CN102507598A - High-speed unordered capsule defect detecting system - Google Patents
High-speed unordered capsule defect detecting system Download PDFInfo
- Publication number
- CN102507598A CN102507598A CN201110341225XA CN201110341225A CN102507598A CN 102507598 A CN102507598 A CN 102507598A CN 201110341225X A CN201110341225X A CN 201110341225XA CN 201110341225 A CN201110341225 A CN 201110341225A CN 102507598 A CN102507598 A CN 102507598A
- Authority
- CN
- China
- Prior art keywords
- capsule
- unordered
- image
- adopts
- edge
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Image Analysis (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The invention discloses a high-speed unordered capsule defect detecting system, which comprises a capsule transmitting device, an image collecting system, a defect identifying system and DSP (Digital Signal Processor) system transplantation, wherein the capsule transmitting device adopts a mode that capsules are randomly arranged and transmitted and is provided with a shaking device in front of a transmission belt; the image collecting system adopts the high-speed shutter technology of an industrial camera of 90 frames per second to solve the problem of smearing and adopts an infrared filter and a lens used for filtering infrared rays as double lenses to solve the switching problem of different detecting light sources; and the defect identifying system uses a method for setting a dynamic gray threshold to separate a foreground from a background and adopts a method based on the threshold to extract the edges of undivided unordered capsules. The defect detecting system adopts a machine vision technology to detect the shape and appearance of solid capsules and does not need contact the detected articles, so the system belongs to non-contact detection. The defect detecting system solves the problem of secondary pollution of the detected capsules under the general technology. The defect detecting system is an environment-friendly detecting system, and the detecting efficiency and precision and the degree of automation are high.
Description
Technical field
The present invention relates to the medicine detection range, be specifically related to the unordered defective capsule detection system of a kind of high speed.
Background technology
It is the important step in pharmaceutical production field that defective capsule detects, for the safety in production and the use of medicine provides important assurance.Present domestic drug quality profile outward appearance detection means mainly relies on naked eyes identification, and it is the main means of ensuring the quality of products that manual work repeats to select, and the headcount of picking capsule often will account for 1/3 of the whole capsule employee of factory sum.Abroad, the capsule detection system and the equipment of comparative maturity is arranged, like SORTOMAT 8D machines of German Bo Xikesi MASCHINPEX company etc.But the home market do not had directly do not supply, have only the pharmaceutical factory of foreign capitals investment in China can be, but ins and outs hold in close confidence to external world from overseas imported with original packaging equipment.And existing equipment is that the later orderly capsule that sorts is detected mostly, and there are two problems in this orderly checkout equipment: the first, and it is less to detect handling capacity.Because after the capsule ordering, the most of space in the visual field is occupied by the capsule pallet, limited to some extent so inspect an interior capsule number, the orderly capsule about 4 is only arranged in the general single visual field.And unordered capsule is owing to closely link to each other between capsule and capsule, and capsule density will be far above orderly capsule in its visual field.The second, the ordering mechanism of the extra increase capsule of needs has increased cost.
Summary of the invention
The object of the present invention is to provide the unordered defective capsule detection system of a kind of high speed, separate the high speed detection that has realized unordered capsule with the system that detects software and hardware combining, have higher realistic meaning through an effective capsule.
The unordered defective capsule detection system of a kind of high speed of the present invention comprises capsule gearing, IMAQ and control system and defect recognition system, also comprises the dsp system transplanting,
The capsule gearing adopts capsule random alignment transmission manner, settles shaking device in transport tape the place ahead, so that its medicine to be detected is placed steadily on travelling belt, do not occurred is overlapping;
Image capturing system adopts the high-speed shutter technology of industrial camera to solve smear problem, and adopts the twin-lens technology; A camera lens has the infrared ray filter, and another camera lens has the filtration infrared function, through the difference of camera lens itself; Draw two width of cloth overlapping image fully after the computing;, and in the image that another width of cloth visible light is taken, do open defect with this edge that obtains and detect as rim detection with infrared image, to solve the problem that different detection light source are switched;
The defect recognition system comprises that method through dynamic gray threshold is set with prospect and background separation, adopts and extracts unsegregated unordered ordering capsule edge based on the method for threshold value;
Dsp system mainly comprises these three modules of image capture module, FPGA module, DM642 module and memory module.
Preferably, in the said image capturing system, industrial camera carried out short-term precharge to bus before each new signal, be cleaned with all information that guarantee relevant front pixel.The short-term precharge pulse of said industrial camera is used for emulation bus is shorted to ground, and its range of application is from VGA to 10Mpixel, and frame per second is from 500fps to 10,000fps, and data throughput is up to 5.5Gpix/s.
