CN1536349A - Method for checking flat medium with pattern and its equipment - Google Patents
Method for checking flat medium with pattern and its equipment Download PDFInfo
- Publication number
- CN1536349A CN1536349A CNA2003101034401A CN200310103440A CN1536349A CN 1536349 A CN1536349 A CN 1536349A CN A2003101034401 A CNA2003101034401 A CN A2003101034401A CN 200310103440 A CN200310103440 A CN 200310103440A CN 1536349 A CN1536349 A CN 1536349A
- Authority
- CN
- China
- Prior art keywords
- defect
- candidate
- module
- defective
- check
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 41
- 230000007547 defect Effects 0.000 claims abstract description 161
- 238000012552 review Methods 0.000 claims abstract description 38
- 238000001514 detection method Methods 0.000 claims abstract description 27
- 238000007689 inspection Methods 0.000 claims abstract description 20
- 230000008569 process Effects 0.000 claims abstract description 19
- 230000002950 deficient Effects 0.000 claims description 55
- 230000033001 locomotion Effects 0.000 claims description 41
- 238000003384 imaging method Methods 0.000 claims description 36
- 230000004304 visual acuity Effects 0.000 claims description 14
- 230000000875 corresponding effect Effects 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 8
- 230000003287 optical effect Effects 0.000 claims description 7
- 230000008859 change Effects 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 4
- 230000035945 sensitivity Effects 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 2
- 238000012790 confirmation Methods 0.000 claims description 2
- 230000002596 correlated effect Effects 0.000 claims description 2
- 238000003860 storage Methods 0.000 claims description 2
- 235000011464 Pachycereus pringlei Nutrition 0.000 claims 1
- 240000006939 Pachycereus weberi Species 0.000 claims 1
- 235000011466 Pachycereus weberi Nutrition 0.000 claims 1
- 238000005259 measurement Methods 0.000 claims 1
- 230000007704 transition Effects 0.000 claims 1
- 239000011521 glass Substances 0.000 abstract description 8
- 230000006872 improvement Effects 0.000 abstract description 3
- 230000000694 effects Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 229910052751 metal Inorganic materials 0.000 description 5
- 239000002184 metal Substances 0.000 description 5
- 230000003068 static effect Effects 0.000 description 5
- 238000005070 sampling Methods 0.000 description 4
- 239000000758 substrate Substances 0.000 description 4
- 238000012546 transfer Methods 0.000 description 4
- 238000005530 etching Methods 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 239000000463 material Substances 0.000 description 3
- 238000011179 visual inspection Methods 0.000 description 3
- XUIMIQQOPSSXEZ-UHFFFAOYSA-N Silicon Chemical compound [Si] XUIMIQQOPSSXEZ-UHFFFAOYSA-N 0.000 description 2
- 230000004075 alteration Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 239000002245 particle Substances 0.000 description 2
- 238000012887 quadratic function Methods 0.000 description 2
- 230000008439 repair process Effects 0.000 description 2
- 229910052710 silicon Inorganic materials 0.000 description 2
- 239000010703 silicon Substances 0.000 description 2
- 240000004859 Gamochaeta purpurea Species 0.000 description 1
- 241000700159 Rattus Species 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 229910021417 amorphous silicon Inorganic materials 0.000 description 1
- 239000003990 capacitor Substances 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 230000001276 controlling effect Effects 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000004438 eyesight Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- AMGQUBHHOARCQH-UHFFFAOYSA-N indium;oxotin Chemical compound [In].[Sn]=O AMGQUBHHOARCQH-UHFFFAOYSA-N 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 229910044991 metal oxide Inorganic materials 0.000 description 1
- 150000004706 metal oxides Chemical class 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000010408 sweeping Methods 0.000 description 1
- 239000010409 thin film Substances 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 238000013316 zoning Methods 0.000 description 1
Images
Landscapes
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
- Image Processing (AREA)
Abstract
A concurrent low resolution/high resolution parallel scanning system is provided as an improvement in the scanning process of an inspection system for planar objects, such as large flat plates employed in panel displays, whereby lower resolution defect detection efficiently overlaps and parallels higher resolution defect review and classification stages in which defects are automatically defined and resolved. Although the invention is a valid solution for the more general problem of optically inspecting the surface of a flat article for defects, the invention is particularly useful for detecting pattern defects on large glass plates deposited with integrated-circuits for forming LCD flat panel displays.
Description
The cross reference of related application
The application is that U.S. Patent application, the sequence number submitted on May 16th, 2003 are 10/439,991 continuation, U.S. Provisional Application, the sequence number that it requires on November 1st, 2002 to submit is the rights and interests of applying date of 60/423.008, and the latter's full content is incorporated into this paper with as a reference.
Technical field
The present invention relates to general field, particularly the defective that occurs is carried out detection and classification based on machine vision on the big flat surface with pattern based on the inspection technology of machine vision.Particularly, the present invention is devoted to being deposited on the inspection such as the material on the big substrate glass plate of liquid crystal display (LCD) panel.Though the present invention can be applicable to generalized case that any flat medium with pattern is checked, The present invention be more particularly directed to the inspection to the glass substrate that is used for preformed thin film transistor (TFT) (TFT) LCD panel.
Background technology
When making the LCD panel, use big, clean thin glass sheet as substrate being used for deposit each material layer, thereby form circuit as a plurality of identical display pannels.This deposit is carried out usually stage by stage, wherein in each stage, one deck formerly (perhaps on glass substrate) is gone up according to the preset pattern that is often determined by mask and is deposited with specific material, as metal, indium tin oxide (ITO), silicon or amorphous silicon.Each the processing stage all comprise various steps, as deposit, shelter, etching and the demoulding.
