is less than the fruit gray level image length of side, by circular mask w and fruit gray level image I
1(x, y) carries out convolution algorithm and obtains fruit gray level image I
2(x, y), to fruit gray level image I
2(x, y) adopts large Tianjin method to carry out the processing of automatic threshold two-value, obtains fruit bianry image I
3(x, y).
The concrete steps that solve respectively the fruit area in Si Ge region in described step 4) are: the central point on the minimum external Ju Getiao of mark limit successively, search for respectively on fruit profile to four points that central point is nearest, centered by the rectangle centre of form, according to the fruit area in Heron's formula difference Integration Solving Si Ge region.
The minimum area of the symmetry index Wei Sige region fruit area in described step 5) and the ratio of maximum area.
The ratio of the maximal value that the symmetry index dispersion in described step 5) is symmetry index and the difference of minimum value and the mean value of symmetry index.
Described fruit is spherical fruit.
Described spherical fruit is apple, oranges and tangerines or navel orange.
Two, a kind of fruit shape pick-up unit:
The present invention includes three kilomega network colour TV cameras, lighting box, astigmatism plate, LED light source, tested fruit, chain conveyor, roller, approach switch and computing machine, in lighting box, chain conveyor is arranged on lighting box bottom, three kilomega network colour TV cameras are arranged on lighting box top, two row's LED light sources are arranged on respectively the both sides between chain conveyor and kilomega network colour TV camera, one row roller is installed on chain conveyor, between roller, there is gap, tested fruit is placed on roller, the light of LED light source sees through astigmatism plate uniform irradiation to tested fruit surface, approach switch be arranged on chain conveyor one side and with roller in same level, three kilomega network colour TV cameras are all connected with computing machine, approach switch is connected with three kilomega network colour TV cameras.
When described roller stops approach switch, trigger three kilomega network colour TV cameras and gather image.
Described tested fruit overturns and translation motion by roller.
The useful effect that the present invention has is:
Whether whether the symmetry index dispersion threshold value of the fruit image by several different attitudes of collecting at a plurality of stations is differentiated fruit shape and is rectified, and has overcome attitude and has changed the impact that fruit shape is differentiated, can differentiate exactly the shape of fruit and rectify.Therefore the present invention's changeable motion fruit shape of test pose rapidly and accurately, guarantees that the shape of the fruit that in high-speed production lines, attitude is changeable differentiates the consistance of result.
Accompanying drawing explanation
Fig. 1 is apparatus structure schematic diagram of the present invention.
Fig. 2 is the image after Apple image arithmetical operation in embodiment.
Fig. 3 is the image after Apple image opening operation in embodiment.
Fig. 4 is the image after Apple image linear filtering in embodiment.
Fig. 5 is the image after the processing of Apple image two-value in embodiment.
Fig. 6 is the external square schematic diagram of the minimum of embodiment Apple image.
In figure: 1, kilomega network colour TV camera, 2, lighting box, 3, astigmatism plate, 4, LED light source, 5, tested fruit, 6, chain conveyor, 7, roller, 8, approach switch, 9, computing machine.
Embodiment
Below in conjunction with the drawings and specific embodiments, the present invention is described in further detail.
As shown in Figure 1, a kind of fruit shape pick-up unit of the present invention comprises three kilomega network colour TV cameras 1, lighting box 2, astigmatism plate 3, LED light source 4, tested fruit 5, chain conveyor 6, roller 7, approach switch 8 and computing machine 9, in lighting box 2, chain conveyor 6 is arranged on lighting box 2 bottoms, three kilomega network colour TV cameras 1 are arranged on lighting box 2 tops, two row's LED light sources 4 are arranged on respectively the both sides between chain conveyor 6 and kilomega network colour TV camera 1, one row roller 7 is installed on chain conveyor 6, between roller 7, there is gap, tested fruit 5 is placed on roller 7, the light of two row's LED light sources 4 sees through astigmatism plate 3 uniform irradiations to tested fruit 5 surfaces, approach switch 8 be arranged on chain conveyor 6 one sides and with roller 7 in same level, three kilomega network colour TV cameras 1 are all connected with computing machine 9, approach switch 8 is connected with three kilomega network colour TV cameras 1.Fruit shape is installed on computing machine 9 and is detected software, by fruit shape, detect the processing that software (carried out software registration, registration name is called fruit shape and detects in real time and hierarchy system, the number of applying for registration of 2013R11L121115) carries out fruit shape detection.