Preferably, adopt variable three coloured light in the said defect recognition system, to realize distinguishing all kinds of Different Rule, the effect of the tablet of different model.
Preferably, background is separated with prospect in the defect recognition system, and concrete grammar is:
At first, on the gray-scale map basis, the threshold value according to histogram setting edge extracting obtains binary map, removes the noise in the image through area and shape filtering again, thereby obtains unsegregated capsule edge;
Then, utilize straight line Hough conversion to separate the edge that the picture element that constitutes fillet is less, demonstrate " point " shape " point " connection capsule; Utilize angle point identification to separate and constitute the edge that the fillet pixel is more, demonstrate " line " shape " line " connection capsule, progressively extract capsule edge single, that separate.
Preferably, the extraction in the said defect recognition system does not separate unordered ordering capsule edge, and concrete grammar is:
Adopt the method for dynamic threshold to come a kind of background colour and two kinds of capsule looks in the differentiate between images earlier;
Remove noise with the open and close computing in the morphology then, and cut off the narrow thin connection that is communicated with in the district;
At last, utilize the area filter method from the capsule prospect, to distinguish imperfect capsule district, single capsule district, many capsules connection district:
Adopt the method for progressively refinement that many capsules connections district is cut apart, extract parallel edges with the Hough conversion earlier, the picture element that constitutes fillet is less, and the capsule that demonstrates " point " shape " point " connection separates; It is more that the method for discerning with angle point again will constitute the fillet pixel, demonstrates the capsule separation that " line " shape " line " connects.
Preferably, said DSP implant system is IMAQ A/D with AD9200, and with TMS320, DM642 is the DSP core processor, and the application network technology is carried out the image transmission, and with FPGA control output and realization image preprocessing function.
Advantage of the present invention is following:
1, lack of alignment capsule high speed detection: the invention solves orderly arrangement capsule and have the handling capacity problem of smaller in detecting; With respect in the single visual field of orderly arrangement capsule the handling capacity about 4 only being arranged, being increased to single visual field can have the capsule about 30.According to the shutter speed of per second 60 times, per minute can detect about about 1500 capsule, has improved 6 times than the speed of about 240 of orderly arrangement capsule per minutes.
2, automated high-precision detects: the each detection as long as capsule is poured in the equipment funnel, open the running test button, and equipment is placed capsule automatically to travelling belt, and through after the field of detection, display screen shows testing result, and adopts two cameras to carry out stereoscopic shooting; One month, more than 60000 time follow-on test on the process production line, rate of accuracy reached to 92%.
3, have self-learning capability: in order to improve the suitable row of equipment, corresponding exploitation has the sample learning module.Before each the detection, equipment is taken through the sampling to the zero defect sample, catches the characteristic of correct sample, adjustment parameter during dynamic operation.
Description of drawings
Fig. 1 is the process flow diagram of the unordered defective capsule detection system of a kind of high speed of the present invention.
Fig. 2 is that unordered ordering capsule edge extracting of the present invention " point " connection is connected synoptic diagram with " line ".
Fig. 3 is that unordered ordering capsule edge extracting cathetus section of the present invention is extracted and the synoptic diagram that is connected.
Fig. 4 is the hardware system synoptic diagram of dsp system of the present invention.
Embodiment
Shown in the process flow diagram of Fig. 1, adopt the flow process of the unordered defective capsule detection system of high speed of the present invention to do, grasp bucket through capsule earlier capsule is put on the streamline; Through gearing capsule is sent to then and gathers original image under the image capturing window; Again the original image that collects is carried out defect recognition; With whether defective signal is passed to device for eliminating, the defective capsule is rejected from travelling belt at last.
1. transmission and shaking device:
Gearing is the necessary link in detecting in capsule detects, through gearing with detected object continuous, in batches flow to detection window.At present main gearing can be divided into orderly arrangement (capsule is transmitted through proper alignment) and random alignment (be positioned over travelling belt on arbitrarily transmit capsule) dual mode; The former plant equipment complicated (needing ordering mechanism); Bad adaptability; Can only transmit the single variety medicine, the efficiency ratio of transmission is lower, but the capsule of arranging in order is easy to the realization of defect recognition software.The latter; Fitness high (capsule or the tablet that can adapt to all kinds of size shape) is arranged; Transmission efficiency high (each detection can reach 30 above capsules or tablet, and sortord has only about 4), but the capsule of random alignment is had higher requirement to defect recognition software.By contrast, we adopt the latter, in order to realize this mode, adopt variable three coloured light, reach to distinguish all kinds of Different Rule, and the tablet of different model lets it notable difference arranged with background colour, thereby has obtained very desirable effect.