Each the processing stage during and in each step in a stage, might introduce the various production defectives that influence structure, their can produce the implication (implication) of electronics and/or vision on final LCD panel product.This defective includes but not limited to short circuit, opens circuit, and external particles, mask problems, the characteristic dimension problem is crossed etching and is owed etching.For final LCD panel is correctly moved, preferably need the defective stage that produces these defectives are detected, classify and situation at needs under repair.The decision of repairing is made according to the accurate classification of defective, particularly according to the differentiation between " fatal ", " recoverable " and " process " defective.
The operation resolution that is used for the system of automatic defect detection usually has direct influence to the expense of inspection speed and system.Therefore, has only the dull and stereotyped Zone Full of the suitable scanning of relatively low resolution.Unfortunately, under this lower resolution, the past can not be carried out simultaneously by the same image data of collecting always and be detected and classification reliably.In addition, low resolution is influential for the performance of detection algorithm, and it usually produces the false alarm that a considerable amount of needs are eliminated.Therefore, after defect detection procedure, need the defect review step, in the defect review step, use higher resolution inspection (passing through camera) to obtain interested defect area, so that the back is to the checking of candidate defect, and after carry out automatically or artificial auxiliary classification.
In this generic operation, the low resolution subsystem is used to detect dull and stereotyped problem area (defects detection subsystem-DDS), and afterwards stage use independent high resolving power camera obtain these problem areas high-definition picture (defect review subsystem-DRS), thus reach more automatically or manual sort.As long as the number of the detected problem area of DDS can remain in the manageable limit,, the high-definition picture of these single-points can keep feasibility for just obtaining.In addition, this number usually has direct influence for cycle length (if checking the words of all defect areas) of system or to the check performance of system (if words of the fixing a limited number of defective of check).
Automatic visual inspection (AOI) equipment has been used to variety of issue, includes but not limited to that printed circuit board (PCB) (PCB) inspection, ultra-large integrated (VLSI) circuit chip of silicon (grain) are checked and the LCD panel is checked.The most of solutions that adopted are all based on spatial domain pattern (spatialdomain pattern) correlation technique, and these technology usually are used in combination with sensor-level pixel or subpixel accuracy collimation technique.
Transfer people's such as Levy U.S. Patent No. 4,579,455 have disclosed a kind of calibration and pattern correlation technique, and wherein a pair of 7 * 7 windows on test and the reference picture are considered, and the error sum of squares on 3 * 3 a plurality of possible subwindows in this window obtains calculating.If the least error in these combinations surpasses a threshold value, then think defectiveness.As if this method can compensate the calibration mismatch to the sensor pixel level downwards.
With by people such as the Levy invention and the relevant statement item of thick calibration accuracy of method that transfers people's such as Specht U.S. Patent No. 4,805,123 a kind of improved calibration and correlation technique that is used to detect defective has been described.In this technology, the big window in test and the reference picture is used to calculate the sensor pixel level correlativity between test and the reference.The smallest point of relevant surfaces that produces sampling is found, and quadratic function is applied to the neighbouring surface of this smallest point.Utilize this suitable quadratic function, just can obtain subpixel accuracy conversion and come calibration testing and reference picture.The calibration image the test and the calibration reference picture on 2 * 2 subwindows on by obtaining comparison with threshold value controlling chart aberration.
Also the someone proposes the changes and improvements to these basic fundamentals, for example, transfer people's such as Yolles U.S. Patent No. 5,907,628, except anything else, it also points out to use the relevant surfaces of sampling to seek the shortcoming of minimum value, and argues because surperficial thick sampling makes this point may be not corresponding to real minimum value.Therefore, they argue, and what effect is follow-up sub pixel interpolation step do not play to improving the minimum value that detects, and will produce wrong calibration result, thus the false alarm in causing detecting.People such as Yolles advise that according to the comparison entity that improves the comparison process being carried out refinement alleviates these problems.
For top any method, use single suitable (relatively low) resolution to scan checked body surface and will cause one group of possible shortcoming.These possible shortcomings necessarily comprise rational shortcoming and false alarm the two since these methods fully filtering test and reference picture between the variation of expection.This has caused the alarm of one group of candidate defect and has increased these candidate defects are verified needs with the real defect map that forms tested object.In addition, have strong need be rational classification of defects being become a plurality of defective classifications, thereby help to dispose inspected object, may allow to repair this object in some applications.
Proposed a solution along this direction in the U.S. Patent No. 5,699,447 that transfers people such as Alumot, it has described a kind of two stage sweeping scheme.Whole front panel is at first scanned by the minor diameter laser beam in the raster scan pattern by the non-charge-coupled device (CCD) optical system irrespective of size of low resolution in a stage at a relatively high speed.Then be to use subordinate phase scanning based on high-resolution CCD optical system.The back one scan stage is extracted the image by the high-resolution of the detected all defect suspected locations of last sweep phase.Though the image of the high-resolution of this solution explanation needs extraction examined object body is for checking, but it and difference of the present invention are, the phase one of checking is by the raster scanning that has the minor diameter laser beam, and twice inspection is to carry out in sequential stages.It comprises more following shortcomings:
(a) increased cycle length: the high-resolution imaging stage arrives after main detection-phase immediately.There is appreciable impact this confrontation of order proof cycle time of high-resolution imaging, because high-definition picture obtains, checks and the required time of classifying has been added to and detects on the required time of scanning.
(b) idle imaging resource: high resolving power defect review imager is the free time when detecting the imager activity, and the check imager detects imager when movable then idle.This causes the poor efficiency utilization of system resource in restriction preset time.
