CN102192713A - Appearance checking device - Google Patents
Appearance checking device Download PDFInfo
- Publication number
- CN102192713A CN102192713A CN2011100350603A CN201110035060A CN102192713A CN 102192713 A CN102192713 A CN 102192713A CN 2011100350603 A CN2011100350603 A CN 2011100350603A CN 201110035060 A CN201110035060 A CN 201110035060A CN 102192713 A CN102192713 A CN 102192713A
- Authority
- CN
- China
- Prior art keywords
- light
- calculated
- shape data
- dimensional shape
- solar cell
- 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
Abstract
The present invention provides an appearance checking device which accurately and quickly calculates out a three-dimensional shape of a solar electric chip. A first shape calculation part and a second shape calculation part (24,34) calculate out the three-dimensional data of the surface of the chip (50) according to the image data of a plurality of light cutting line images which are shot with a preset frame frequency through a photographing part (20). Hereon, the first shape calculation part (24) performs searching for a prior frame in a period when the cameras (21-23) perform photographing for the light cutting line image of the current frame, and simultaneously performs gravity center calculation processing for the light cutting line image of a previous second frame.
Description
Technical field
The present invention relates to a kind of appearance inspection device of checking the outward appearance of solar cell wafer.
Background technology
In recent years, solar electrical energy generation receives publicity as regenerative resource, and the market scale of solar cell enlarges rapidly.Raw-material solar cell wafer as solar cell can form the groove that is called the muscle shape that cuts trace (saw mark) but be known in when cutting by with fret-saw silicon ingot cutting being obtained.And, be formed with a plurality of cutting traces with length direction towards the mode of certain orientation roughly on the surface of solar cell wafer.
So, there is the deep cutting trace of a plurality of grooves and shaggy solar cell wafer becomes unacceptable product, can't go on the market as commodity.Therefore whether, need to check the surfaceness of solar cell wafer, carrying out the solar cell wafer is the visual examination of unacceptable product.
In the past, the visual examination of solar cell wafer is carried out visual examination by the people always, but because expansion of solar cell market scale in recent years and the various reasons such as artificial wafer breakage when reducing the solar cell wafer inspection, research attempts utilizing special-purpose testing fixture to carry out visual examination.
As such testing fixture is known the cross sectional shape that for example uses the laser extensometer to measure a line correspondence of wafer surface arranged.
In addition, in patent documentation 1, disclose following technology: take the two sides of many crystal semiconductors wafer and the view data of the image of each face of photographing is according to pixels compared two opposed positions of face, when comparative result satisfies predefined stained condition, judge that some in two faces exists stained.
In patent documentation 2, the defective that is used as the wire that glass substrate was comprised of slim panel with inspection is a purpose, following technology is disclosed: will be divided into strip as the view data of the body that is taken of checking object, by split image i.e. first data of deep or light one-dimensional data of having calculated projection, according to these first data, determine to comprise defective candidate's scope of line defect candidate by split image, use the view data corresponding to calculate second data of having emphasized defective candidate's scope, according to these second data being accumulated the defective that the data that obtain are determined wire with the image of the defective candidate's scope that should determine.
In patent documentation 3, muscle shape defective with detection LCD panel is a purpose, select as the concerned pixel of checking object, with the concerned pixel is the zone that given size is selected at the center, in the zone of selecting, extract specified quantity according to the size order of brightness value, the pixel of the brightness value that pre-set threshold is above, obtain each pixel of extraction and the accumulation minimum value of the distance of the angle reference line that passes through concerned pixel, will according to the ballot value of the accumulation minimum value of obtaining in the voting space corresponding with pixel to voting with the corresponding part of concerned pixel, detect muscle shape defective according to voting results.
Patent documentation 1: TOHKEMY 2007-67102 communique
Patent documentation 2: TOHKEMY 2008-134196 communique
Patent documentation 3: TOHKEMY 2005-3452960 communique
Yet, in the method for above-mentioned use laser extensometer, only detect the cross sectional shape of a line correspondence on the wafer, also exist on the line of mensuration concavo-convex unshowy but at the concavo-convex tangible wafer of other parts, at this moment, the possibility of the wafer of unacceptable product listing improves.
In addition, in patent documentation 1, compare by view data and to judge stained having or not, but do not obtain the three-dimensional shape data on the surface of many crystal semiconductors wafer the image of each face of many crystal semiconductors wafer of taking.
In addition, in patent documentation 2, though from determining the defective of wire by taking the view data that obtains as the body that is taken of checking object, this technology is also same with patent documentation 1, does not obtain the three-dimensional shape data on the surface of the body that is taken.
In addition, in patent documentation 3, carry out Flame Image Process by shooting LCD panel and to the view data that obtains, thereby detect muscle shape defective, still, this technology is also same with patent documentation 1,2, does not obtain the three-dimensional shape data on the surface of the body that is taken.
Therefore, in the technology of patent documentation 1～3, can't measure the 3D shape of the cutting trace that is formed on the solar cell wafer.
Summary of the invention
The object of the present invention is to provide a kind of can be correctly and judge the appearance inspection device of the 3D shape of solar cell wafer at high speed.
(1) appearance inspection device of an aspect of of the present present invention is checked the outward appearance of solar cell wafer, wherein, delivery section, it carries described solar cell wafer along the length direction of the cutting trace that forms with certain speed on described solar cell wafer; Irradiation portion, it cuts off line to solar cell wafer edge of being carried by described delivery section and the direction irradiates light that throughput direction intersects; Photographic unit, its with some cycles continuously pickup light cut off line image, this light cuts off line image and comprises that the light by the irradiation of described irradiation portion cuts off line; Shape is calculated mechanism, it cuts off the view data of line image according to several light that photograph by described photographic unit, calculate the three-dimensional shape data of described solar cell wafer, described shape calculate continuous photographing process that mechanism and the described light that is undertaken by described photographic unit cuts off line image carry out concurrently described solar cell wafer three-dimensional shape data calculate processing.
According to this structure, the solar cell wafer of being carried with certain speed by delivery section cuts off line by the illuminated light of direction that edge and throughput direction intersect, thereby cutting off line, and take continuously as light cut-out line image with some cycles by photographic unit with the mode irradiates light that intersects of cutting trace.Therefore, photographic unit can cut off moment that the light that shines when once light cut off the shooting of line image before line shone cuts off the immediate vicinity of line at light and takes next light and cut off line image.Therefore, can calculate the three-dimensional shape data of solar cell wafer with high precision.
In addition since the light that is undertaken by photographic unit cut off the continuous photographing process of line image and calculate by shape the solar cell wafer that mechanism carries out three-dimensional shape data calculate the processing executed in parallel, therefore can obtain the 3D shape of solar cell wafer at a high speed.
(2) preferred: cut off at the light that is carried out present frame by described photographic unit during the shooting of line image, described shape is calculated mechanism and the frame that photographed before described present frame is carried out is used to calculate the processing of cutting off the altitude information of the corresponding described solar cell wafer of the irradiation position of line with described light.
According to this structure, the light of the present frame that is undertaken by photographic unit cuts off the photographing process of line image and is used for the frame that photographed before present frame is calculated with the processing of altitude information that light cuts off the solar cell wafer of the corresponding row of the irradiation position of line and can be carried out simultaneously.Therefore, can obtain the 3D shape of solar cell at a high speed.
(3) preferred: described shape is calculated mechanism and is possessed first shape and calculate portion, this first shape portion of calculating carries out the search processing and center of gravity is calculated processing, it is to cut off in the line image at each light that photographs by described photographic unit that described search is handled, set many lines abreast with described throughput direction, and the high-high brightness pixel of searching for each line, described center of gravity is calculated and handled is basis is handled each line that searches by described search high-high brightness pixel, calculate the barycentric coordinates of the high-high brightness of each line with sub-pix unit, described first shape is calculated portion and is cut off during the shooting of line image at the light that is carried out present frame by described photographic unit, cut off line image for the light of the former frame of described present frame and carry out described search and handle, cut off line image for light simultaneously and carry out described center of gravity and calculate processing from several forward second frames of described present frame.