Because machine vision can only be discerned over against the defective image capture window, that be in the checking matter surface; And can't discern back to the defective of image capture window or side; So settle shaking device in transport tape the place ahead; It is overlapping that its medicine to be detected is not occurred, and separate between capsule and the capsule as far as possible, finally guarantees that no omission medicine passes through detection zone.
2. image capturing system:
Image capturing system is the important step of Machine Vision Recognition Technology, and the quality quality of images acquired directly influences the complexity of image recognition accuracy and programmed algorithm.Image capturing system mainly comprises camera lens, image pick-up card, illumination condition three parts, takes the smear problem and different detection light source switching problems that is brought when it mainly solves capsule to be detected and moves.
When capsule passed through to take window on travelling belt, can there be smear problem (ghost image on the x direction) in mobile capsule.This is caused by the relatively large RC constant of the wide emulation bus of sheet.For the signal on the bus, owing to the time that the precision of handling 10 bits is used is longer, so the partial information of a last pixel possibly is detained on the current pixel.This ghost image is difficult to proofread and correct in subsequent image processing.
We adopt the high-speed shutter technology of the industrial camera of per second 90 frames to solve smear problem.Industrial camera carried out of short duration precharge to bus before each new signal.All information that so just guaranteed relevant front pixel are washed off.This technical requirement produces the short-term precharge pulse.These pulses are used for emulation bus is shorted to ground.Its range of application is from VGA to 10Mpixel, and frame per second is from 500fps to 10,000fps, and data throughput is up to 5.5Gpix/s.
We find through infrared ray effective especially to the rim detection (being that profile detects) of capsule in the actual experiment process; And it is not enough to some extent to its surface defects detection (being that outward appearance detects) compared with visible light; Because the foundation that the data that rim detection obtains detect as follow-up outward appearance is so can not separately take.We adopt the twin-lens technology for this reason, and first has the infrared ray filter with two mirrors, and another has the filtration infrared function; Difference through camera lens itself; Draw fully overlapping image of two width of cloth after the computing, as rim detection, and in the image that another width of cloth visible light is taken, do open defect with this edge that obtains and detect with infrared image; Thereby solved the problem of two light source detection, guaranteed for follow-up Machine Vision Recognition provides reliable image.
3. defect recognition software:
The defect recognition algorithm is the core of this software section, also is a difficult point of entire equipment.The defect recognition algorithm is with the image that collects, and carries out analyzing and processing, thereby whether is had the conclusion of defective, comprises following step in the algorithm process process.
1) background is separated with prospect
Background and the foundation that prospect is separated are that both gray-scale values are different, thus method that can be through gray threshold is set with prospect and background separation, concrete static threshold and two kinds of methods of dynamic threshold of can being divided into again.
Static threshold obviously is not suitable for this problem; Detect on-the-spot at capsule; Even the environment of IMAQ (comprising illumination, shooting distance, time shutter, streamline speed) is identical; Because the distributing position of random alignment capsule is different, the gray distribution of image that causes collecting is still different, so employed threshold value need dynamically arrange when extracting the edge.Trough in the grey level histogram can be used as the foundation that dynamic gray threshold is provided with.
Owing to comprise a kind of background colour in the image, two kinds of capsule looks, totally three kinds of colors, so three crests and two troughs are arranged in the histogram, a trough can be used to distinguish background and prospect, another can be used to distinguish two kinds of colors of capsule itself.First trough in these two troughs can be set to the threshold value of capsule edge extracting, and the foundation of judgement is the grey level difference at this trough place and the difference of adjacent peaks height.
Concrete grammar is: at first, on the basis of gray-scale map, the threshold value according to histogram setting edge extracting obtains binary map, removes the noise in the image through area and shape filtering again, thereby obtains unsegregated capsule edge.Then, utilize straight line Hough conversion to separate the edge that " point " connects capsule; Utilize angle point identification to separate the edge that " line " connects capsule, progressively extract capsule edge single, that separate.This method can significantly promote the detection speed (to the each quantity that detects of the capsule of orderly arrangement about 10, at every turn can detect 30 to the capsule of random alignment) of capsule.