(c) unoptimizable review processes: classify and production environment time restriction when having stoped the imaging of all candidate defect positions when needs, stay some difficult tasks may for the user of system and decide what and which candidate defect will check that imager is collected and processing with high resolving power.
Needed is complete automatic and overlapping high resolving power defect review system, and itself and quick and precise classification technology binding operation are to improve the speed and the accuracy of inspection and reparation.
Summary of the invention
According to the present invention, it provides a kind of concurrent low-resolution/high-resolution parallel sweep system with as the improvement to the scanning process of check system, makes the low resolution defects detection stage overlapping and parallel with high resolving power defect review and sorting phase that defective is wherein defined automatically and solves effectively thus.Though the present invention is effective solution for the more generally problem of the defective of the flat body surface of optical check, the pattern defective that the present invention is deposited with detection on the big glass plate of integrated circuit LCD surface plate is particularly useful.Therefore, the present invention will obtain explanation about this application-specific.
The invention provides a kind of improved system, this system mechanically with electric on can be concurrent in time the body surface that bears inspection be carried out low resolution and high-resolution imaging.This comprises two independent with parallel imaging subsystems with different resolution, and each system can both obtain image concurrently from the checking matter surface.The low resolution imaging stage comprises defects detection subsystem (DDS), and the high-resolution imaging stage comprises defect review subsystem (DRS).In addition, each these subsystem can have one or more imaging passage successively.
DDS covers the whole surface of examined object body, typically uses a plurality of image optics parts identical, relatively low resolution (imaging optics) and photoelectric transformer (such as CCD device or complementary metal oxide semiconductor (CMOS) (CMOS) photosensitive device).Having a definite purpose of this subsystem is to distinguish candidate defect also alternatively according to presorting them with respect to the defective locations of TFT grid feature.This presorting can be sorted out candidate defect in: data line, door line, transistor, capacitor and ITO electrode.Because the number of the defective that can check in a preset time is limited, therefore can use the defective result that presorts to distinguish the priority ranking of defective competition, so that check.The priorization of candidate defect is to realize by specifying a check corresponding to the result that presorts to be worth the factor for each candidate defect.
DDS can be installed on the device of motion, and this device moves to pass through a plurality of bouts (pass) possibly along the examined object body, perhaps is fixedly installed to alternatively on the translational surface that is loaded with the examined object body.
DRS utilizes the identical image optics parts have relative high-resolution on a small quantity and photoelectric transformer to cover the very little inspection zone that is subjected to.The special purpose of this subsystem be with high resolving power to be identified and carry out imaging by the candidate defect that the DDS subsystem is presorted so that the final classification of defects that carries out automatically.DRS imaging passage can be installed on the device of motion, and this device and DDS synchronously move along the examined object body.Each DRS passage can also move through tested object independently.
By overlapping low resolution defects detection and high resolving power defect review and classification feature, the present invention seeks to reduce significantly the total cycle time that is used for the big flat object with pattern is carried out automatic visual inspection.Consider the inspection problem that is applied on the TFT LCD, this measure has increased the value of (automatic visual inspection instrument) AOI instrument for the user who operates under the restriction cycle length of being everlasting tight, has therefore increased the AOI system and has been used for the dull and stereotyped chance of checking of online 100%TFT-LCD.
Mechanical motion by DRS imaging passage can make inspection obtain simplifying, and this motion is dynamically adjusted by dispatching algorithm.This algorithm is sought by number maximization that makes the high priority candidate defect that is obtained by DRS and the distance minimization that subsystem module is moved, thereby optimizes the effort of defect review.The input of this algorithm is to be worth the factor by the space distribution of DDS detected candidate defect in the glass plate plane and by the check that DDS is assigned to candidate defect.The result of dispatching algorithm is the optimization motor pattern of each independent DRS module.This motor pattern is carried out by the DRS kinematic system then.
Therefore, the optimal scheduling algorithm minimizes the number of candidate defect of high-resolution imaging check of the DRS of camera that can not be by having given number.It also allows the prior candidate defect of high-resolution imaging is obtained priority.In a word, it or to cause be that the distribute number of needed camera of a given candidate defect reaches and minimizes, thereby reduce system cost potentially, perhaps utilize same number of camera module and improve the performance of system.
A certain embodiments of dispatching algorithm can be based on graph theory, and the purpose for demonstration obtains describing in detail in the disclosure.Yet the alternative realization of other of dispatching algorithm also is possible, and the present invention is not limited to the specific detail of the algorithm narrated.
In the motion (On-the-fly) (OTF) to focus on be a characteristic feature required for the present invention, thereby in order to carry out clear focusing and realize having low and high-resolution overlapping concurrent DDS/DRS system by high-resolution imaging system being taken to the candidate defect position.It has improved the time efficiency of overall plan, and DRS can ceaselessly be operated along the y axle.
As the Real-time and Concurrent check of the part of DRS and classification the people intervenes in the checking process needs are reached and minimize, and improved efficient cycle length of system.
The present invention and parts thereof have played the effect of the streamline of inspection and relevant treatment, and all critical stages that it makes inspection are movable fully and be used during the tested object of scanning.
With reference to the detailed description of doing below in conjunction with accompanying drawing, the present invention will be better understood.
Description of drawings
Fig. 1 is the vertical view of a zonule of faceplate part, and it shows possible defect type;
The stereographic map of Fig. 2 A and Fig. 2 B shows the stereographic map of a kind of possible implementation of the present invention respectively corresponding to relative direction of scanning;
Fig. 3 A and Fig. 3 B show feasible border;
Fig. 4 A, Fig. 4 B and Fig. 4 C show access strategy;
Fig. 5 A and Fig. 5 B show the automatic focus strategy.