According to this structure, during the light of the present frame that is undertaken by photographic unit cuts off the shooting of line image, carry out search processing and center of gravity simultaneously and calculate processing, cut off line image for every shooting one width of cloth light, can access the centre of gravity place of the high-high brightness of each bar line.That is, can obtain the centre of gravity place of the high-high brightness of each bar line in the mode of streamline (pipeline).
Therefore, the centre of gravity place of the high-high brightness of each bar line can be in the shooting of solar cell wafer, obtained in real time, the 3D shape of solar cell wafer can be obtained at high speed.
(4) preferred: described shape is calculated mechanism and is possessed second shape and calculate portion, this second shape is calculated portion and is calculated the three-dimensional shape data of described solar cell wafer by carrying out following processing repeatedly, this is treated to: according to the barycentric coordinates of each line of being calculated by the described first shape portion of calculating and the elevation angle of described photographic unit and irradiation portion, calculate the altitude information of the described solar cell wafer of row corresponding with the irradiation position of described light cut-out line.
According to this structure, according to calculate the barycentric coordinates of handling each the bar line calculate and the elevation angle of photographic unit and irradiation means by center of gravity, calculate the altitude information of the solar cell wafer of row corresponding with the irradiation position of width of cloth light cut-out line image, carry out above-mentioned processing by each light is cut off line image, calculate the three-dimensional shape data of solar cell wafer.Therefore, can obtain whole 3D shape of solar cell wafer accurately.
(5) preferred: described photographic unit is taken from the top to described solar cell wafer, and described irradiation portion is from the described solar cell wafer of oblique irradiation.
According to this structure, irradiation portion is from the whole zone of the Width of oblique irradiation solar cell wafer, and photographic unit is taken in the whole zone to the Width of solar cell wafer from the top, therefore can obtain whole three-dimensional shape data of solar cell wafer in the resolution that improves altitude information.In addition, because photographic unit is taken from the top to the solar cell wafer, therefore photographic unit is provided with easily.
(6) preferred: as to take for described solar cell wafer is divided into a plurality of subregions, described photographic unit has a plurality of, described shape is calculated the three-dimensional shape data that mechanism calculates the each several part zone respectively, for the zone of repeating with adjacent subregion in the subregion, overlap with the three-dimensional shape data of another subregion by the three-dimensional shape data that makes a subregion, thereby calculate the three-dimensional shape data of described solar cell wafer.
According to this structure, by a plurality of photographic units the solar cell wafer is divided into a plurality of subregions and takes, and calculate the three-dimensional shape data in each several part zone respectively.And the zone for the part subregion and adjacent repeats overlaps the three-dimensional shape data that obtains whole of solar cell wafer by the three-dimensional shape data that makes a subregion with the three-dimensional shape data of another subregion.Therefore, compare, can access high-resolution light and cut off line image with the situation of the solar cell wafer being taken by a photographic unit.Consequently, can calculate high-resolution three-dimensional shape data.
(7) preferred: described shape is calculated mechanism and is used least square method that the three-dimensional shape data in each several part zone is carried out the plane revisal, for the zone of described repetition, with the three-dimensional shape data weighting summation after the plane revisal of the three-dimensional shape data after the plane revisal of a subregion and another subregion.
According to this structure, the use least square method is carried out the plane revisal to the three-dimensional shape data in each several part zone.Thus, can obtain the three-dimensional shape data in each several part zone, and reduce error because of the concavo-convex three-dimensional shape data that causes of the sensitive surface of each photographic unit.
In addition, zone for subregion in the subregion and adjacent repetition, therefore 3D shape zone after the plane revisal of the three-dimensional shape data after the plane revisal of a subregion and another subregion is weighted addition, can successfully the subregion be bonded to each other and calculates the three-dimensional shape data of whole of solar cell wafer.
(8) preferred: described appearance inspection device also possesses evaluation of estimate and calculates mechanism, and this evaluation of estimate is calculated mechanism by calculate the evaluation of estimate that three-dimensional shape data that mechanism calculates is calculated the roughness of the described solar cell wafer of expression by described shape; Decision mechanism, it judges according to described evaluation of estimate whether described solar cell wafer is qualified.
According to this structure, calculate the evaluation of estimate of the roughness of expression solar cell wafer from whole three-dimensional shape data of high-resolution solar cell wafer, and utilize this evaluation of estimate to judge whether the solar cell wafer is qualified, therefore can judge correctly whether the solar cell wafer is unacceptable product.
(9) preferred: described evaluation of estimate is calculated mechanism and is carried out the processing of removing low-frequency component from described three-dimensional shape data, and calculates described evaluation of estimate by the three-dimensional shape data of removing behind the low-frequency component.
According to this structure, on the basis of the roughness of estimating the solar cell wafer, from three-dimensional shape data, remove unwanted low-frequency component, therefore can access the evaluation of estimate of the roughness that correctly reflects the solar cell wafer, can judge correctly whether the solar cell wafer is unacceptable product.
(10) preferred: described evaluation of estimate is calculated mechanism the three-dimensional shape data of each row is used Gaussian filter and extracts low-frequency component, deducts low-frequency component of each row of extraction from former each row three-dimensional shape data, thereby removes low-frequency component.
According to this structure, with the unit of classifying as, use Gaussian filter from three-dimensional shape data, to extract low-frequency component, and deduct the processing of the low-frequency component of extraction from original three-dimensional shape data, thereby from three-dimensional shape data, remove low-frequency component, therefore can remove low-frequency component accurately.
(11) preferred: if described evaluation of estimate is judged to be unacceptable product greater than the then described decision mechanism of threshold value, if be judged to be certified products less than the then described decision mechanism of threshold value.
According to this structure, by evaluation of estimate and threshold value are compared, can be the certified products or the judgement of unacceptable product, therefore can promptly carry out the certified products determination processing.
The invention effect
According to the present invention, can be correctly and measure whole 3D shape of solar cell wafer at high speed.
Description of drawings
Fig. 1 is the one-piece construction figure that the appearance inspection device of embodiments of the present invention is shown.
Fig. 2 is the block diagram of the electric structure of appearance inspection device shown in Figure 1.
Fig. 3 is the process flow diagram of master routine (main routine) that the appearance inspection device of embodiments of the present invention is shown.
Fig. 4 is the process flow diagram that the detailed process of search processing is shown.
Fig. 5 illustrates the process flow diagram that center of gravity is calculated the detailed process of processing.
Fig. 6 is the sequential chart of processing that shows the process flow diagram of Fig. 4 by the time sequence.
Fig. 7 is the figure of the flow process of schematically illustrated search processing and the center of gravity processing of calculating processing.
Fig. 8 illustrates the figure that light cuts off an example of line image.
Fig. 9 is the chart that the distribution of the brightness value when being the center with the high-high brightness pixel on the i line is shown.
Figure 10 is the key diagram of calculating the processing of altitude information.
Figure 11 is the figure of the state that is provided with of schematically illustrated light source and camera.
Figure 12 is the chart that the cross sectional shape data of being calculated by the shape portion of calculating are shown.
Figure 13 is the figure of three-dimensional shape data of certain subregion of simulation or schematically illustrated wafer of being calculated by the shape portion of calculating.
Figure 14 illustrates whole three-dimensional shape data to generate the process flow diagram of handling.
Figure 15 is the figure that explanation is weighted the processing of addition.
Figure 16 illustrates to estimate the process flow diagram of handling.
Figure 17 illustrates the cross sectional shape data of removing behind the low-frequency component.
Figure 18 is simulation or the schematically illustrated figure that removes the three-dimensional shape data behind the low-frequency component.
The one-piece construction figure of the appearance inspection device when Figure 19 illustrates light source and camera and is.