When roller 7 stops approach switch 8, trigger three kilomega network colour TV cameras 1 and gather image.
Tested fruit 5 overturns and translation motion by roller 7.
During work, under externally power drives, while tested fruit 5 overturns translation by lighting box 2 under chain conveyor 6 drives, when roller 7 stops approach switch 8, trigger the 12 width images that 3 kilomega network colour TV cameras 1 gather the tested fruit 5 of 4 positions simultaneously, as shown in Figure 1.View data is transferred to computing machine by PCI-Express, for fruit SHAPE DETECTION software, processes.
Described chain conveyor 6 and the people's such as Rao Xiuqin patent of invention " detection method of fruit size and device " (application number: the chain conveyor 200510049489.2) is identical.
The step of a kind of fruit shape detection method of the present invention is as follows:
1), for each fruit of same kind fruit, gather the fruit coloured image of the different attitudes of 9-20 width.
2) every width fruit coloured image is carried out to background segment and target extraction, obtain fruit bianry image.
3) solve the external square of minimum of fruit in every width fruit bianry image.
4) solve with the fruit area in two Si Ge regions that center line was formed of minimum external square in every width fruit bianry image.
5) solve the symmetry index in every width fruit bianry image, at least 1 shape of usining is rectified and the intermediate value of the symmetry index dispersion of the same kind fruit that 1 shape is not rectified as the threshold value of this kind fruit, the threshold value of the symmetry index dispersion of any same kind fruit and this kind fruit is compared to judge that whether fruit shape proper.
The concrete steps of every width fruit coloured image being carried out to background segment and target extraction are: extract the R in fruit coloured image, G, B component, by the arithmetical operation between component, obtain the fruit gray level image I (x after merging, y), the definition length of side is less than the rectangular configuration element S of the fruit gray level image length of side, by rectangular configuration element S and fruit gray level image I (x, y) carry out opening operation, the ground unrest of eliminating residual fruit gray level image obtains fruit gray level image I
1(x, y), then define the circular mask w that diameter is less than the fruit gray level image length of side, by circular mask w and fruit gray level image I
1(x, y) carries out convolution algorithm and obtains fruit gray level image I
2(x, y), to fruit gray level image I
2(x, y) adopts large Tianjin method to carry out the processing of automatic threshold two-value, obtains fruit bianry image I
3(x, y).
As Figure 2-Figure 5, every width fruit coloured image is carried out to background segment and target extraction, obtains in accordance with the following steps fruit bianry image:
1) arithmetical operation: extract red component R (x, the y) image of fruit coloured image, green component G (x, y) image, blue component B (x, y) image.To R, G, B component image carries out arithmetical operation by formula 1, obtains image I (x, y) as Fig. 2.
I(x,y)＝a*R(x,y)+b*G(x,y)+c*B(x,y) （1）
In formula, a is red component coefficient, and b is green component coefficient, and c is blue component coefficient, x=0, and 1,2 ..., M-1, y=0,1,2 ..., N-1, M is image pixel line number, and N is image pixel columns, and x and y represent respectively the row and column of fruit coloured image.
2) opening operation: structure size is the rectangular configuration element S of h * l, and h and l are respectively the height and width of rectangular configuration element S.Press formula 2, rectangular configuration element S carries out opening operation to image I (x, y), obtains image I
1(x, y) is as Fig. 3.
In formula, o represents opening operation,! Represent erosion operation,
represent dilation operation.
3) linear filtering: the circular mask w that structure radius size is m, utilizes the image I that 3 pairs of sizes of formula are M * N
1(x, y) carries out convolution algorithm, obtains image I
2(x, y) is as Fig. 4.
In formula, g=h=(m-1)/2, s=-g ,-g+ 1 ,-g+ 2 ..., g-2, g-1, g, t=-h,-h+1 ,-h+ 2 ..., h-2, h-1, h, x=0,1,2, ..., M-1, y=0,1,2 ..., N-1, s and t represent respectively the row and column of circular mask w, g and h represent respectively the summation value range of row and column.