Embodiment
With reference to figure 1, wherein show some the possible defectives that in LCD panel manufacture process, can find.These defectives comprise the metal bump 110 that protrudes in the ITO layer 112, protrude into the ITO projection 114 in the metal 116, the so-called damaged by rats crack 118 in the metal 112, open circuit 120 in the metal 116, transistor short circuit 124 in the transistor 122 of pixel, and the external particle in any zone 126.These defectives of each type must be subjected to an examination and classify and obtain repairing under possible situation.
According to the present invention, check that platform 10 (Fig. 2 A or 2B) is equipped with defect review subsystem (DRS) 12.DRS12 comprises a plurality of essentially identical image optics parts and photoelectric transformer with high-resolution, they form one group of camera module 14,16,18 together, and these modules can be on stand (gantry) 20 move with the scanning motion of defects detection subsystem (DDS) 24 length direction along tested object 22.Each identical camera 14,16,18 can also move through the scanning motion of defects detection subsystem 24 independently.
A kind of suitable mechanical layout of this system is illustrated among Fig. 2 A and 2B that is in two relative direction of scanning, and wherein DSR12 is placed on its scanning stand 20, and this stand 20 can move along the direction of low-resolution scan.Independent high resolving power DRS module 14,16,18 is operated with motion passes main scanning direction.For illustrative purposes, there is shown 20 DDS24 module 26-45 and 3 DRS modules 14,16,18.The actual number of module also can be by the decision that needs of application- specific.DRS module 14,16,18 not necessarily will be on independent stand 20, and they also can be installed on the same stand, as DDS stand 47.In addition, though figure the stand that uses Linear Driving here, when considering to carry out image scanning on being examined the surface, it is not unique method.Other alternative approach of carrying out this one scan can be arranged, as have the inclination rotation mirror scanner of the static optical parts of suitable design, and can correspondingly adopt algorithm disclosed herein.
The operation of DDS/DRS concurrent system 10 comprises that the low resolution DDS24 that impels on the stand 47 scans the whole zone of tested object, and scanning can be passed through a plurality of bouts (pass), and wherein each bout has covered the part of the total area.During each bout, DDS carries out basic defect inspection and classification, and produces the tabulation of defective and candidate defect.These candidate defects are lined up then, so as with DRS12 with high-resolution dynamic dispatching imaging.
Each this candidate defect also is worth to estimate with DRS and is associated, with the severity level of indication with this particular candidate defective of high-resolution imaging.Carry out the optimal motion pattern with the DRS12 of DDS24 cooperative motion, so that the number of the candidate defect that obtains at current bout and be worth maximum.The candidate defect that can not obtain at current bout is arranged at follow-up bout.Narrate in the details of this process chapters and sections below.
The minor alteration of Pingdu on scanning area of being examined big object is necessary conditions of DRS module dynamic focusing scheme in conjunction with the very little depth of field (depth-of-field) of the typical optics of DRS module 14,16,18.Therefore, the present invention has been incorporated into the DRS module 14,16,18 that can carry out dynamic auto focusing when advancing to the candidate defect position.Narrate in the details of this feature chapters and sections below.
The DRS module schedules
A feature of the present invention is the process that available high resolving power DRS module 14,16,18 is dispatched to arranged candidate defect place, to be used in the effort optimization of defect review, this point is to realize by the distance minimization that number that makes the candidate defect that obtains and DRS value maximization make module move simultaneously.A certain embodiments of this dispatching algorithm is based on graph theory and will obtain description for illustrative purposes.The invention is not restricted to the specific detail of this algorithm, also can use the alternative approach that dynamic dispatching is provided.
Defect distribution according to expection is uniform this hypothesis, can think that each DRS module 14,16,18 all is independently, and has the interaction that required minimum was divided and prevented to collide to the x span that makes each responsibility of solution, is illustrated below.
When DDS24 scanned, it disclosed the candidate defect for the treatment of by one of high resolving power DRS module 14,16,18 concurrent imaging.Because machinery quickens and to the speed limit of system design, defer to the DRS module 14,16,18 of these restrictions and perhaps can not obtain the candidate defect that all disclose in bout.The full details of particular schedule algorithm does not provide in the disclosure, but the restriction of some keys of necessary consideration that provided the tabulation of committed step and diagrammatic illustration herein, so that those of ordinary skill in the art can propose suitable dispatching algorithm.The dispatching algorithm that is used for will selected DRS module 14,16,18 being dispatched to the candidate defect place has been considered the mechanical constraint to block motion, the value that the DRS of the demand of for example maximum acceleration and speed, dynamic focusing, the defective number that can be obtained by the alternating movement pattern and each independent defective is worth.Suitable dispatching algorithm is also handled and carry out many bouts scan condition on scanning area, can not be acquired in follow-up bout at the candidate defect that obtains on the current bout thereby make effectively.
DRS module schedules process should go up in the principal direction (being expressed as the y axle in the x-y plane) of scanning considers fixed or changed window (look-ahead window) in advance, is subjected to the dynamic listing candidate defect of the optimizing scheduling processing of DRS image-forming module with decision.In addition, be considered at all untreated candidate defects and carry out under the situation of this optimizing scheduling, this window can cover whole glass span.For any set a distance of giving along the y axle, single DRS module has the suitable visit field (FOA) in a zone, and in this zone, this module can arrive any point in it, suppose that this module is static when beginning, and when moving to terminal point, be tending towards once more static.Fig. 3 A and 3B show suitable visit field.
Shown in Fig. 3 A and 3B, the window in advance of preliminary election has caused specific visit field on specific each initial position of DRS module and speed and the y axle.Which defective this of access region determines can arrive from the current location of DRS module.Yet in case this module begins towards the target location motion that might be candidate defect, this zone will change and must recomputate.If this module is not static to the x axle when the zoning, then this regional shape will change.