Symbol description
10 irradiation portions
20 shoot parts
21～23 cameras
24 first shapes are calculated portion's (shape is calculated mechanism)
30 control parts
31 carry control part
32 irradiation control parts
33 shooting control part
34 second shapes are calculated portion's (shape is calculated mechanism)
35 evaluations of estimate are calculated portion's (evaluation of estimate is calculated mechanism)
36 detection units (decision mechanism)
40 delivery section
50 wafers
CL, CL1, CL2, CL3 light cut off line
Dk, Dk+1 subregion
Embodiment
Below, the appearance inspection device of one embodiment of the present invention is described.Fig. 1 illustrates the one-piece construction figure of the appearance inspection device of embodiments of the present invention.As shown in Figure 1, this appearance inspection device possesses irradiation portion 10, shoot part 20, control part 30 and delivery section 40.In Fig. 1, the Y direction indication utilizes delivery section 40 to carry the throughput direction of solar cell wafer 50.In addition, directions X is represented and Y direction quadrature and the direction parallel with surface level.Z direction indication and directions X and Y direction be the short transverse of quadrature respectively.Below, solar cell wafer 50 notes are made wafer 50.
In addition, the intersection of light source 11～13 light that shines respectively and the wafer of carrying by delivery section 40 50 is that light cuts off line CL1, CL2, CL3.In the present embodiment, light cuts off the position Y1 that coordinate that line CL1, CL2, CL3 for example form the Y direction is positioned at regulation, length direction and directions X almost parallel.In addition, light cuts off the coordinate that line CL2 for example is set at the Y direction and is positioned at the position Y2 by upstream side than position Y1, length direction and directions X almost parallel.That is, the whole area illumination light of the Width (directions X) of wafer 50 is cut off line, therefore can access whole three-dimensional shape data of wafer 50 by light cut-out line CL1, CL2, CL3.
The setting that such light cuts off line CL1～CL3 can realize easily by the direction of light that position or ejaculation are set of regulating light source 11～13.Need to prove that light cut-out line CL1, CL2, CL3 are that light cuts off line CL not having to record and narrate under other situation of special section.
In addition, light source 11～13 is installed in the lower surface of base station 14 respectively via the support material 15 of L word shape roughly, with from oblique irradiate wafer 50.
Shoot part 20 for example is made of three cameras 21～23.Camera 21～23 for example disposes along directions X by this order, and is configured to take wafer 50 from upside.At this, camera 21～23 is installed in the lower surface of base station 25 respectively, so that sensitive surface is the same position of z direction.The x direction of camera 21～23 and the camera angle of y direction are identical, and the position of the z direction of sensitive surface is also identical, and therefore vertical the and horizontal width of the subregion of conduct shooting area separately is identical.In addition, the image description of camera 21～23 being taken respectively that comprises light cut-out line is that light cuts off line image.In addition, the sensitive surface of camera 21～23 is a rectangle, and the limit of a side of sensitive surface is parallel with the Y direction, and the limit of opposite side is parallel with directions X.
At this, camera 21～23 is by can constituting with the CMOS camera of frame frequency (for example 250fps) photographic images of regulation, and the view data that the light that photographs is cut off the simulation of line image is converted to the view data of numeral, and with the frame frequency of regulation to control part 30 outputs.In addition, the shooting of camera 21～23 constantly realizes synchro control by control part 30, for example to become synchronization.
In addition, in the present embodiment, first shape is calculated the example that portion 24 camera 21～23 in addition is equivalent to photographic unit.
At this, be made as α if light is cut off the width of the Y direction of line CL, the cycle of camera 21～23 is 1/250=0.004s, if transporting velocity is set at α/0.004, then can seamlessly scan wafer 50, therefore for example being set at as transporting velocity, α/0.004 gets final product.In addition, wafer 50 is so that the direction of cutting trace is that the mode mounting of throughput direction is in delivery section 40.Thus, wafer 50 is carried along the direction of cutting trace, and along with the direction of cutting trace roughly the direction irradiates light of quadrature cut off line.
Fig. 2 is the block diagram that the electric structure of appearance inspection device shown in Figure 1 is shown.Camera 21～23 possesses first shape separately and calculates portion 24 (shape is calculated an example of mechanism).In addition, control part 30 possesses and carries that control part 31, irradiation control part 32, shooting control part 33, second shape are calculated portion 34 (shape is calculated an example of mechanism), evaluation of estimate is calculated portion 35 and detection unit 36.In addition, carrying control part 31～detection unit 36 to carry out by for example CPU is used to make computing machine to realize as the control program of control part 30 performance functions.
At this, calculate portion 24 and second shape by first shape and calculate portion 34 and constitute shape and calculate mechanism.It is parallel with the continuous photographing process that the light that is undertaken by shoot part 20 cuts off line image that shape is calculated mechanism, carry out wafer 50 three-dimensional shape data calculate processing.In detail, shape is calculated mechanism and is cut off at the light of the present frame that is undertaken by shoot part 20 during the shooting of line image, the frame that photographs before the present frame is carried out be used to calculate the processing of altitude information of cutting off the wafer 50 of the corresponding row of the irradiation position of line with light.
First shape is calculated portion's 24 execution search processing and center of gravity is calculated processing, it is to cut off at each light that photographs by camera 21～23 to set the many bar lines parallel with throughput direction in the line image that described search is handled, and the high-high brightness pixel of searching for each line, described center of gravity is calculated and handled is according to handling the high-high brightness pixel of each line that searches and calculate the barycentric coordinates of the high-high brightness of each line with sub-pix (sub pixel) unit by search.
At this, first shape is calculated portion 24 during the light of the present frame that is undertaken by camera 21～23 cuts off the shooting of line image, when execution was handled for the search of the light cut-out line image of the previous frame of present frame, execution was calculated processing for the center of gravity of the light cut-out line image of two frames before the present frame.
In addition, first shape is calculated search processing and the center of gravity that portion 24 carries out the each several part zone corresponding with camera 21～23 respectively and is calculated processing.
Second shape is calculated portion 34 according to the barycentric coordinates of being calculated each line that portion 24 calculates by first shape and the elevation angle of camera 21～23 and light source 11～13, calculate the altitude information of the described solar cell wafer of row corresponding with the irradiation position of light cut-out line, by carrying out above-mentioned processing repeatedly, thereby calculate whole three-dimensional shape data of wafer 50.
In addition, second shape is calculated the three-dimensional shape data that portion 34 calculates the each several part zone corresponding with camera 21～23 respectively, for the zone of repeating with adjacent subregion in the subregion, overlap with the three-dimensional shape data of another subregion by the three-dimensional shape data that makes a subregion, thereby calculate the three-dimensional shape data of 50 whole of wafers.
At this, second shape is calculated portion 34 and is used least square methods that the three-dimensional shape data in each several part zone is carried out the plane revisal, for the zone of repeating with adjacent subregion in the subregion, by the three-dimensional shape data after the plane revisal of the three-dimensional shape data after the plane revisal of a subregion and another subregion is weighted addition, thereby calculate the three-dimensional shape data of 50 whole of wafers.
Evaluation of estimate is calculated portion 35 by calculate the evaluation of estimate that three-dimensional shape data that portion 34 calculates is calculated the expression roughness of wafer 50 by second shape.At this, evaluation of estimate is calculated portion 35 and is carried out from calculating the processing that three-dimensional shape data that portion 34 calculates is removed low-frequency component by shape, and calculates evaluation of estimate by the three-dimensional shape data after removing.
Specifically, evaluation of estimate is calculated portion 35 and is used Gaussian filter to extract low-frequency component from the three-dimensional shape data of each row, and deducts the low-frequency component of each row of extraction from the three-dimensional shape data of each original row, removes low-frequency component thus.
Operating portion 60 for example is made of keyboard and mouse, receives from the various instructions of operating personnel's input.Display part 70 for example is made of liquid crystal panel, shows that the various application drawing pictures or second shape calculate image of three-dimensional shape data that portion 34 calculates or the result of determination that detection unit 36 is shown etc.
Next, use process flow diagram that the detailed process of the processing of this appearance inspection device is described.Fig. 3 is the process flow diagram of master routine that the appearance inspection device of embodiments of the present invention is shown.Need to prove that following processing is carried out respectively the view data that the light that photographs cuts off line image in each camera 21～23.In addition, in Fig. 3, the light that camera 21～23 is taken X frame respectively cuts off line image.At first, drive delivery section 40, the conveying (step S1) of beginning wafer 50 by carrying control part 31.
Next, 21～23 pairs of wafers 50 of camera are taken, and the light of obtaining first frame cuts off the view data (step S2 (1)) of line image.