4) two-value is processed: according to large Tianjin method, automatically select threshold value, by the image I after linear filtering
2(x, y) is processed into bianry image I
3(x, y), as Fig. 5, fruit image value is 1, and background image value is 0.
The concrete grammar that solves the fruit area in Si Ge region is: the central point on the minimum external Ju Getiao of mark limit successively, search for respectively on fruit profile to four points that central point is nearest, centered by the rectangle centre of form, according to the fruit area in Heron's formula difference Integration Solving Si Ge region.
Solve the symmetry index Os in every width fruit bianry image, the minimum area of symmetry index OsWei Sige region fruit area and the ratio of maximum area.As shown in Figure 6, calculate the symmetry index Os of fruit shape: the center point P 1 on the external Ju Getiao of the minimum limit of the tested fruit 5 of mark successively, P2, P3 and P4, on the profile of tested fruit 5, search respectively and center point P 1, P2,4 nearest some Q1 that P3 and P4 are corresponding, Q2, Q3 and Q4, the region that line segment OQ1, OQ2 and arc Q1Q2 are surrounded is denoted as A1, and the region that line segment OQ2, OQ3 and arc Q2Q3 surround is denoted as A2, the region that line segment OQ3, OQ4 and arc Q3Q4 surround is denoted as A3, and the region that line segment OQ4, OQ1 and arc Q4Q1 surround is denoted as A4.To region A1, A2, A3 and A4, adopt Heron's formula to calculate respectively its area S1, S2, S3 and S4.By the minimum value of area S1, S2, S3 and S4, divided by maximal value, obtain the symmetry index Os of tested fruit 5.
Symmetry index dispersion is the maximal value max{Os of symmetry index
iand minimum value min{Os
idifference and the mean value mean{Os of symmetry index
iratio.First adopt formula 4 to calculate the symmetry index dispersion ε of the proper same kind fruit of shape.
At least 1 shape of usining is rectified and the intermediate value of the symmetry index dispersion of the same kind fruit that 1 shape is not rectified as the threshold value of this kind fruit.Then for other any fruit, carry out shape test, repeat the symmetry index dispersion that above-mentioned steps obtains this fruit, the threshold value of the symmetry index dispersion of any fruit of this kind and this kind fruit is compared to judge whether fruit shape is rectified, if ε is in threshold range for this symmetry index dispersion, shape is rectified; Otherwise shape is not rectified.
Fruit is spherical fruit.
Spherical fruit is apple, oranges and tangerines or navel orange.
Embodiments of the invention:
Adopt 13 apples, wherein by 5 of artificial naked eyes judgements, for shape, rectified, 8 is that shape is not rectified, and respectively each apple is gathered the coloured image of the different attitudes of 12 width.
Every width fruit coloured image is carried out to background segment and target extraction, as shown in Figure 2, extract fruit color
Red component R (x, the y) image of color image, green component G (x, y) image, blue component B (x, y) image.To R, G, B component image is pressed formula 1, gets red component coefficient a=1.8, green component coefficient b=-1.1, blue component coefficient c=-1.2, image pixel line number M=384, image pixel columns N=384, carries out arithmetical operation, obtains image I (x, y).
As shown in Figure 3, structure size is the rectangular configuration element S of h * l.Wherein, high h=5, wide l=15, presses formula 2, uses structural element S to carry out opening operation to image I (x, y), obtains image I
1(x, y).
As shown in Figure 4, the circular mask w that structure radius size is m=20, utilizes the image I that 3 pairs of sizes of formula are M * N
1(x, y) carries out convolution algorithm, obtains image I
2(x, y).
As shown in Figure 5, according to large Tianjin method, automatically select threshold value, by the image I after linear filtering
2(x, y) is processed into bianry image I
3(x, y), fruit image value is 1, background image value is 0.