2) unordered ordering capsule edge extracting
The separation edge extraction is not meant the capsule and the background of the random alignment of piling shape is cut apart, to obtain accurate many capsule connections district that tries one's best.The method extracted of separation edge does not mainly contain following several types: based on threshold value, based on gradient, based on the zone, based on cluster with based on the detection method under some specific environments.This patent adopts the method based on threshold value that the edge is detected.
We adopt the method for dynamic threshold to come a kind of background colour and two kinds of capsule looks in the differentiate between images earlier.Remove noise with the open and close computing in the morphology then, and cut off the narrow thin connection that is communicated with in the district.At last, utilize the area filter method from the capsule prospect, to distinguish imperfect capsule district, single capsule district, many capsules connection district.The method of distinguishing is: if the area in i district is Ai, be limited to A under the area of the single capsule of established standards
L, on be limited to A
H, then can the connection in the image be divided into three types:
(1) noise or be in the imperfect capsule on the image boundary, A
i<a
L
(2) single capsule is communicated with district, A
L<a
i<a
H
(3) many capsules are communicated with district, A
i>A
HAnd n=(A
iMod (A
L+ A
H))/2, n is that many capsules are communicated with capsule number in the district.
3) " point " connects the capsule separation
After accomplishing the extraction of separation edge not; Only realized separating of prospect and background; For the district that is communicated with that also can constitute some many capsules between capsule closely adjacent in the prospect and the capsule; These connection number, angle, link positions that are communicated with the district are all uncertain, even the connection of many capsules multiple spot might occur, and parallel connected situation.Connect closely that degree can be divided into " point " and connect by many capsules and be connected two kinds of situation with " line "." point " connects and to be meant that the picture element that constitutes fillet is less, demonstrate " point " shape, like 1. 4. two places among Fig. 2; " line " connects and to be meant that to constitute the fillet pixel more, demonstrate " line " shape, like 2. 3. two places among Fig. 2.We adopt the method for progressively refinement that many capsules connections district is cut apart, and extract parallel edges with the Hough conversion earlier, with the capsule separation of " point " connection; With the method for angle point identification the capsule that " line " connects is separated again.
The edge feature information of capsule is complete basically under " point " connection situation, and capsular parallel edges, center line, two ends shade are all more complete, is not damaged because of connection.Thus, we are the foundation of the parallel edges in the capsule border to separating as " point " connection.Parallel edges extracts coupling through Hough conversion and straight line characteristic, extracts capsule that the back just can connect to " point " and separate distinguishing from being communicated with, like 1. 4. two places among Fig. 2.
(1) straight-line segment is extracted in the Hough conversion
The Hough conversion is with (x, y) point on the cathetus of plane is mapped to that (ρ, θ) point on the plane is sought peak value through statistical property again.The method has the advantage of strong interference immunity and rotational invariance, result such as Fig. 3 that straight line extracts.
(2) straight-line segment connects
In Fig. 3, " point " connects the parallel edges situation of having interrupted the part capsule, and promptly a long straight-line segment has been cut into two or many short lines sections.Therefore to connect these short lines, to recover original straight line information; Meanwhile, " point " connects separated.After being connected like 5 among Fig. 3,6 and 7,8, two places at capsule edge " point " connects and accomplished separation.
The foundation that straight-line segment can connect is angle and the position relation between the straight-line segment.For two straight line L
1And L
2(like 5 among Fig. 3 and 6), minor increment is L between their end points
n, be L to the maximum
f, straight line L
1Mid point to straight line L
2Distance be d
1, straight line L
2Mid point to straight line L
1Distance be d
2,, then connect two straight lines as satisfying following 3 criterions simultaneously.
Criterion 1:| angle (L
1, L
2) |<t
l(angle), perhaps min (L
1, L
2)/max (L
1, L
2)<0.15, and | angle (L
1, L
2) |<ε
0* t
l, ε wherein
0>1.
Criterion 2: satisfy one of following two conditions: (1) (d
1+ d
2)/2<t
2, (2) | angle (L
1, L
2) |<ε
1* t
l, and (d
1+ d
2)/2<(1+ ε
2) * t
2, wherein 0<ε
1<1,0<ε
2<1.
Criterion 3: satisfy one of following two conditions: (1) L
n<t
3, and L
f>L
1+ L
2, (2) | angle (L
1, L
2) |<ε
3* t
l, (d
1+ d
2)/2<ε
3* t
2, L
n<t
3, and L
f>Max (L
1, L
2), wherein 0<ε
3<1.