Fig. 8 illustrates the figure that light cuts off an example of line image.In addition, light shown in Figure 8 cuts off in the line image, and the pixel count of directions X (vertical direction) is M, and the pixel count of Y direction (throughput direction) is N, that is, light shown in Figure 8 cut off line image be M capable * view data of N row.In addition, the pixel value of each pixel is for example represented with 0～255 256 gray scales.Below, it is brightness value that pixel value is recorded and narrated.
As shown in Figure 8, occurred as can be known along the light cut-out line CL of the wire of directions X.Return Fig. 3 and describe, in step S2 (2), 21～23 pairs of wafers 50 of camera are taken, and the light of obtaining second frame cuts off the view data of line image.Meanwhile, first shape is calculated the view data execution search processing of the light cut-out line image of 24 pairs first frames of portion.
In step S2 (3), 21～23 pairs of wafers 50 of camera are taken, and the light of obtaining the 3rd frame cuts off the view data of line image.Meanwhile, the light that first shape is calculated 24 pairs second frames of portion cuts off the view data execution search processing of line image, and search light cuts off the high-high brightness pixel of each line of line image.Meanwhile, first shape is calculated the light cut-out line image execution center of gravity of 24 pairs first frames of portion and is calculated processing, and calculates the barycentric coordinates of the high-high brightness of each line with sub-pix unit.
In the moment that step S2 (3) finishes, can access the barycentric coordinates of cutting off the first corresponding row of line with the light that occurs at first frame.
After, at step S2 (4)～S2 (X), carry out and the same processing of step S2 (3) repeatedly.In the moment that step S2 (X) finishes, can access the barycentric coordinates of first row～the X-2 row.
In step S2 (X+1), because the shooting that camera 21～23 carries out finishes, therefore only carry out search processing and center of gravity and handle, can access the barycentric coordinates of X-1 row.
In step S2 (X+2), because handling, search also finishes, therefore only carrying out center of gravity calculates processing, can access the barycentric coordinates of X row.
From above pipeline processes, at every turn the light of taking a frame through camera 21～23 cut off line image during, just can access one and be listed as pairing barycentric coordinates.
Fig. 4 is the process flow diagram that the detailed process of search processing is shown.At first, first shape is calculated portion 24 and the light of the former frame of the present frame of camera 21～23 current shooting is cut off the light that line image sets as process object is cut off line image (step S211).
Next, first shape is calculated portion 24 0 substitution being used for of setting on light cuts off line image is represented the variable i of the wire size of each line, with i initialization (step S212).At this moment, as shown in Figure 8, cutting off on the line image at light as can be known has a line along the Y direction setting.In addition, represent that with i=0 light shown in Figure 8 cuts off first row of line image, i=1 represents that light shown in Figure 8 cuts off the mode of second row of line image, and a line cuts off the delegation of line image corresponding to light, and variable i is corresponding with each row that light cuts off line image.
Next, to calculate portion 24 searches for the brightness maximum in i bar line pixel be brightness maximum pixel (step S213) to first shape.At this moment, first shape is calculated portion 24 in i bar line shown in Figure 8, searches for the brightness maximum pixel by will for example setting successively as concerned pixel to the pixel of right-hand member from the pixel of left end.Specifically, at first the pixel of left end is set as concerned pixel, its brightness value and coordinate are stored in the illustrated memory buffer of omission.At this, can adopt expression to count round values into which pixel from the pixel of left end as coordinate.
Next, the right side adjacent pixels being set as concerned pixel, is under the situation about being stored in more than the brightness value of memory buffer at the brightness value of concerned pixel, and the brightness value and the coordinate of memory buffer with concerned pixel upgraded.On the other hand, under the situation of brightness value less than the brightness value that is stored in memory buffer of concerned pixel, updated stored is not in the brightness value and the coordinate of memory buffer.Carry out such processing repeatedly, the coordinate decision of finally storing in the memory buffer is the coordinate Xp of high-high brightness pixel, search the high-high brightness pixel of i bar line.
And the coordinate Xp of high-high brightness pixel that makes the i bar line of obtaining is corresponding with variable i and store into and omit in the illustrated memory buffer.
Next, first shape is calculated portion 24 (among step S214 " being (YES) ") under the situation of the processing end of the coordinate Xp that whole lines is obtained the high-high brightness pixel returns processing, under the unclosed situation of processing of the coordinate Xp that whole lines is obtained the high-high brightness pixel ("No" among the step S214 (NO)), make i add 1 (step S215), and make processing return step S213.
That is, first shape is calculated portion 24 and is obtained the coordinate Xp of high-high brightness pixel that light shown in Figure 8 cuts off whole lines of line image by the processing of repeating step S213～S215.
Fig. 5 illustrates the process flow diagram that center of gravity is calculated the detailed process of processing.At first, first shape is calculated portion 24 and the light of two frames before the present frame of camera 21～23 current shooting is cut off the light that line image sets as process object is cut off line image (step S221).
Next, first shape is calculated portion 24 and is searched for processing similarly with 0 substitution variable i, with i initialization (step S222).Next, first shape calculate portion 24 with the high-high brightness pixel that in i bar line, searches as the center, about extract n neighboring pixel out, and use high-high brightness pixel and 2n neighboring pixel are obtained the barycentric coordinates Xsub (step S223) of the high-high brightness of i bar line.
Fig. 9 is the chart that the distribution of the brightness value so that the high-high brightness pixel is the center in the i bar line is shown.Among Fig. 9, coordinate Xp is the coordinate of high-high brightness pixel, 8 on right side, 8 in left side are shown and add up to the distribution of the brightness value of 17 pixels.
And first shape is calculated portion 24 and is used for example following formula to obtain the barycentric coordinates of high-high brightness.
[formula 1]
Wherein, Xsub represents the barycentric coordinates of high-high brightness, and Xj represents the coordinate of the j pixel on the i bar line, Kj represents the brightness value of Xj, Xp represents the coordinate of high-high brightness pixel, and Kp represents the brightness value of high-high brightness pixel, and n is the label that is used for determining neighboring pixel.
Thus, in i bar line, the barycentric coordinates Xsub of high-high brightness is the value of the following mixed decimal point of 1 pixel, promptly obtains with sub-pix unit.In addition, first shape is calculated portion 24 and is pre-determined which position of the barycentric coordinates Xsub of high-high brightness being obtained radix point, if the barycentric coordinates Xsub of high-high brightness surpasses this figure place, then round up, cast out or carry etc. handled and got final product.
In the example of Fig. 9, the X value of the maximum of the chart of being represented by solid line becomes the barycentric coordinates Xsub of dominant bit brightness.In addition, in the example of Fig. 9, the number that makes neighboring pixel is n=8, but this only is an example, then also can adopt other suitable values in the not huge scope of calculated amount as long as be more than 1.
Returning Fig. 5 describes, first shape is calculated portion 24 under the situation of the processing end of the barycentric coordinates Xsub that whole lines is obtained high-high brightness ("Yes" among the step S224 (YES)), make to handle and return, under the unclosed situation of processing of whole lines being obtained barycentric coordinates Xsub (NO among the step S224), make i add 1 (step S225), and make processing return step S223.
That is, first shape is calculated portion 24 by carrying out the processing of step S223～S225 repeatedly, thereby obtains the barycentric coordinates Xsub of high-high brightness that light shown in Figure 8 cuts off whole lines of line image.
In addition, it is corresponding and store in the illustrated memory buffer of omission to make barycentric coordinates Xsub and the light of the high-high brightness of whole lines cut off the frame number of line image and variable i.
Fig. 6 is the sequential chart of processing that the process flow diagram of Fig. 3 is shown with time series.T1～T during shown in Figure 6 (X+1) represents that respectively frame light of camera 21～33 shootings cuts off needed time of line image, the frame period of being.
During T1, carry out step S2 (1) shown in Figure 3, the light of taking first frame cuts off line image.During T2, carry out step S2 (2) shown in Figure 3, the search that the light that the light of second frame cuts off the shooting of line image and first frame cuts off line image is handled and is carried out simultaneously.
During T3, carry out step S2 (3) shown in Figure 3, the center of gravity that the search that the light that the light of the 3rd frame cuts off shooting, second frame of line image cuts off line image is handled and the light of first frame cuts off line image is calculated processing and is carried out simultaneously.