As shown in Figure 6, adopt minimum external square method to obtain the external square of minimum of apple in every width apple bianry image.Then, solve with the apple area in two Si Ge regions that center line was formed of minimum external square in every width apple bianry image, then the symmetry index that calculates 13 apples is as table 1:
The symmetry index of 13 tested apples of table 1
| Numbering | No.1-1 | No.1-2 | No.1-3 | No.1-4 | No.1-5 | No.2-1 | No.2-2 | No.2-3 | No.2-4 | No.2-5 | No.2-6 | No.2-7 | No.2-8 |
| 1 | 0.9393 | 0.8966 | 0.9322 | 0.8651 | 0.9393 | 0.7657 | 0.9765 | 0.9170 | 0.9023 | 0.7848 | 0.7745 | 0.7885 | 0.8980 |
| 2 | 0.9869 | 0.8837 | 0.8561 | 0.8757 | 0.9234 | 0.9140 | 0.8890 | 0.7664 | 0.9376 | 0.9229 | 0.8561 | 0.9207 | 0.8704 |
| 3 | 0.9305 | 0.9056 | 0.9645 | 0.9200 | 0.9055 | 0.9539 | 0.9883 | 0.7332 | 0.9097 | 0.8481 | 0.9418 | 0.8719 | 0.9408 |
| 4 | 0.9252 | 0.8918 | 0.8549 | 0.9185 | 0.9553 | 0.8456 | 0.9146 | 0.8671 | 0.9106 | 0.7951 | 0.9046 | 0.9182 | 0.7897 |
| 5 | 0.9765 | 0.9022 | 0.8859 | 0.8947 | 0.9322 | 0.8198 | 0.9124 | 0.8921 | 0.9136 | 0.8802 | 0.9715 | 0.9642 | 0.9421 |
| 6 | 0.9374 | 0.9124 | 0.8692 | 0.9051 | 0.9305 | 0.8828 | 0.9010 | 0.7992 | 0.9674 | 0.9107 | 0.9314 | 0.8180 | 0.8490 |
| 7 | 0.9770 | 0.9043 | 0.9587 | 0.9284 | 0.9083 | 0.9628 | 0.9343 | 0.7810 | 0.8984 | 0.8956 | 0.8455 | 0.7940 | 0.9034 |
| 8 | 0.8879 | 0.8811 | 0.8763 | 0.9258 | 0.9444 | 0.8819 | 0.8805 | 0.9142 | 0.9259 | 0.8553 | 0.9542 | 0.8261 | 0.7579 |
| 9 | 0.9025 | 0.9572 | 0.9183 | 0.8784 | 0.9294 | 0.8343 | 0.8535 | 0.8306 | 0.9618 | 0.8686 | 0.9105 | 0.8710 | 0.9714 |
| 10 | 0.9274 | 0.8563 | 0.8769 | 0.9341 | 0.9295 | 0.8983 | 0.9276 | 0.8657 | 0.9272 | 0.9707 | 0.9548 | 0.7999 | 0.8362 |
| 11 | 0.9159 | 0.9633 | 0.9751 | 0.9524 | 0.9059 | 0.9666 | 0.9185 | 0.9226 | 0.7906 | 0.9083 | 0.9277 | 0.8908 | 0.8600 |
| 12 | 0.9214 | 0.9134 | 0.9021 | 0.9502 | 0.9701 | 0.8992 | 0.7686 | 0.8485 | 0.7768 | 0.8614 | 0.9752 | 0.8902 | 0.8664 |
The symmetry index dispersion that obtains apple is as table 2:
The symmetry index dispersion of the tested apple of table 2
| Numbering | No.1-1 | No.1-2 | No.1-3 | No.1-4 | No.1-5 | No.2-1 | No.2-2 | No.2-3 | No.2-4 | No.2-5 | No.2-6 | No.2-7 | No.2-8 |
| Os | 0.1058 | 0.1181 | 0.1327 | 0.0957 | 0.0694 | 0.2268 | 0.2426 | 0.2243 | 0.2114 | 0.2124 | 0.2201 | 0.2037 | 0.2443 |
Wherein, the symmetry index dispersion of 5 apples that the shape that artificial vision judges is rectified is in 0.0694～0.1327 scope, the symmetry index dispersion of 8 apples that the shape that artificial vision judges is not rectified is in 0.2037～0.2443 scope, get 0.1327 and 0.2037 intermediate value 0.1682 as threshold value, symmetry index dispersion and this threshold value of then any apple being carried out obtaining after shape test compare to judge whether fruit shape is rectified, be less than 0.1682 rectify for shape, be greater than 0.1682 do not rectify for shape.Finally, the apple of other 200 samples is also calculated to symmetry index dispersion ε with formula 4, find that accuracy rate is more than 80.15%.