According to the specific embodiment of dispatching algorithm described here, to calculate by suitable visit field, each of DRS module of the present invention is all handled the step of visible candidate defect figure below carrying out in the y axle window of selecting:
(a) structure forward direction mobile (forward flow) figure, it has the node corresponding to candidate defect and current module position, and and the corresponding camber line of feasible motion (arc) from the current location to the candidate defect and between candidate defect;
(b) each camber line for the block motion of expression from a candidate defect to another candidate defect calculates a value factor.This value factor comprises the distance value of the candidate defect of physics to candidate defect, lose the punishment of other defective (because new suitable visit field that reposition produces), and the negative cost (negative cost) (or be worth the benefit of decision by its check) that obtains the target defective;
(c) obtain chart as a result for the minimum cost path (minimum cost path) of end from current location to the y axle window of considering;
(d) calculate the movement locus that needs for corresponding DRS module.
Fig. 4 A shows said process to 4C.Do not change as long as fall into go ahead of the rest this group candidate defect of window of the y axle of being considered, optimum is separated in motion.Yet, appearing in the visual field of this window in case may have the new defective of different value values, it may no longer be optimum that current motion is separated, and should repeat said process and think the related new optimal path of DRS module searching.When calculating new separating, related DRS module may be in the motion of defective.According to new optimum solution, this module can change direction with the position towards different candidate defects.
As a feature of the present invention, the camber line cost is determined as obtaining the suitable function of the negative cost (that is benefit) of target candidate defective, and when total distance of and motion selected when camber line and cost of energy and this skew correlation, the cost of candidate defect is lost.The single factor of in the total cost each is arrived compatible dynamic range by standard, and is subjected to weighting to produce final camber line cost according to the needs of application-specific.
The graph theoretic problem of " shortest path searching " that produces by solve be used for from current module position up to the window in advance that is included in module in the minimum cost path along the last candidate defect of y axle.Behind the last candidate defect in obtaining computation sequence, module keeps the x stationary state, and in another new candidate defect appeared at the window of going ahead of the rest, process repeated in this case.Solution to graph theoretic problem produces by use the algorithm of well setting up in this field.For example, as if the known algorithm that is used for the Dijkstra of shortest path or its increment variant in this field is fit to this task very much, but also can consider other alternative algorithm.
Another feature of the present invention is by the added limitations of suitable visit field is incorporated into the existence of boundary condition and adjacent block in the chart of clearly expressing.All be assigned the scheduling district that waits size that has suitable lap therebetween in the adjacent DRS module of passing same x axle each.The x span subregion that is divided into equal-sized zone is based on the equally distributed hypothesis of defective and makes.
Candidate defect might fall within two overlay regions between the adjacent DRS module, will consider separating of these two modules in this case.If have only a module or do not have module to comprise that specific defects in their optimal scheduling track, then this situation can not cause any conflict condition.All comprise the occasion of this specific defects at two optimal paths, can selection algorithm considering the total optimal result cost that calculates of these two modules, and defective is distributed to minimized that module that causes two total optimal costs.For example, establish m
iAnd m
I+1Be two adjacent blocks, C
* iAnd C
* I+1Be two optimal result costs, then this defective is assigned to the module k that meets the following conditions:
After finding optimal scheduling to separate for all DRS modules, corresponding exercise data obtains calculating and being sent to DRS motion control hardware.This relation is applicable to the various stand combinations that the present invention imagines.
The dynamic auto focusing of DRS camera module
Another feature of the present invention is dynamically to adjust the process that DRS camera module focused on and searched optimum focal point.This measure is to finish in the motion of module head for target candidate defect position, will make an explanation below.
In the whole span of checked flat object greatly, there is random deviation with Pingdu.These deviations are assumed to be and have low spatial frequency (that is slowly take place along the x-y plane).This effect be used for the very narrow depth of field that the needed very high microscope of high-resolution imaging amplifies and combine.Therefore, for each by the reposition of imaging, focusing optical parts again.Maximum for defective is covered, this focusing need be finished at the volley, that is when module just when move the candidate defect position.In DRS module quilt and DDS were installed in embodiment on the same stand, this point especially needed, because the latter has the constant y axle motion by low resolution line sweep imaging requirements regulation.
In one embodiment, the present invention is being by carrying out Image Acquisition completely a series of or part during the moving of candidate defect, to seek to realize the best focus to the candidate defect position.In any embodiment, the distance of dynamic focusing process in the close enough target location starts and finishes, thereby makes the focus that obtains effective in target location.
This process is undertaken by following step:
(a), and during this block motion, obtain a series of images data from DRS camera module from leaving a preset distance of target candidate defective;
(b) by utilizing these a series of images data of extracting from the camera module and being combined in the focus quality measured values that these images calculate, with sampling focus figure-of-merit curve;
(c) utilize these samples of smooth function interpolation, the z stage of thinking the focusing optical parts moves and determines maximum focus;
(d) the z driver is guided to the position that makes focus quality criteria maximum, to obtain focusing on the most clearly on the target area.
Fig. 5 A and 5B show module from candidate defect n
2To n
5Motion, wherein suppose n
5Still suitable under added limitations.Character according to required optics, for example use required line sweep sensor or area scan sensors, (a) image acquisition step in both can be during any diagonal motion of module, and (also in the motion of x direction of principal axis) carries out (Fig. 5 A), also can be only becomes x execution (Fig. 5 B) when static when camera.In back one occasion, the purpose that be to focus on has been considered additional x quiescent period at the exercise end place to the calculating of suitable visit field.