After, T (X) extremely, the center of gravity that the search that the light that the light of present frame cuts off shooting, the former frame of line image cuts off line image is handled and the light of preceding second frame cuts off line image is calculated processing and carried out simultaneously, and is every through during one, the barycentric coordinates Xsub that then calculates the high-high brightness of 1 row.
Fig. 7 is the figure of the flow process of schematically illustrated search processing and the center of gravity processing of calculating processing.Longitudinal axis express time axle shown in Figure 7 is carved with scale by the frame period of camera 21～23.Need to prove that as mentioned above, the frame frequency of camera 21～23 is 250fps, so the frame period is 4msec.In addition, in Fig. 7, the line number of setting in frame light cut-out line image is 480 of i=0～479.
During T (n), take wafer 50, the light of obtaining the n frame cuts off line image.Next, during T (n+1), the wafer 50 that nothing is carried with certain speed towards the left side with stopping being taken, the light of obtaining the n+1 frame cuts off line image.Next, during T (n+2), the wafer 50 that nothing is carried with certain speed towards the left side with stopping being taken, the light of obtaining the n+2 frame cuts off line image.
And, during T (n), the light of n-1 frame is cut off line image sets i=0～479 line, in the high-high brightness pixel of each line of search, light to the n-2 frame cuts off line image setting i=0～279 line, and the barycentric coordinates Xsub that calculates the high-high brightness of each line.
In addition, during T (n+1), T (n+2), also carry out with during the same processing of T (n).Thus, T during every process (n) then can access the barycentric coordinates Xsub of the high-high brightness of i=0～479.
Returning Fig. 3 describes, in step S3, second shape calculates that portion 34 will calculate that portion is 24 that calculate by first shape, the barycentric coordinates Xsub of the row that line image calculates of high-high brightness cut off to(for) the light of frame substitution following formula respectively, calculate altitude information, with the altitude information of calculating series arrangement, calculate the cross sectional shape data according to variable i.
h(μm)＝R·Xsub·cosθ/sin(θ+φ)
Wherein, h represents altitude information, and R represents visual field resolution, and θ represents the elevation angle of light source 11～13, and φ represents the elevation angle of camera.
Figure 10 is the key diagram of calculating the processing of altitude information.Figure 11 is the figure of the state that is provided with of schematically illustrated light source 11～13 and camera 21～23.Quadrilateral shown in Figure 10 represents that light cuts off line image, and the light that the thick line shown in the quadrilateral represents to utilize the barycentric coordinates Xsub of the high-high brightness of whole lines to describe cuts off line CL '.
As shown in figure 11, the elevation angle separately of light source 11～13 is that benchmark is set at θ with the Z direction, and the elevation angle separately of camera 21～23 is that benchmark is set at φ with the Z direction.And the position of inciding camera 21～23 from the light of light source 11～13 irradiation is according to the height of wafer 50 and in the front and back along Y direction shown in Figure 10.Thus, by barycentric coordinates Xsub substitution following formula, can obtain the altitude information of the position of wafer 50 with high-high brightness.
In addition, in Figure 10, if the length longitudinally that makes light cut off line image is ML (μ m), horizontal length is NL (μ m), makes longitudinally that pixel count is M, and horizontal pixel count is N, and then resolution R in the visual field is R=NL/N.
And second shape is calculated portion 34 and will be cut off M the altitude information h that each Xsub on the line CL ' obtains to light and be arranged in row.Thus, can access the cross sectional shape data of wafer 50.
In addition, in Figure 11, preferred θ+φ is 90 degree.In addition, θ for example is 45 degree～82 degree, preferred 60 degree～82 degree.In addition, φ is for example 45 degree～8 degree, preferred 30 degree～8 degree.Thus, light source 11～13 is from oblique irradiate wafer 50, and camera 21～23 can be taken wafer 50 from the top, can access high-resolution altitude information.
Figure 12 illustrates the chart of calculating the cross sectional shape data that portion 34 calculates by second shape, and the longitudinal axis is represented height with the μ m of unit, and transverse axis represents that light cuts off the line direction, promptly with the mm of unit represent shown in Figure 10 vertically.As shown in figure 12 as can be known, can correctly calculate cross sectional shape data small concavo-convex of wafer 50.
Return Fig. 3 and describe, in step S4, second shape is calculated portion 34 and is arranged in order according to frame number by the cross sectional shape data that will obtain, thereby calculates whole three-dimensional shape data of subregion.
In the example of Fig. 3, be the X width of cloth because light cuts off the frame number of line image, therefore can access the three-dimensional shape data of cross sectional shape data with the subregion of X row arrangement.In addition, the platform number of camera is 3, therefore can access the three-dimensional shape data with respect to three subregions.
Figure 13 is the figure of the three-dimensional shape data of simulation and schematically illustrated certain subregion of calculating the wafer 50 that portion 34 calculates by second shape.As shown in figure 13, by with cross sectional shape data ordering shown in Figure 12, thereby can reproduce the 3D shape of the subregion of wafer 50.In addition, as shown in figure 13 as can be known, a plurality of grooves occur, reproduced the cutting trace truly along the Y direction.
Next, return Fig. 3 and describe, second shape calculate portion 34 carry out three-dimensional shape data that each subregion is obtained in conjunction with and whole three-dimensional shape data obtaining the three-dimensional shape data of 50 whole of wafers generates and handles (step S5).Figure 14 illustrates whole three-dimensional shape data to generate the process flow diagram of handling.
At first, second shape is calculated portion 34 the variable k of camera 21～23 is represented in 1 substitution, with k initialization (step S31).In the present embodiment, because camera is 3, so k gets a certain round values in k=1～3, the k=1～corresponding camera 21～23 of 3 difference.
Next, second shape is calculated portion 34 and is obtained the three-dimensional shape data (step S32) corresponding with the subregion of k platform, k+1 platform camera.Next, second shape is calculated portion 34 and is used least square methods pair to carry out plane revisal (step S33) respectively with the regional corresponding three-dimensional shape data of each several part.
Specifically, second shape is calculated portion 34, and to make the formula on the plane of subregion be z=ax+by+c, obtains a, the b, the c that make following E minimum, thereby a, the b that obtains, the formula on the above-mentioned plane of c substitution are obtained the formula on the plane of subregion.
E＝∑(z-(ax+by+c))
2
Wherein, x, y, z are corresponding with X, Y, Z direction shown in Figure 1.
Then, second shape is calculated the three-dimensional shape data of the subregion of portion's 34 usefulness (x ', y ', z ') expression k platform camera, thereby utilizes z '-z that the three-dimensional shape data of subregion is carried out the plane revisal.
Wherein, the z of z '-z passes through the value that z=ax '+by '+c obtains.
Thus, can obtain the three-dimensional shape data of subregion, and the error of the concavo-convex caused three-dimensional shape data of the sensitive surface of elimination camera.
Next, second shape is calculated the zone that portion 34 repeats mutually for the subregion of k platform camera and the subregion of k+1 platform camera, is weighted the zone (step S34) that repeat in phase Calais revisal two parts zone by the three-dimensional data with two parts zone.
Figure 15 is the figure that explanation is weighted the processing of addition.As shown in figure 15, the scope of the X coordinate of the subregion Dk of k platform camera is p～r, and the scope of the X coordinate of the subregion Dk+1 of k+1 platform camera is q～s, has the relation of p＜q＜r＜s.Therefore, the x of subregion Dk and subregion Dk+1 is overlapping in the scope of q≤x≤r.In addition, as shown in Figure 1, the value of p～s can be in advance obtained by the elevation angle of camera 21～23 and camera angle etc., adopts this value to get final product.
Therefore, second shape is calculated portion 34 and is used following formulas to be weighted addition for the three-dimensional shape data of overlapping areas after with the plane revisal of subregion Dk and subregion Dk+1.
f
new(x，y)＝((f
k(x，y)·(r-x)+f
k+1(x，y)·(x-q))/(r-q)(q≤x≤r)
Wherein, f
New(x, the y) altitude information behind the expression weighting summation, f
k(x, the y) altitude information after the plane revisal of expression subregion Dk, f
K+1(x, y) altitude information after the plane revisal of expression subregion Dk+1.