Above-mentioned embodiment is used for the present invention that explains, rather than limits the invention, and in the protection domain of spirit of the present invention and claim, any modification and change that the present invention is made, all fall into protection scope of the present invention.
Claims (10)
1. a fruit shape detection method, is characterized in that the step of the method is as follows:
1), for each fruit of same kind fruit, gather the fruit coloured image of the different attitudes of 9-20 width;
2) every width fruit coloured image is carried out to background segment and target extraction, obtain fruit bianry image;
3) solve the external square of minimum of fruit in every width fruit bianry image;
4) two of minimum external square center lines are divided into Si Ge region by the fruit in every width fruit bianry image, solve respectively the fruit area in Si Ge region;
5) solve the symmetry index of fruit in every width fruit bianry image, at least 1 shape of usining is rectified and the intermediate value of the symmetry index dispersion of the same kind fruit that 1 shape is not rectified as the threshold value of this kind fruit, the threshold value of the symmetry index dispersion of any same kind fruit and this kind fruit is compared to judge that whether fruit shape proper.
2. a kind of fruit shape detection method according to claim 1, it is characterized in that, every width fruit coloured image carried out to the concrete steps that background segment and target extract be described step 2): extract the R in fruit coloured image, G, B component, by the arithmetical operation between component, obtain the fruit gray level image after merging, the definition length of side is less than the rectangular configuration element S of the fruit gray level image length of side, by rectangular configuration element S and fruit gray level image, carry out opening operation, the ground unrest of eliminating residual fruit gray level image obtains fruit gray level image, define again the circular mask w that diameter is less than the fruit gray level image length of side, by circular mask w and fruit gray level image, carry out convolution algorithm and obtain fruit gray level image, to fruit gray level image, adopt large Tianjin method to carry out the processing of automatic threshold two-value, obtain fruit bianry image.
3. a kind of fruit shape detection method according to claim 1, it is characterized in that, the concrete steps that solve respectively the fruit area in Si Ge region in described step 4) are: the central point on the minimum external Ju Getiao of mark limit successively, search for respectively on fruit profile to four points that central point is nearest, centered by the rectangle centre of form, according to the fruit area in Heron's formula difference Integration Solving Si Ge region.
4. a kind of fruit shape detection method according to claim 1, is characterized in that: the minimum area of the symmetry index Wei Sige region fruit area in described step 5) and the ratio of maximum area.
5. a kind of fruit shape detection method according to claim 1, is characterized in that: the ratio of the maximal value that the symmetry index dispersion in described step 5) is symmetry index and the difference of minimum value and the mean value of symmetry index.
6. a kind of fruit shape detection method according to claim 1, is characterized in that: described fruit is spherical fruit.
7. a kind of fruit shape detection method according to claim 6, is characterized in that: described spherical fruit is apple, oranges and tangerines or navel orange.
8. for implementing the claims a kind of fruit shape pick-up unit of method described in 1, it is characterized in that: comprise three kilomega network colour TV cameras (1), lighting box (2), astigmatism plate (3), LED light source (4), tested fruit (5), chain conveyor (6), roller (7), approach switch (8) and computing machine (9), in lighting box (2), chain conveyor (6) is arranged on lighting box (2) bottom, three kilomega network colour TV cameras (1) are arranged on lighting box (2) top, two row's LED light sources (4) are arranged on respectively the both sides between chain conveyor (6) and kilomega network colour TV camera (1), one row roller (7) is installed on chain conveyor (6), between (7), there is gap in roller, tested fruit (5) is placed on roller (7), the light of LED light source (4) sees through astigmatism plate (3) uniform irradiation to tested fruit (5) surface, approach switch (8) be arranged on chain conveyor (6) one sides and with roller (7) in same level, three kilomega network colour TV cameras (1) are all connected with computing machine (9), approach switch (8) is connected with three kilomega network colour TV cameras (1).
9. a kind of fruit shape pick-up unit according to claim 8, is characterized in that: when described roller (7) stops approach switch (8), trigger three kilomega network colour TV cameras (1) and gather image.