The focusing quality that step (b) is used is estimated can be based on the contrast of image and the highest frequency component in this image, and this composition is by modulation transmissions function (MTF) restriction of employed lens combination.
Candidate defect classification in the DRS motion
By using the high-definition picture that is collected by DRS, this system can be to check of candidate defect execution concurrence and the classification of finding on the examined object body automatically.
DDS produces a candidate defect stream, and their queuings also are scheduled to carry out imaging by a plurality of DRS modules.By the method scheduling DRS module of in front chapters and sections explanation so that outstanding candidate defect is carried out high-resolution imaging.So just produce a candidate defect stream related with high resolution image data.These candidate defects experience two stage processing, comprising:
(a) check automatically; And
(b) classification automatically.
During check (AR) was handled automatically, the high resolving power candidate image was compared with the reference picture (data representation of the reference picture of perhaps more specifically saying so) that is stored in the system storage, and reference picture obtains with same module or disparate modules in advance.This step relate to test and with reference between the compensation of known variant, comprise such as camera sensitivity, sensor pixel change of sensitivity, and the unregulated item in the space on the sensor pixel level is proofreaied and correct.Because the high resolving power of DRS module, so do not need the sub-pixel calibration.
Automatically there are reasonable defective in the result of check processing or confirmation in alternate location, or refuse the defective as " vacation " defective, that is are the artefacts (artifact) of the known limitation of low resolution DDS.Rationally defective is forwarded to automatic sorting phase.
In that automatically classification (AC) is during the stage, high resolving power defect image and the output in AR stage are combined, and being used to extract the correlated characteristic of defective, and make final decision about defect type by the classification processing.May interested single feature depend on specific application, but should comprise flaw size, defective locations, defect shape and signal level.
Dull and stereotyped inspection of the TFT-LCD that considers is to be used under the situation of product defects, and the main output of categorizing system is whether the decision defective is:
A) process defect,
B) but repair-deficiency, perhaps
C) critical defect.
Subclass within these major decisions is then considered according to specific user's needs.
Another one feature of the present invention is the ability in above-mentioned stage of Parallel Implementation (that is concurrent processing).For example, AR and AC stages operating are as the concurrent activity of a part of DRS operation, and DDS also carries out the low-resolution scan on the whole surface of tested object simultaneously.
Abovely explain the present invention with reference to certain embodiments.Other embodiment is obvious for the person of ordinary skill of the art.Therefore except being pointed out by claims, the intention of above-mentioned explanation is not a limitation of the present invention.
Claims (14)
1. an inspection has the method for the flat medium of pattern, comprising:
Utilize relatively low resolution imaging and location protocol to have the flat medium of pattern by the imaging device inspection; Concomitantly simultaneously
Utilization is carried out imaging based on higher relatively resolution imaging and location protocol, so that the defective of being pointed out by above-mentioned detection step is checked.
2. method according to claim 1 is characterized in that, the automatic focus imaging in the motion of described check step use dynamic optimization.
3. defect inspection method carries out simultaneously with check and classification in the motion of candidate defect in the flat object of being tested with pattern, and described method comprises following parallel work-flow:
Obtain the image of object by the defects detection subsystem; Detect the candidate defect of described object; And specify check value value for described candidate defect, and described defects detection subsystem has a plurality of defects detection subsystem module, and described module is operated according to the first relatively low operation resolution; And concomitantly simultaneously
Obtain described candidate defect image all around by the defect review subsystem than the zonule; Check described than the zonule; And classify than the zonule to described, described defect review subsystem has a plurality of defect review subsystem module, and described module is operated according to the second higher relatively operation resolution.
4. method according to claim 3 is characterized in that comprising:
Utilize dynamic defect check subsystem module dispatching algorithm, make the number maximum of the candidate defect of the higher priority that obtains by described defect review subsystem module, and make the distance that moves by described defect review subsystem module minimum, so that the motion in the check of described concurrent high-resolution is optimized.
5. method according to claim 4 is characterized in that, described dispatching algorithm further comprises:
For each defect review subsystem and when each repeat of described motion optimization the suitable motion chart forward on the structural map opinion meaning, described chart has the node corresponding to the current location of each candidate defect and correlation module thereof, and has the camber line corresponding to the suitable motion between described node;
The suitable cost function of selecting the function that is worth according to check from the distance of representing the cost of losing other defect, required motion and target candidate defective, each described camber line is related with cost, obtain chart as a result thus, described camber line has been represented first second the block motion in the described candidate defect from described candidate defect;
Terminal point from the current location of defect review subsystem to a window of considering along the direction of scanning solves chart as a result for searching the minimum cost path, and described minimum cost path is represented to the order through sorting of defective transition by defective; And
For the defect review subsystem module is calculated exercise data, to be used to control the motion of defect review subsystem module.
6. method according to claim 3 is characterized in that comprising:
Begin from a preset distance that leaves the target candidate position, and between the moving period of defect review subsystem module, the automatic focus image-forming component;
At least utilize the sample of the focusing quality tolerance of on image, calculating to obtain a focusing quality tolerance curve;
Use the sample of smooth function interpolation focusing quality tolerance, think that the z stage that is used for mobile focusing optical parts determines maximum focus point; And
The z stage guided to make the maximized z shaft position of described focusing quality tolerance curve, to obtain focusing on the most clearly in the target candidate position.
7. method according to claim 6 is characterized in that, described focus steps comprises:
Obtain a series of images data from the image-forming component of defect review subsystem; And wherein said obtaining step comprises
Use a series of images data and be combined in the described focusing quality measurement result of calculating on the image focusing quality curve is sampled.