Promptly, along with x near q, be weighted addition so that the composition of the altitude information of subregion Dk increases to the mode bigger than subregion Dk+1, along with x near r be weighted addition so that the composition of the altitude information of subregion Dk+1 increases to the mode bigger than subregion Dk.
Next, second shape is calculated portion 34 under the situation that the processing to the three-dimensional shape data of the subregion of all cameras finishes ("Yes" among the step S35 (YES)), make to handle and return, under to the unclosed situation of processing of the three-dimensional shape data of the subregion of all cameras ("No" among the step S35 (NO)), make to handle to advance to step S36.
In step S36, second shape is calculated portion 34 subregion Dk is combined (step S36) with the three-dimensional shape data of subregion Dk+1.In this case, second shape is calculated portion 34 for the (p≤x＜q) of the zone underlapped with subregion Dk+1 among the Dk of subregion, three-dimensional shape data after the plane revisal of directly adopting subregion Dk is shown below, for (r＜the x≤s) of the zone underlapped among the Dk+1 of subregion with subregion Dk, be shown below, directly adopt the three-dimensional shape data after the plane revisal of subregion Dk+1, for subregion Dk and subregion Dk+1 overlapping areas (q≤x≤r), the three-dimensional shape data that adopts above-mentioned weighting summation to obtain.
f
new(x，y)＝f
k(x，y)(p≤x＜q)
f
new(x，y)＝f
k+1(x，y)(r＜x≤s)
Next, second shape is calculated portion 34 makes k add 1 (step S37), and makes processing be back to step S32.Thus, obtain the three-dimensional shape data of 50 whole of wafers.
Return Fig. 3 and describe, in step S6, evaluation of estimate calculates portion 35 and detection unit 36 is carried out the evaluation processing.Figure 16 illustrates to estimate the process flow diagram of handling.At first, evaluation of estimate is calculated the three-dimensional shape data (step S41) that portion 35 obtains 50 whole of wafers.Next, evaluation of estimate is calculated the variable y that portion 35 is used to specify 0 substitution the row of three-dimensional shape data, with y initialization (step S42).
Next, evaluation of estimate is calculated portion 35 and is removed low-frequency component (step S43) from the cross sectional shape data of y row.At this, evaluation of estimate is calculated portion 35 and is at first utilized Gaussian filter to extract low-frequency component to each altitude information of the cross sectional shape data that constitute y row.
Specifically, evaluation of estimate is calculated portion 35 and is used following formula extraction low-frequency component.
w(x)＝∫-∞
∞p(x’)·s(x-x’)dx’
Wherein, the low-frequency component of the cross sectional shape data of w (x) expression y row, the cross sectional shape data of p (x) expression y row.
In addition, s (x) is the normal distribution that is expressed from the next.
s(x)＝(1/α·λc)·exp(-(x/α·λc)
2)
Wherein,
Promptly, evaluation of estimate is calculated portion 35 and in the cross sectional shape data of y row the altitude information from the altitude information of number one to last number is set at the concern altitude information successively, by paying close attention to altitude information is that center pair cross-section shape data p (x) multiply by s (x) and integration with each, thereby obtains the low-frequency component of respectively paying close attention to altitude information.
And evaluation of estimate is calculated portion 35 after respectively being paid close attention to the low-frequency component w (x) of altitude information, removes low-frequency component w (x) by r (x)=p (x)-w (x) from cross sectional shape data p (x).Thus, obtain the radio-frequency component r (x) of cross sectional shape data p (x).
Figure 17 illustrates the cross sectional shape data of removing behind the low-frequency component.As shown in figure 17 as can be known, the big complications of representing among Figure 12 are flattened, and the cross sectional shape data are that benchmark changes with a certain height level.
Next, evaluation of estimate is calculated portion 35 and (be "Yes" (YES) among the step S44) under the situation that the processing of whole three-dimensional shape data of wafer 50 being removed low-frequency component finishes, make to handle and advance to step S46, under the unclosed situation of processing of whole three-dimensional shape data of wafer 50 being removed low-frequency component, (be "No" (NO) among the step S44), make to handle to advance to step S45.
In step S45, evaluation of estimate is calculated portion 35 makes y add 1, and makes processing return step S43.That is,, carry out the processing of removing low-frequency component repeatedly, finally remove low-frequency component from whole three-dimensional shape data of wafer 50 from a row three-dimensional shape data by the processing of step S43～S45 repeatedly.
Figure 18 is simulation or the schematically illustrated figure that removes the three-dimensional shape data behind the low-frequency component.As shown in figure 18 as can be known, the big complications that show among Figure 13 are flattened, and three-dimensional shape data is that benchmark changes with a certain height level parallel with X-Y plane.
In step S46, evaluation of estimate is calculated the summation that portion 35 obtains the absolute value of removing the altitude information behind the low-frequency component, and this summation divided by whole numbers of altitude informations, is calculated the evaluation of estimate of the roughness of expression wafer 50 thus.
Next, detection unit 36 is ("Yes" among the step S47 (YES)) under the situation of evaluation of estimate greater than threshold value, judging that wafer 50 be unacceptable product (step S48), is that judgement wafer 50 is certified products (step S49) under the situation below the threshold value ("No" among the step S47 (NO)) in evaluation of estimate.Thus, finish the processing of this appearance inspection device.In addition, adopt predefined expression wafer 50 to get final product as threshold value for the value of unacceptable product.
So, according to this appearance inspection device, the wafer of being carried with certain speed by delivery section 40 50 cuts off line by the illuminated light of direction that edge and throughput direction intersect, and cutting off line, and take continuously as light cut-out line image with some cycles by shoot part 20 with the mode irradiates light that intersects of cutting trace.Therefore, shoot part 20 can be taken next light in the moment that makes light cut off the immediate vicinity of the light cut-out line that shines when once light cut off the shooting of line image before line shone and cut off line image.
Thus, can utilize light cross-section method to cut off line image and obtain the cross sectional shape data of wafer 50 that light cuts off the irradiation position of line, can access whole three-dimensional shape data of wafer 50 by arranging these cross sectional shape data from each light.
In addition, be divided into a plurality of subregions by 21～23 pairs of wafers 50 of a plurality of cameras and take, and calculate the three-dimensional shape data in each several part zone respectively.And,, overlap the three-dimensional shape data that obtains 50 whole of wafers with the three-dimensional shape data of another subregion by the three-dimensional shape data that makes a subregion in the subregion and adjacent partly overlapping zone.Therefore, compare, can access high-resolution light and cut off line image with the situation of wafer 50 being taken by a camera.Consequently can calculate high-resolution three-dimensional shape data.
And, calculate the evaluation of estimate of the roughness of expression wafer 50 by whole three-dimensional shape data of the high-resolution wafer 50 that so obtains, and use this evaluation of estimate to judge the quality of wafer 50, therefore can judge correctly whether wafer 50 is unacceptable product.
Need to prove, light source is set at three in the above-described embodiment, but the present invention is not limited to this, also can be a light source.In addition, also can constitute irradiation portion 10, the light from light source is penetrated from injection part by a light source and a plurality of injection part.At this moment, cutting off line from the light of irradiation portion 10 irradiation need be to the whole area illumination of the Width of wafer 50.In addition, the platform number of camera is set at three, but the present invention is not limited to this, also can be made of a plurality of cameras more than two or four.In addition, corresponding one to one by making each camera with light source or irradiation portion, and make the configuration relation of each camera and each light source or irradiation portion all constant, can make the resolution of the light cut-out line image that photographs by each camera identical thus.
The one-piece construction figure of the appearance inspection device when Figure 19 illustrates light source and camera and is.This mode for example is applicable to that with the little wafer 50 of size be the situation of checking object.In addition,,, therefore need not handle by the above-mentioned weighting summation that carry out with adjacent subregion overlapping areas in the subregion, can realize the simplification of handling owing to there is not the subregion according to this mode.