10. a kind of fruit shape pick-up unit according to claim 8, is characterized in that: described tested fruit (5) overturns and translation motion by roller (7).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310301746.1A CN103514452B (en) | 2013-07-17 | 2013-07-17 | A kind of fruit shape detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310301746.1A CN103514452B (en) | 2013-07-17 | 2013-07-17 | A kind of fruit shape detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103514452A true CN103514452A (en) | 2014-01-15 |
| CN103514452B CN103514452B (en) | 2016-09-28 |
Family
ID=49897145
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310301746.1A Active CN103514452B (en) | 2013-07-17 | 2013-07-17 | A kind of fruit shape detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103514452B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105597911A (en) * | 2016-01-07 | 2016-05-25 | 何天柱 | Fruit sorting method |
| CN105809085A (en) * | 2014-12-29 | 2016-07-27 | 深圳Tcl数字技术有限公司 | Human eye positioning method and device |
| CN106290359A (en) * | 2016-07-22 | 2017-01-04 | 南京农业大学 | A kind of method of the lossless classification of apple crisp slices quality |
| CN107860316A (en) * | 2017-10-30 | 2018-03-30 | 重庆师范大学 | Corn kernel three-dimensional parameter measurement apparatus and its measuring method |
| CN108288388A (en) * | 2018-01-30 | 2018-07-17 | 深圳源广安智能科技有限公司 | A kind of intelligent traffic monitoring system |
| CN108436619A (en) * | 2018-03-21 | 2018-08-24 | 洛阳久德轴承模具技术有限公司 | A kind of roller dimension on-Line Monitor Device for coordinating with roller grinding lathe |
| CN109410209A (en) * | 2018-11-19 | 2019-03-01 | 浙江大学 | A kind of exogenous foreign matter detecting method of nut based on deep learning classification |
| CN113421297A (en) * | 2021-07-02 | 2021-09-21 | 浙江德菲洛智能机械制造有限公司 | Strawberry shape symmetry analysis method |
| CN115046478A (en) * | 2022-08-10 | 2022-09-13 | 深之蓝海洋科技股份有限公司 | Underwater relative pose measuring method and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1394699A (en) * | 2002-08-03 | 2003-02-05 | 浙江大学 | Fruit quality real time detection and grading robot system |
| CN1664502A (en) * | 2005-03-28 | 2005-09-07 | 浙江大学 | Multi dimension energy detection method and apparatus for fruit shape |
-
2013
- 2013-07-17 CN CN201310301746.1A patent/CN103514452B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1394699A (en) * | 2002-08-03 | 2003-02-05 | 浙江大学 | Fruit quality real time detection and grading robot system |
| CN1664502A (en) * | 2005-03-28 | 2005-09-07 | 浙江大学 | Multi dimension energy detection method and apparatus for fruit shape |
Non-Patent Citations (2)
| Title |
|---|
| WANG FUJIE等: "Research on description of fruit shape based on machine vision", 《AMERICAN SOCIETY OF AGRICULTURAL AND BIOLOGICAL ENGINEERS ANNUAL INTERNATIONAL MEETING 2012》 * |
| 王福杰等: "苹果图像的背景分割与目标提取", 《农业机械学报》 * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105809085A (en) * | 2014-12-29 | 2016-07-27 | 深圳Tcl数字技术有限公司 | Human eye positioning method and device |
| CN105809085B (en) * | 2014-12-29 | 2019-07-26 | 深圳Tcl数字技术有限公司 | Human-eye positioning method and device |
| CN105597911A (en) * | 2016-01-07 | 2016-05-25 | 何天柱 | Fruit sorting method |
| CN106290359B (en) * | 2016-07-22 | 2019-01-11 | 南京农业大学 | A kind of method of the lossless classification of apple crisp slices quality |
| CN106290359A (en) * | 2016-07-22 | 2017-01-04 | 南京农业大学 | A kind of method of the lossless classification of apple crisp slices quality |
| CN107860316A (en) * | 2017-10-30 | 2018-03-30 | 重庆师范大学 | Corn kernel three-dimensional parameter measurement apparatus and its measuring method |
| CN108288388A (en) * | 2018-01-30 | 2018-07-17 | 深圳源广安智能科技有限公司 | A kind of intelligent traffic monitoring system |
| CN108436619A (en) * | 2018-03-21 | 2018-08-24 | 洛阳久德轴承模具技术有限公司 | A kind of roller dimension on-Line Monitor Device for coordinating with roller grinding lathe |
| CN109410209A (en) * | 2018-11-19 | 2019-03-01 | 浙江大学 | A kind of exogenous foreign matter detecting method of nut based on deep learning classification |