8. method according to claim 3 is characterized in that comprising:
In the defects detection subsystem, produce a series of candidate defects;
The order that queuing and arrangement are carried out imaging by a plurality of defect review subsystem module;
Scheduling defect review subsystem module is carried out the imaging of relative high-resolution to outstanding candidate defect, to produce a series of and the related candidate defect of the view data of relative high-resolution;
Make candidate defect experience two stage processing, comprising:
Automatically check is handled; And
Automatically classification is handled;
During automatic review processes, the reference picture of high resolving power candidate image with the known defect state of storing in system storage compared, wherein, described comparison process comprise to the test and with reference between known variant compensate, it comprises revises one of following content at least:
A) imaging device sensitivity,
B) sensor pixel change of sensitivity;
Compensation space is not calibrated on the sensor pixel level, exists the confirmation of reasonable defective or refusal as the defective of Artifact to be created in the position candidate place, comprises the artefact of the known limitation of low resolution DDS;
Handle the information of transmission about reasonable defective for classifying automatically; Afterwards
In classification is handled automatically, be used in combination the output that imaging of relative high-resolution defective and automatic classification are handled, to extract the correlated characteristic of defective;
Handle the final conclusion of making defect type by classification.
9. device that is used to check the flat medium with pattern comprises:
Detection subsystem is used for using relatively low resolution imaging and location protocol to check candidate defect by imaging device;
The check subsystem, it uses higher relatively resolution imaging and location protocol concomitantly the defective of being pointed out by described detection subsystem to be checked.
10. device according to claim 9 is characterized in that, described detection subsystem can also be assigned to described candidate defect with check value value.
11. a device that is used for defects detection, it carries out defect review and classification to the phenomenon in the testee concomitantly at the volley, and described device comprises:
The defects detection subsystem, it has a plurality of defects detection subsystem module, is used for obtaining image, the detection candidate defect of object and specifying check value value for described candidate defect with the first relatively low operation resolution; And
The defect review subsystem, it has a plurality of defect review subsystem module, described module can be obtained the image than the zonule around the candidate defect concomitantly, utilizes relative high-resolution that described candidate defect is checked and described candidate defect is categorized as defective.
12. device according to claim 11 is characterized in that, described defects detection subsystem is installed on the first movable stand, and described defect review subsystem then is installed on the second movable stand.
13. device according to claim 11, it is characterized in that, described defects detection subsystem comprises a plurality of detection modules that are fixedly mounted on the first movable stand, and described defect review subsystem comprises a plurality of defect review subsystem module that are mounted to along the second movable stage motion.
14. device according to claim 13 is characterized in that, the motion of first in the described check module is by second position limit in the described check module, and described device further comprises controller, and described controller can be carried out following operation:
Structure forward-flow cardon, it has the node corresponding to the current location of one of candidate defect and defect review subsystem module, and has and from the current location of defect review subsystem module to first selected candidate defect and the corresponding camber line of feasible motion of second middle selected candidate defect;
Each camber line for the block motion of expression from a candidate defect to another candidate defect, cost is related with camber line with the function as the value factor, the value of the defective that comprises the expense of losing other defective, the distance that must move and obtain, thus chart as a result obtained;
Solve chart as a result, so that the cost route minimum of terminal point from the current location of defect review subsystem module to the y axle window of considering; And
For the defect review subsystem module is calculated exercise data, to be used to control the motion of defect review subsystem module.
Applications Claiming Priority (5)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US42300802P | 2002-11-01 | 2002-11-01 | |
| US60/423,008 | 2002-11-01 | ||
| US10/439,991 US20040086166A1 (en) | 2002-11-01 | 2003-05-16 | Method and apparatus for flat patterned media inspection |
| US10/439,991 | 2003-05-16 | ||
| US10/688,326 US7386161B2 (en) | 2002-11-01 | 2003-10-17 | Method and apparatus for flat patterned media inspection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1536349A true CN1536349A (en) | 2004-10-13 |
| CN100520379C CN100520379C (en) | 2009-07-29 |
Family
ID=34381918
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2003101034401A Expired - Lifetime CN100520379C (en) | 2002-11-01 | 2003-11-03 | Method for checking flat medium with pattern and its equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100520379C (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102221753A (en) * | 2011-06-06 | 2011-10-19 | 深圳市华星光电技术有限公司 | Method and device for detecting pixel array |
| CN103196915A (en) * | 2013-02-26 | 2013-07-10 | 无锡微焦科技有限公司 | Object detection system |
| WO2016192534A1 (en) * | 2015-06-03 | 2016-12-08 | Saint-Gobain Glass France | Optical device for detecting an internal flaw of a transparent substrate and method for the same |