Criterion 1 is used to judge the degree of approximation of two straight lines on direction.Criterion 2 is used to judge the position relation of two straight lines.Criterion 3 is used to judge the size in gap between two straight lines.
(1) parallel edges coupling
After straight-line segment connects, need these straight lines be mated in twos, it is right to form parallel edges.Utilize parallel edges to recovering the shade at capsule two ends again, separate thereby connect " point " on the shade.The foundation of parallel edges coupling is that single capsule has two parallel edges, and their length is suitable, between distance be the minor axis length of capsule.Extract in the straight-line segment process in the Hough conversion, straight line is transformed into (ρ, θ) plane.If two straight lines are in that (ρ, θ) coordinate on the plane is respectively (ρ
1, θ
1) and (ρ
2, θ
2), their length is respectively L
1With L
2, the capsule shaft length of standard is d; When then following three conditions satisfy simultaneously, judge two straight line parallels.
Condition 1:| θ
1-θ
2|<ε
4
Condition 2:| ρ
1-ρ
2|-d<ε
5
Condition 3:| L
1-L
2|<ε
6
4) " line " connects the capsule separation
After the separation of " point " connection capsule, remainder is " line " bonding pad.Produce reason that " line " connect and be two or more capsules and be arranged in parallel and closely link to each other, caused the disappearance of parallel edges, so can't use the parallel edges method of identification to extract the capsule that " line " connects at the parallel edges place." line " though connection situation parallel edges lacks, the parallel edges two ends in disappearance have formed the bigger angle point of curvature.Through can recover the parallel edges of disappearance to the angle point Feature Recognition.
(1) Freeman chain code
Article one, discrete curve can be defined as one group of a limited number of 8 connectivity points in the Z territory.Therefore, a digitizing two-value curve can be used Freeman chain code (direction chain code) expression.The Freeman chain code is the possible direction values of 8 kinds of adjacent two pixel lines.The boundary curve of capsule forms n chain code direction after by discretize, and finally this curving chain code can be expressed as { a
i}
n, a wherein
i={ 0,1,2,3,4,5,6,7}, i are the index value of pixel, a
iIt is the direction chain code that points to pixel (i+1) by pixel (i).
(2) angle point identification
Angle point is the point of curvature maximum value on violent point of two dimensional image luminance transformation or the image border curve.These points also are the key characters in the image graphics, and the content of information is very high.The method of Corner Detection can adopt people such as Rosenfeld A and Freeman H to propose the detection method based on boundary curve.
The edge of capsule can adopt based on the chain code difference of boundary curve and estimate curvature after Freeman chain representation.For i point in the chain code curve, the curvature that defines based on the chain code difference is:
The chain yardage that wherein
, n are is the center with i chain code.When carrying out Corner Detection, threshold value Cs is set, calculates the chain yardage n of local field, if | C (i, n) |>Cs, then this is an angle point.In Fig. 4, use this method can detect A, B, C, D, E, F, G, eight angle points of H.4. 1. " point " connect the angle point of two place's annexes along with the separation that " point " connects capsule disappears.
(3) pseudo-angle point is removed
To the angle point among Fig. 4, there are two types of pseudo-angle points, the first kind is near the field the actual angle point a plurality of angle points to be arranged, this type angle point need be merged into an actual angle point, as: D, E and G, H need be merged into two angle points respectively; Second type is the pseudo-angle point that capsule cap causes along bending own, and the pseudo-angle point of this type needs deletion, as: A, B place.
The pseudo-angle point of the first kind:
The size of angle point curvature, having embodied should the puppet angle point and the degree of closeness of actual angle point, and curvature is high more near more from actual angle point, curvature low more then leave far away more.To certain pixel is square field (the half the r of being of the square length of side, the unit picture element point number at center.) in these pseudo-angle points carry out weighted mean (is weight with curvature), try to achieve the coordinate of actual angle point.The coordinate of k pseudo-angle point is respectively (x in the field
1, x
2..., x
k), (y
1, y
2..., y
k), its approximate curvature be (C (and 1, n), C (2, n) ..., C (k, n)), actual corner location (X, computing formula Y) is following:
Second type of pseudo-angle point:
The curvature direction of this type of pseudo-angle point curvature direction and actual angle point is just in time opposite, removes second type of pseudo-angle point so a threshold value e can be set.
After above-mentioned two types of pseudo-angle points were removed, remainder was actual angle point; According to the parallel edges slope that does not lack, actual angle point is matched again, finally restore the disappearance parallel lines, separate thereby connect a capsule to " line ".