In addition, in the above description, first shape calculates portion 24 and second shape portion of calculating was arranged in 34 minutes, but also can one constitute.At this moment, each camera 21～23 can be located at, also control part 30 can be located at.In addition, in the above description, photographing process, search are handled and center of gravity is calculated processing and handled parallel processing with three class pipeline (pipeline), but comprised in this parallel processing that the processing of the cross sectional shape data of obtaining row also can.At this moment, make the processing that photographing process, search are handled, center of gravity is calculated the cross sectional shape data of handling and obtain row realize getting final product with the level Four pipeline processes.
Claims (11)
1. appearance inspection device, the outward appearance that it checks the solar cell wafer wherein, possesses:
Delivery section, it is along the length direction of the cutting trace that forms on described solar cell wafer and carry described solar cell wafer with certain speed;
Irradiation portion, it cuts off line to solar cell wafer edge of being carried by described delivery section and the direction irradiates light that throughput direction intersects;
Photographic unit, its with some cycles continuously pickup light cut off line image, this light cuts off line image and comprises that the light by the irradiation of described irradiation portion cuts off line;
Shape is calculated mechanism, and it cuts off the view data of line image according to several light that photograph by described photographic unit, calculates the three-dimensional shape data of described solar cell wafer,
Described shape calculate continuous photographing process that mechanism and the described light that is undertaken by described photographic unit cuts off line image carry out concurrently described solar cell wafer three-dimensional shape data calculate processing.
2. appearance inspection device according to claim 1 is characterized in that,
Cut off at the light that is carried out present frame by described photographic unit during the shooting of line image, described shape is calculated mechanism and the frame that photographed before described present frame is carried out is used to calculate the processing of cutting off the altitude information of the corresponding described solar cell wafer of the irradiation position of line with described light.
3. appearance inspection device according to claim 1 is characterized in that,
Described shape is calculated mechanism and is possessed first shape and calculate portion, this first shape portion of calculating carries out the search processing and center of gravity is calculated processing, it is to cut off in the line image at each light that photographs by described photographic unit that described search is handled, set many lines abreast with described throughput direction, and the high-high brightness pixel of searching for each line, it is according to the high-high brightness pixel of handling each line that searches by described search that described center of gravity is calculated processing, calculates the barycentric coordinates of the high-high brightness of each line with sub-pix unit
Described first shape is calculated portion and is cut off during the shooting of line image at the light that is carried out present frame by described photographic unit, cut off line image for the light of the former frame of described present frame and carry out described search and handle, cut off line image for light simultaneously and carry out described center of gravity and calculate processing from several forward second frames of described present frame.
4. appearance inspection device according to claim 3 is characterized in that,
Described shape is calculated mechanism and is possessed second shape and calculate portion, this second shape is calculated portion and is calculated the three-dimensional shape data of described solar cell wafer by carrying out following processing repeatedly, this is treated to: according to the barycentric coordinates of each line of being calculated by the described first shape portion of calculating and the elevation angle of described photographic unit and irradiation portion, calculate the altitude information of the described solar cell wafer of row corresponding with the irradiation position of described light cut-out line.
5. appearance inspection device according to claim 1 is characterized in that,
Described photographic unit is taken described solar cell wafer from the top,
Described irradiation portion is from the described solar cell wafer of oblique irradiation.
6. appearance inspection device according to claim 1 is characterized in that,
Take for described solar cell wafer is divided into a plurality of subregions, described photographic unit has a plurality of,
Described shape is calculated the three-dimensional shape data that mechanism calculates the each several part zone respectively, for the zone of repeating with adjacent subregion in the subregion, overlap with the three-dimensional shape data of another subregion by the three-dimensional shape data that makes a subregion, thereby calculate the three-dimensional shape data of described solar cell wafer.
7. appearance inspection device according to claim 6 is characterized in that,
Described shape is calculated mechanism and is used least square method that the three-dimensional shape data in each several part zone is carried out the plane revisal, for the zone of described repetition, with the three-dimensional shape data weighting summation after the plane revisal of the three-dimensional shape data after the plane revisal of a subregion and another subregion.
8. according to each described appearance inspection device in the claim 1～7, it is characterized in that,
Described appearance inspection device also possesses evaluation of estimate and calculates mechanism, and this evaluation of estimate is calculated mechanism by calculate the evaluation of estimate that three-dimensional shape data that mechanism calculates is calculated the roughness of the described solar cell wafer of expression by described shape;
Decision mechanism, it judges according to described evaluation of estimate whether described solar cell wafer is qualified.
9. appearance inspection device according to claim 8 is characterized in that,
Described evaluation of estimate is calculated mechanism and is carried out the processing of removing low-frequency component from described three-dimensional shape data, and calculates described evaluation of estimate by the three-dimensional shape data of removing behind the low-frequency component.
10. appearance inspection device according to claim 9 is characterized in that,
Described evaluation of estimate is calculated mechanism the three-dimensional shape data of each row is used Gaussian filter and extracts low-frequency component, and deducts low-frequency component of each row of extraction from each original row three-dimensional shape data, thereby removes low-frequency component.
11. each described appearance inspection device is characterized in that according to Claim 8,
If described evaluation of estimate is judged to be unacceptable product greater than the then described decision mechanism of threshold value, if be judged to be certified products less than the then described decision mechanism of threshold value.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2010025288A JP2011163852A (en) | 2010-02-08 | 2010-02-08 | Visual inspection device |
| JP2010-025288 | 2010-02-08 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102192713A true CN102192713A (en) | 2011-09-21 |
Family
ID=44594718
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2011100350603A Pending CN102192713A (en) | 2010-02-08 | 2011-01-30 | Appearance checking device |
Country Status (3)
| Country | Link |
|---|---|
| JP (1) | JP2011163852A (en) |
| CN (1) | CN102192713A (en) |
| TW (1) | TWI415014B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103038603A (en) * | 2010-07-30 | 2013-04-10 | 克拉-坦科股份有限公司 | Apparatus and method for three dimensional inspection of wafer saw marks |
| CN104956210A (en) * | 2013-01-30 | 2015-09-30 | 住友化学株式会社 | Image generating device, defect inspecting device, and defect inspecting method |
| CN105190228A (en) * | 2013-03-27 | 2015-12-23 | 株式会社尼康 | Shape measurement device, structure production system, shape measurement method, structure production method, and shape measurement program |
| CN108604880A (en) * | 2015-10-26 | 2018-09-28 | 应用材料公司 | Detect the method and system of the gap in solar wafer |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6212926B2 (en) * | 2013-04-26 | 2017-10-18 | 住友金属鉱山株式会社 | Object inspection apparatus and object inspection method |