| CN107843599A (en) * | 2017-10-24 | 2018-03-27 | 武汉精测电子集团股份有限公司 | The methods of a kind of panel detection based on AOI is sentenced and device |
| CN108475417A (en) * | 2016-01-04 | 2018-08-31 | 科磊股份有限公司 | It is applied for semiconductor and high-definition picture is generated by low-resolution image |
| CN109959666A (en) * | 2019-04-11 | 2019-07-02 | 京东方科技集团股份有限公司 | A kind of array substrate method of determining defects, processor and decision-making system |
| CN109975321A (en) * | 2019-03-29 | 2019-07-05 | 深圳市派科斯科技有限公司 | A kind of defect inspection method and device for FPC |
| CN110823923A (en) * | 2019-10-15 | 2020-02-21 | 广东炬森智能装备有限公司 | Display screen internal circuit detection device |
| CN111307823A (en) * | 2020-04-13 | 2020-06-19 | 国家电网有限公司 | System and method for detecting typical visual defects of power transformation equipment based on edge cloud cooperation |
| CN112634199A (en) * | 2015-06-26 | 2021-04-09 | 柯达阿拉里斯股份有限公司 | Camera subassembly dust and defect detection system and method |
| CN112733824A (en) * | 2021-04-06 | 2021-04-30 | 中国电力科学研究院有限公司 | Transformer equipment defect diagnosis method and system based on video image intelligent front end |
-
2003
- 2003-11-03 CN CNB2003101034401A patent/CN100520379C/en not_active Expired - Lifetime
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102221753B (en) * | 2011-06-06 | 2013-04-24 | 深圳市华星光电技术有限公司 | Method and device for detecting pixel array |
| CN102221753A (en) * | 2011-06-06 | 2011-10-19 | 深圳市华星光电技术有限公司 | Method and device for detecting pixel array |
| CN103196915A (en) * | 2013-02-26 | 2013-07-10 | 无锡微焦科技有限公司 | Object detection system |
| CN103196915B (en) * | 2013-02-26 | 2015-05-27 | 无锡微焦科技有限公司 | Object detection system |
| US10429317B2 (en) | 2015-06-03 | 2019-10-01 | Saint-Gobain Glass France | Optical device for detecting an internal flaw of a transparent substrate and method for the same |
| WO2016192534A1 (en) * | 2015-06-03 | 2016-12-08 | Saint-Gobain Glass France | Optical device for detecting an internal flaw of a transparent substrate and method for the same |
| CN112634199A (en) * | 2015-06-26 | 2021-04-09 | 柯达阿拉里斯股份有限公司 | Camera subassembly dust and defect detection system and method |
| CN108475417A (en) * | 2016-01-04 | 2018-08-31 | 科磊股份有限公司 | It is applied for semiconductor and high-definition picture is generated by low-resolution image |
| CN108475417B (en) * | 2016-01-04 | 2023-04-04 | 科磊股份有限公司 | System and method for generating high resolution images from low resolution images for semiconductor applications |
| CN107843599A (en) * | 2017-10-24 | 2018-03-27 | 武汉精测电子集团股份有限公司 | The methods of a kind of panel detection based on AOI is sentenced and device |
| CN107843599B (en) * | 2017-10-24 | 2021-07-06 | 武汉精测电子集团股份有限公司 | AOI-based panel detection and judgment method and device |
| CN109975321A (en) * | 2019-03-29 | 2019-07-05 | 深圳市派科斯科技有限公司 | A kind of defect inspection method and device for FPC |
| CN109959666A (en) * | 2019-04-11 | 2019-07-02 | 京东方科技集团股份有限公司 | A kind of array substrate method of determining defects, processor and decision-making system |
| CN109959666B (en) * | 2019-04-11 | 2021-08-03 | 京东方科技集团股份有限公司 | Array substrate defect judgment method, processor and judgment system |
| CN110823923A (en) * | 2019-10-15 | 2020-02-21 | 广东炬森智能装备有限公司 | Display screen internal circuit detection device |
| CN111307823A (en) * | 2020-04-13 | 2020-06-19 | 国家电网有限公司 | System and method for detecting typical visual defects of power transformation equipment based on edge cloud cooperation |
| CN112733824A (en) * | 2021-04-06 | 2021-04-30 | 中国电力科学研究院有限公司 | Transformer equipment defect diagnosis method and system based on video image intelligent front end |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100520379C (en) | 2009-07-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7386161B2 (en) | Method and apparatus for flat patterned media inspection | |
| CN110126467B (en) | Fused ink detection device and method suitable for large-area substrate printing | |
| CN1954204B (en) | Improved inspection method of TFT LCD panels using on-demand automated optical inspection sub-system | |
| US7792352B2 (en) | Method and apparatus for inspecting pattern defects | |
| CN1080911C (en) | Object observing method and device | |
| CN1221926C (en) | Method and device for examining pre-determined area of printed circuit board | |
| KR100863700B1 (en) | Vision inspection system and method for inspecting workpiece using the same | |
| CN100520379C (en) | Method for checking flat medium with pattern and its equipment | |
| CN114113114B (en) | Automatic process method for detecting and repairing micro defects on surface of large-caliber element | |
| CN1928535A (en) | Machine vision based LCD spot flaw detection method and system | |
| JP2010102050A (en) | Defect correction device and defect correction method | |
| US20210035305A1 (en) | Image-based classification system | |
| US8810646B2 (en) | Focus offset contamination inspection | |
| JP2010256338A (en) | Imaging inspection device and imaging inspection method | |
| JP2003085558A (en) | Method of pattern inspection and inspection device | |
| JP2004354250A (en) | Defect inspection device | |
| CN209992407U (en) | Large-caliber ultra-clean smooth surface defect detection device combined with linear array camera | |
| CN1518085B (en) | High-speed in-line electro-optics testing method and system for defects on chip | |
| JP2005315792A (en) | Defect inspecting/classifying apparatus | |
| JP4074624B2 (en) | Pattern inspection method | |
| JP4662424B2 (en) | Glass substrate inspection method and inspection apparatus, and display panel manufacturing method | |
| US20110270580A1 (en) | Method, apparatus, and computer program product for focus prediction | |
| JP2000097869A (en) | Defect inspection method for pattern and its device | |
| JPH0979946A (en) | Inspection device for display device | |
| CN1947069A (en) | Exposure apparatus |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210419 Address after: Yavnei, Israel Patentee after: ORBOTECH Ltd. Address before: California, USA Patentee before: PHOTON DYNAMICS, Inc. |
|
| CX01 | Expiry of patent term | ||
| CX01 | Expiry of patent term |
Granted publication date: 20090729 |