4) DSP transplants:
Defects detection equipment is from reliability and the consideration that reduces cost, and defective capsule identification software ultimate demand is moved on dsp chip.So need and to be transplanted on the DSP at the identification software of developing on the PC.The actual needs of, high-speed transfer big, complex calculation and networking to data volume in the machine vision algorithm; Designed with AD9200 is image capture module; With TMS320; DM642 is a DSP core digital signal processor, and application network technology (comprising data memory module, program storage block, serial interface module and ethernet module etc.) is carried out the image transmission, and with FPGA module and output of IO module controls and realization image preprocessing function.Total system mainly comprises three modules, image capture module, FPGA module, DM642 module and memory module (combining shown in Figure 4).
The foregoing description just is to let the one of ordinary skilled in the art can understand content of the present invention and enforcement according to this in order technical conceive of the present invention and characteristics to be described, to be its objective is, can not limit protection scope of the present invention with this.The variation or the modification of every equivalence that the essence of content has been done according to the present invention all should be encompassed in protection scope of the present invention.
Claims (7)
1. the unordered defective capsule detection system of high speed comprises capsule gearing, IMAQ and control system and defect recognition system, it is characterized in that, also comprises the dsp system transplanting,
The capsule gearing adopts capsule random alignment transmission manner, settles shaking device in transport tape the place ahead, so that its medicine to be detected is placed steadily on travelling belt, do not occurred is overlapping;
Image capturing system adopts the high-speed shutter technology of industrial camera to solve smear problem, and adopts the twin-lens technology; A camera lens has the infrared ray filter, and another camera lens has the filtration infrared function, through the difference of camera lens itself; Draw two width of cloth overlapping image fully after the computing;, and in the image that another width of cloth visible light is taken, do open defect with this edge that obtains and detect as rim detection with infrared image, to solve the problem that different detection light source are switched;
The defect recognition system comprises that method through dynamic gray threshold is set with prospect and background separation, adopts and extracts unsegregated unordered ordering capsule edge based on the method for threshold value;
Dsp system mainly comprises these three modules of image capture module, FPGA module, DM642 module and memory module.
2. the unordered defective capsule detection system of high speed according to claim 1; It is characterized in that; In the said image capturing system, industrial camera carried out short-term precharge to bus before each new signal, be cleaned with all information that guarantee relevant front pixel.
3. the unordered defective capsule detection system of high speed according to claim 2; It is characterized in that the short-term precharge pulse of said industrial camera is used for emulation bus is shorted to ground; Its range of application is from VGA to 10Mpixel; Frame per second is from 500fps to 10,000fps, and data throughput reaches 5.5Gpix/s.
4. the unordered defective capsule detection system of high speed according to claim 1 is characterized in that, adopts variable three coloured light in the said defect recognition system, to realize distinguishing all kinds of Different Rule, the effect of the tablet of different model.
5. the unordered defective capsule detection system of high speed according to claim 1 is characterized in that background is separated with prospect in the defect recognition system, and concrete grammar is:
At first, on the gray-scale map basis, the threshold value according to histogram setting edge extracting obtains binary map, removes the noise in the image through area and shape filtering again, thereby obtains unsegregated capsule edge;
Then, utilize straight line Hough conversion to separate the edge that the picture element that constitutes fillet is less, demonstrate " point " shape " point " connection capsule; Utilize angle point identification to separate and constitute the edge that the fillet pixel is more, demonstrate " line " shape " line " connection capsule, progressively extract capsule edge single, that separate.
6. the unordered defective capsule detection system of high speed according to claim 1 is characterized in that, the extraction in the said defect recognition system does not separate unordered ordering capsule edge, and concrete grammar is:
Adopt the method for dynamic threshold to come a kind of background colour and two kinds of capsule looks in the differentiate between images earlier;
Remove noise with the open and close computing in the morphology then, and cut off the narrow thin connection that is communicated with in the district;
At last, utilize the area filter method from the capsule prospect, to distinguish imperfect capsule district, single capsule district, many capsules connection district:
Adopt the method for progressively refinement that many capsules connections district is cut apart, extract parallel edges with the Hough conversion earlier, the picture element that constitutes fillet is less, and the capsule that demonstrates " point " shape " point " connection separates; It is more that the method for discerning with angle point again will constitute the fillet pixel, demonstrates the capsule separation that " line " shape " line " connects.