| JP6225594B2 (en) * | 2013-09-20 | 2017-11-08 | 大日本印刷株式会社 | Height data processing apparatus, height data processing method, program, storage medium, embossed plate manufacturing method, sheet manufacturing method |
| US9523771B2 (en) * | 2014-01-13 | 2016-12-20 | Facebook, Inc. | Sub-resolution optical detection |
| KR102073803B1 (en) * | 2016-07-27 | 2020-02-05 | 주고꾸 도료 가부시키가이샤 | Three-dimensional surface roughness evaluating device, three-dimensional surface roughness evaluating method, three-dimensional surface roughness data acquiring device, and three-dimensional surface roughness data acquiring method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4929843A (en) * | 1989-06-28 | 1990-05-29 | General Electric Company | Apparatus and method for determining a dimension of an object |
| JPH07260444A (en) * | 1994-03-24 | 1995-10-13 | Nippon Avionics Co Ltd | Method and apparatus for measuring object three-dimensionally by light section method |
| CN1474160A (en) * | 2003-07-02 | 2004-02-11 | 西安邮电学院 | Multiple-section synthesizing three-dimensional profile measuring method |
| JP2004226240A (en) * | 2003-01-23 | 2004-08-12 | Nippon Steel Corp | Optical shape measuring method |
Family Cites Families (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0749942B2 (en) * | 1990-05-29 | 1995-05-31 | 日本無線株式会社 | Road surface unevenness measuring device |
| JP2930746B2 (en) * | 1991-01-22 | 1999-08-03 | 株式会社豊田中央研究所 | Parts inspection equipment |
| JPH09229632A (en) * | 1996-02-27 | 1997-09-05 | Toray Ind Inc | Image formation output apparatus and method, and shape measuring apparatus and method |
| JPH10281737A (en) * | 1997-04-04 | 1998-10-23 | Nikon Corp | Method for synthesizing wave front |
| JPH11218413A (en) * | 1998-01-30 | 1999-08-10 | Sumitomo Metal Ind Ltd | Moving object shape-measuring method and equipment therefor |
| JP3152203B2 (en) * | 1998-05-27 | 2001-04-03 | 株式会社東京精密 | Appearance inspection device |
| JP2000292132A (en) * | 1999-04-08 | 2000-10-20 | Sumitomo Special Metals Co Ltd | Work quality inspecting method and apparatus |
| JP4472845B2 (en) * | 2000-08-28 | 2010-06-02 | パナソニック電工株式会社 | Image processing inspection apparatus and image processing inspection method |
| JP2002221408A (en) * | 2001-01-26 | 2002-08-09 | Minolta Co Ltd | Optical measuring device |
| JP2002243656A (en) * | 2001-02-14 | 2002-08-28 | Nec Corp | Method and equipment for visual inspection |
| JP2003148936A (en) * | 2002-10-30 | 2003-05-21 | Nippon Avionics Co Ltd | Three-dimensional measurement method for object by light-section method |
| JP3878165B2 (en) * | 2003-11-05 | 2007-02-07 | シーケーディ株式会社 | 3D measuring device |
| JP4370570B2 (en) * | 2004-03-31 | 2009-11-25 | オムロン株式会社 | Sensor system |
| JP2005345290A (en) * | 2004-06-03 | 2005-12-15 | Seiko Epson Corp | Streak-like flaw detecting method and streak-like flaw detector |
| JP2006170922A (en) * | 2004-12-20 | 2006-06-29 | Topcon Corp | Visual inspection method and its apparatus |
| JP4563183B2 (en) * | 2005-01-05 | 2010-10-13 | 大日本印刷株式会社 | Inspection method and apparatus for streaky unevenness defect |
| JP2006234553A (en) * | 2005-02-24 | 2006-09-07 | Dainippon Screen Mfg Co Ltd | Visual inspection device and visual inspection method |
| JP4713278B2 (en) * | 2005-08-30 | 2011-06-29 | シャープ株式会社 | Polycrystalline semiconductor wafer visual inspection method and visual inspection apparatus |
| JP4160991B2 (en) * | 2006-11-29 | 2008-10-08 | シャープ株式会社 | Linear defect detection apparatus and semiconductor substrate manufacturing apparatus, linear defect detection method and semiconductor substrate manufacturing method, computer connected to the detection apparatus or ... |
| JP4914761B2 (en) * | 2007-05-16 | 2012-04-11 | オリンパス株式会社 | Appearance inspection device |
| JP2009170643A (en) * | 2008-01-16 | 2009-07-30 | Mitsubishi Materials Techno Corp | Separation device and inspecting apparatus of substrate |
-
2010
- 2010-02-08 JP JP2010025288A patent/JP2011163852A/en active Pending
- 2010-12-22 TW TW99145229A patent/TWI415014B/en not_active IP Right Cessation
-
2011
- 2011-01-30 CN CN2011100350603A patent/CN102192713A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4929843A (en) * | 1989-06-28 | 1990-05-29 | General Electric Company | Apparatus and method for determining a dimension of an object |
| JPH07260444A (en) * | 1994-03-24 | 1995-10-13 | Nippon Avionics Co Ltd | Method and apparatus for measuring object three-dimensionally by light section method |
| JP2004226240A (en) * | 2003-01-23 | 2004-08-12 | Nippon Steel Corp | Optical shape measuring method |
| CN1474160A (en) * | 2003-07-02 | 2004-02-11 | 西安邮电学院 | Multiple-section synthesizing three-dimensional profile measuring method |
Non-Patent Citations (3)
| Title |
|---|
| 张明慧 等: "视频图像判读技术在实时测量系统中的应用", 《测控技术》 * |
| 李建科 等: "基于TMS320DM642的嵌入式实时图像处理系统设计及应用", 《自动化与仪器仪表》 * |
| 陈杰来 等: "数字图像处理技术在指针式仪表读数系统中的应用", 《江南大学学报（自然科学版）》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103038603A (en) * | 2010-07-30 | 2013-04-10 | 克拉-坦科股份有限公司 | Apparatus and method for three dimensional inspection of wafer saw marks |
| US9140546B2 (en) | 2010-07-30 | 2015-09-22 | Kla-Tencor Corporation | Apparatus and method for three dimensional inspection of wafer saw marks |
| CN104956210A (en) * | 2013-01-30 | 2015-09-30 | 住友化学株式会社 | Image generating device, defect inspecting device, and defect inspecting method |
| CN104956210B (en) * | 2013-01-30 | 2017-04-19 | 住友化学株式会社 | Image generating device, defect inspecting device, and defect inspecting method |
| CN105190228A (en) * | 2013-03-27 | 2015-12-23 | 株式会社尼康 | Shape measurement device, structure production system, shape measurement method, structure production method, and shape measurement program |
| CN105190228B (en) * | 2013-03-27 | 2018-11-30 | 株式会社尼康 | Shape measuring apparatus, structure manufacture system, process for measuring shape, structure manufacturing method and measuring shape formula |
| CN108604880A (en) * | 2015-10-26 | 2018-09-28 | 应用材料公司 | Detect the method and system of the gap in solar wafer |
| CN108604880B (en) * | 2015-10-26 | 2020-06-12 | 应用材料公司 | Method and system for detecting a breach in a solar wafer |
Also Published As
| Publication number | Publication date |
|---|---|
| JP2011163852A (en) | 2011-08-25 |
| TW201135609A (en) | 2011-10-16 |
| TWI415014B (en) | 2013-11-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102192713A (en) | Appearance checking device | |
| CN106705897B (en) | Method for detecting defects of arc-shaped glass panel for curved-surface electronic display screen | |
| CN100483116C (en) | Method for detecting 3D defects on surface of belt material | |
| CN103038601A (en) | Device and method for inspecting tyre shape | |
| CN106949848A (en) | A kind of high-precision laser 3D profiles phone structural detection method | |
| US20080316476A1 (en) | System and Device for the Optical Inspection of Glass Panels | |
| JP2019100917A (en) | Inspection program generation system, generation method of inspection program and generation program of inspection program | |
| BR112019016351B1 (en) | ONLINE SYSTEM AND METHOD FOR MEASURING THE OPTICAL CHARACTERISTICS OF A GLASS SLIDE | |
| CN103097879A (en) | Method and device for analyzing the optical quality of a transparent substrate | |
| CN110148106A (en) | A kind of system and method using deep learning model inspection body surface defect | |
| CN103075979A (en) | Three-dimensional surface detecting device and three-dimensional surface detecting method | |
| US20090290788A1 (en) | System and method for performing multi-image training for pattern recognition and registration | |
| CN108548825A (en) | A kind of transparent plate defect detecting device and method based on two-dimentional illumination | |
| CN110398215A (en) | Image processing apparatus and method, system, article manufacturing method, storage medium | |
| JP2015232481A (en) | Inspection device, inspection method, and program | |
| CN104655041B (en) | A kind of industrial part contour line multi-feature extraction method of additional constraint condition | |
| Pernkopf | 3D surface acquisition and reconstruction for inspection of raw steel products | |
| Mohammadikaji | Simulation-based planning of machine vision inspection systems with an application to laser triangulation | |
| US8165383B2 (en) | Method, system and computer program product for edge detection | |
| US20220092756A1 (en) | Feature detection based on neural networks | |
| JP6425406B2 (en) | INFORMATION PROCESSING APPARATUS, INFORMATION PROCESSING METHOD, AND PROGRAM | |
| CN107607537A (en) | A kind of online vision detection system of ice cream stick quality | |
| CN115564828A (en) | Machine vision positioning method and system based on ambiguity algorithm | |
| Adamo et al. | An online defects inspection system for satin glass based on machine vision | |
| Garretson et al. | An algorithm for measuring the hydraulic jump height of an airfoil in a water table |
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: 20110921 |