| CN109410209B (en) * | 2018-11-19 | 2022-04-12 | 浙江大学 | Method for detecting exogenous foreign matters in nuts based on deep learning classification |
| CN113421297A (en) * | 2021-07-02 | 2021-09-21 | 浙江德菲洛智能机械制造有限公司 | Strawberry shape symmetry analysis method |
| CN113421297B (en) * | 2021-07-02 | 2023-06-27 | 浙江德菲洛智能机械制造有限公司 | Shape symmetry analysis method for strawberries |
| CN115046478A (en) * | 2022-08-10 | 2022-09-13 | 深之蓝海洋科技股份有限公司 | Underwater relative pose measuring method and device |
| CN115046478B (en) * | 2022-08-10 | 2022-12-02 | 深之蓝海洋科技股份有限公司 | Underwater relative pose measuring method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103514452B (en) | 2016-09-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103514452A (en) | Method and device for detecting shape of fruit | |
| Jana et al. | Automatic fruit recognition from natural images using color and texture features | |
| Martins et al. | Automatic detection of surface defects on rolled steel using computer vision and artificial neural networks | |
| CN106228612B (en) | The method and apparatus for rebuilding potato three-dimensional surface using System of Rotating about Fixed Axis profile diagram | |
| CN106096660B (en) | Convolutional neural networks based on independent composition analysis algorithm | |
| Alipasandi et al. | Classification of three varieties of peach fruit using artificial neural network assisted with image processing techniques. | |
| CN104597052A (en) | High-speed lossless potato grading detection method and system based on multi-characteristic fusion | |
| Costa et al. | Quantitative evaluation of Tarocco sweet orange fruit shape using optoelectronic elliptic Fourier based analysis | |
| CN105678341B (en) | A kind of woollen and cashmere recognizer based on Gabor wavelet analysis | |
| CN103646255A (en) | Face detection method based on Gabor characteristics and extreme learning machine | |
| Li et al. | Cucumber detection based on texture and color in greenhouse | |
| Cetinic et al. | Automated painter recognition based on image feature extraction | |
| CN105574540A (en) | Method for learning and automatically classifying pest image features based on unsupervised learning technology | |
| Ji et al. | Apple grading method based on features of color and defect | |
| CN106127253A (en) | A kind of method for detecting infrared puniness target utilizing sample characteristics learning classification | |
| Jana et al. | Intra-class recognition of fruits using color and texture features with neural classifiers | |
| CN106872473A (en) | A kind of potato defects detection identifying system design based on machine vision | |
| Zhang et al. | Grasping point detection of randomly placed fruit cluster using adaptive morphology segmentation and principal component classification of multiple features | |
| CN102680488B (en) | Device and method for identifying massive agricultural product on line on basis of PCA (Principal Component Analysis) | |
| Sidehabi et al. | The Development of Machine Vision System for Sorting Passion Fruit using Multi-Class Support Vector Machine. | |
| Lee et al. | Intelligent classification methods of grain kernels using computer vision analysis | |
| CN102332087A (en) | Face recognition method based on sparse representation | |
| Jeong et al. | Development of an automatic sorting system for fresh ginsengs by image processing techniques | |
| Arigbabu et al. | Gender recognition on real world faces based on shape representation and neural network | |
| Goga et al. | Hand gesture recognition using 3D sensors |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210520 Address after: 310012 room 1102, block B, Lishui digital building, 153 Lianchuang street, Wuchang Street, Yuhang District, Hangzhou City, Zhejiang Province Patentee after: Hangzhou nuotian Intelligent Technology Co.,Ltd. Address before: 310027 No.38, Zheda Road, Xihu District, Hangzhou City, Zhejiang Province Patentee before: ZHEJIANG University |
|
| TR01 | Transfer of patent right |