7. the unordered defective capsule detection system of high speed according to claim 1; It is characterized in that said DSP implant system is IMAQ A/D with AD9200; With TMS320; DM642 is the DSP core processor, and the application network technology is carried out the image transmission, and with FPGA control output and realization image preprocessing function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110341225XA CN102507598A (en) | 2011-11-02 | 2011-11-02 | High-speed unordered capsule defect detecting system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110341225XA CN102507598A (en) | 2011-11-02 | 2011-11-02 | High-speed unordered capsule defect detecting system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102507598A true CN102507598A (en) | 2012-06-20 |
Family
ID=46219705
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201110341225XA Pending CN102507598A (en) | 2011-11-02 | 2011-11-02 | High-speed unordered capsule defect detecting system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102507598A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102879404A (en) * | 2012-10-07 | 2013-01-16 | 复旦大学 | System for automatically detecting medical capsule defects in industrial structure scene |
| CN103218820A (en) * | 2013-04-22 | 2013-07-24 | 苏州科技学院 | Camera calibration error compensation method based on multi-dimensional characteristics |
| CN105092589A (en) * | 2015-07-07 | 2015-11-25 | 东北大学 | Detection method for defects of capsule head |
| CN105136817A (en) * | 2015-08-28 | 2015-12-09 | 赵玉洁 | Capsule visible light defect recognition method based on wireless network |
| CN105139384A (en) * | 2015-08-11 | 2015-12-09 | 北京天诚盛业科技有限公司 | Defective capsule detection method and apparatus |
| CN105160671A (en) * | 2015-08-28 | 2015-12-16 | 赵玉洁 | Capsule visible light defect identification apparatus based on wireless network |
| CN105784711A (en) * | 2014-12-25 | 2016-07-20 | 锐视（厦门）科技有限公司 | Cod liver oil soft capsule appearance defect visual detection method and device |
| CN108387586A (en) * | 2018-02-06 | 2018-08-10 | 深圳市华星光电半导体显示技术有限公司 | Break bar detection device and break bar detection method |
| CN108805854A (en) * | 2018-01-09 | 2018-11-13 | 湖南科技大学 | Tablet quick counter and integrality detection method under a kind of complex environment |
| CN110135256A (en) * | 2019-04-12 | 2019-08-16 | 红云红河烟草(集团)有限责任公司 | Horizontal cigarette judgment method and equipment |
| CN111398308A (en) * | 2020-03-27 | 2020-07-10 | 上海健康医学院 | Automatic detection method and system for packaging quality of aluminum-plastic bubble caps of tablets and capsules |
| CN112712538A (en) * | 2020-12-29 | 2021-04-27 | 合肥联宝信息技术有限公司 | Display screen positioning method, electronic equipment and storage medium |
| CN113129288A (en) * | 2021-04-22 | 2021-07-16 | 安徽大学 | Tablet surface defect detection method based on deep learning image semantic segmentation and automatic processing device thereof |
-
2011
- 2011-11-02 CN CN201110341225XA patent/CN102507598A/en active Pending
Non-Patent Citations (2)
| Title |
|---|
| 吴宏杰 等: "一种提取随机排列胶囊分离边缘的方法", 《微计算机信息》 * |
| 吴宏杰 等: "一种胶囊缺陷检测系统的设计与研究", 《电脑知识与技术》 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102879404B (en) * | 2012-10-07 | 2015-06-17 | 复旦大学 | System for automatically detecting medical capsule defects in industrial structure scene |
| CN102879404A (en) * | 2012-10-07 | 2013-01-16 | 复旦大学 | System for automatically detecting medical capsule defects in industrial structure scene |
| CN103218820A (en) * | 2013-04-22 | 2013-07-24 | 苏州科技学院 | Camera calibration error compensation method based on multi-dimensional characteristics |
| CN103218820B (en) * | 2013-04-22 | 2016-02-10 | 苏州科技学院 | A kind of camera calibration error compensating method based on multidimensional characteristic |
| CN105784711A (en) * | 2014-12-25 | 2016-07-20 | 锐视（厦门）科技有限公司 | Cod liver oil soft capsule appearance defect visual detection method and device |
| CN105784711B (en) * | 2014-12-25 | 2019-06-18 | 厦门威芯泰科技有限公司 | A kind of cod-liver oil soft capsule open defect visible detection method and its device |
| CN105092589A (en) * | 2015-07-07 | 2015-11-25 | 东北大学 | Detection method for defects of capsule head |
| CN105092589B (en) * | 2015-07-07 | 2018-08-03 | 东北大学 | A kind of capsule head defect inspection method |
| CN105139384A (en) * | 2015-08-11 | 2015-12-09 | 北京天诚盛业科技有限公司 | Defective capsule detection method and apparatus |
| CN105139384B (en) * | 2015-08-11 | 2017-12-26 | 北京天诚盛业科技有限公司 | The method and apparatus of defect capsule detection |
| CN105160671A (en) * | 2015-08-28 | 2015-12-16 | 赵玉洁 | Capsule visible light defect identification apparatus based on wireless network |
| CN105136817A (en) * | 2015-08-28 | 2015-12-09 | 赵玉洁 | Capsule visible light defect recognition method based on wireless network |
| CN108805854A (en) * | 2018-01-09 | 2018-11-13 | 湖南科技大学 | Tablet quick counter and integrality detection method under a kind of complex environment |
| CN108805854B (en) * | 2018-01-09 | 2022-02-08 | 湖南科技大学 | Method for rapidly counting tablets and detecting completeness of tablets in complex environment |
| CN108387586A (en) * | 2018-02-06 | 2018-08-10 | 深圳市华星光电半导体显示技术有限公司 | Break bar detection device and break bar detection method |
| CN110135256A (en) * | 2019-04-12 | 2019-08-16 | 红云红河烟草(集团)有限责任公司 | Horizontal cigarette judgment method and equipment |
| CN111398308A (en) * | 2020-03-27 | 2020-07-10 | 上海健康医学院 | Automatic detection method and system for packaging quality of aluminum-plastic bubble caps of tablets and capsules |
| CN111398308B (en) * | 2020-03-27 | 2023-01-17 | 上海健康医学院 | Automatic detection method and system for packaging quality of aluminum-plastic bubble caps of tablets and capsules |
| CN112712538A (en) * | 2020-12-29 | 2021-04-27 | 合肥联宝信息技术有限公司 | Display screen positioning method, electronic equipment and storage medium |
| CN112712538B (en) * | 2020-12-29 | 2022-02-08 | 合肥联宝信息技术有限公司 | Display screen positioning method, electronic equipment and storage medium |
| CN113129288A (en) * | 2021-04-22 | 2021-07-16 | 安徽大学 | Tablet surface defect detection method based on deep learning image semantic segmentation and automatic processing device thereof |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102507598A (en) | High-speed unordered capsule defect detecting system | |
| CN105181714B (en) | A kind of bottling liquid Impurity Detection System and its detection method based on image procossing | |
| CN104504388B (en) | A kind of pavement crack identification and feature extraction algorithm and system | |
| CN105044114A (en) | Electrolytic capacitor appearance package defect image detection system and electrolytic capacitor appearance package defect image detection method | |
| CN101907453B (en) | Online measurement method and device of dimensions of massive agricultural products based on machine vision | |
| CN104297255B (en) | A kind of dixie cup defective vision detection method and system and device | |
| CN202010662U (en) | Real-time inspection and grading system device for fruit appearance quality | |
| CN102020036A (en) | Visual detection method for transparent paper defect of outer package of strip cigarette | |
| CN107084992A (en) | A kind of capsule detection method and system based on machine vision | |
| CN109724984A (en) | A kind of defects detection identification device and method based on deep learning algorithm | |
| CN204789357U (en) | Electrolytic capacitor outward appearance packing defect image detection system | |
| CN105403147A (en) | Embedded bottle pre-form detection system and detection method | |
| CN111709935A (en) | Real-time coal gangue positioning and identifying method for ground moving belt | |
| CN111047655A (en) | High-definition camera cloth defect detection method based on convolutional neural network | |
| CN201707305U (en) | Inspection system for inspecting quality of container caps | |
| CN102034098A (en) | Automatic bolt sorting system and automatic bolt sorting method | |
| CN206132657U (en) | Gilt quality intelligent detecting system based on machine vision | |
| CN102879404B (en) | System for automatically detecting medical capsule defects in industrial structure scene | |
| CN107784469A (en) | A kind of articles from the storeroom checking method based on unmanned plane image technique | |
| CN112131982A (en) | Tea tree tender shoot identification method based on convolutional neural network | |
| CN102998316A (en) | Transparent liquid impurity detection system and detection method thereof | |
| WO2021135675A1 (en) | Automatic bar counting method and apparatus | |
| CN106248680A (en) | A kind of engine commutator quality detecting system based on machine vision and detection method | |
| CN104048966A (en) | Big-law-based cloth cover defect detection and classification method | |
| CN109509173A (en) | A kind of method of counting and counting device of bacterial clump |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20120620 |