TWI273216B - Method for inspecting patterns - Google Patents

Abstract

A method for the inspection of electrical circuit patterns including carrying out an initial inspection of a sequentially acquired image of an electrical circuit pattern to determine potential defects in the electrical circuit pattern; upon identifying a potential defect in the electrical circuit pattern in the course of the initial inspection, interrupting the initial inspection and carrying out a secondary evaluation of a portion of the sequentially acquired image including the potential defect; and following completion of the secondary inspection, resuming the initial inspection.

Description

1273216, (1) 玖, invention description (the description of the invention should be stated: the technical field, prior art, content, implementation and schematic description of the invention) Related patent cross-references This application claims to be in February 2001 3 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and system for pattern inspection that is particularly useful for circuit inspection during manufacturing. Prior Art Circuit fabrication, such as printed circuit boards, typically involves one or more stages during which a conductor pattern is deposited in a printed circuit board to form one or more substrate layers. At least some of the substrate layers of the substrate layers are, for example, V-3 0 0TM available from 〇rbotech L td., Inc., Yavne, Israel.

Automated optical inspection (A0I) of the Inspire 9 0 60TM, SK-75TM* ICP8000TM AOI system. A variety of methods are used in Α , I to optically inspect the defects of the circuit pattern. Some methods include checking a bitwise comparison of a circuit image of a reference image. Other methods include an inspection circuit analysis to identify the type and location of the various components that form a circuit to determine whether all components are provided and properly placed, and to measure various characteristics, such as conductor width and space between conductors; And decide whether these meet the predetermined design specifications. SUMMARY OF THE INVENTION In general, the present invention is directed to providing a pattern inspection system and method for facilitating the use of software that operates on multiple purpose hardware for inspection of patterns, and

1273216 (2) Check the circuit pattern. The general idea of the present invention relates to a pattern inspection system operating using a first algorithmic group to compare the selected portion of the input data with a reference and to quickly filter out portions of the input data that closely resemble the reference. The input image data stream is evaluated. When the system encounters a portion of the input data that is not very similar to the reference, the evaluation of the input data is interrupted, in order to use a second algorithm group to further evaluate the input data that is not very similar to this part of the reference. When the evaluation is further performed, the continuously obtained input image data stream is temporarily stored in the memory. As long as the second algorithm group is used to complete the further evaluation is not very similar to the reference part, the system will return and use the first algorithm group to evaluate the input data stream, first evaluate the unprocessed part of the memory collection until I encountered another part of the input that was not very similar to the reference. According to an embodiment of the invention, the second algorithm group is using more resources than the first algorithm group. The second algorithm group is preferably better than the first algorithm group to determine whether a portion that is not very similar to the reference is still sufficiently similar, so it should not be considered defective. According to an embodiment of the invention, the first algorithm set registers a check pattern image as a reference using a general registration procedure, and then a first set of attributes in the structure and position of the image, for example, in the reference The corresponding attributes in the comparison. Thereafter, the second algorithm group will not be very similar to each part of the input data of the reference, so it can be accurately registered with a corresponding part in the reference. After micro-registration, the system will again compare the first set of attributes in the part with the corresponding attributes in the reference.

1273216 0) Comparison. Alternatively, in accordance with an embodiment of the present invention, the first set of algorithms may evaluate a first set of attributes or characteristics, such as a contour position; and the second set of algorithms may evaluate a second set of attributes or characteristics, wherein the second set A group is a computing resource that requires a larger and/or more than the first group, such as one or more statistical opportunities that represent a profile. Another general idea of the present invention is a system for comparing an inspection pattern image with a reference, wherein it is known that the inspection pattern image should resemble the reference _, and some differences between the image and the reference are Indicates a defect in the pattern. A normal image portion of the continuous stream is provided to a processor. As long as a first characteristic is compared, a very similar reference image is found, that is, the portion of the image that meets or exceeds a high quality threshold is not further considered. However, the image portion of the inspection pattern that does not meet the high quality threshold is then further evaluated using a different evaluation method. Different evaluation methods may use a further step of precise alignment between the image and the reference, or consider additional and/or different characteristics presented in the evaluation portion. According to some embodiments of the present invention, after a portion of the image is found to be inconsistent with the high quality threshold, and before all of the image portions are compared with the reference, the high quality threshold can be used for evaluation, which can be performed quickly. . Moreover, in accordance with some embodiments of the present invention, further evaluation may be performed by the same processor to compare the image to the reference and to evaluate the comparison portion using a high quality threshold. An additional general aspect of the present invention is related to methods of inspecting circuits using the above described systems, and such methods can be used to fabricate circuits, including:

1273216 (4) A conductor member is formed in a predetermined pattern on a circuit substrate; the system and method described herein are used to optically inspect the pattern, and then the substrate in which the defect is found is discarded or fixed. Embodiments Referring now to Figures 1 and 2, there is illustrated the construction and operation of a pattern inspection system 10 constructed and arranged in accordance with a preferred embodiment of the present invention. The system 10 includes a detector module 12, which is preferably a portion of the scanner, and operates to continuously obtain a series of image portions 14 of the inspection pattern 16. For example, pattern 16 includes a conductor member 18 pattern formed on surface 20 of circuit substrate 22. The detector module 12 operates as a line scanner to continuously obtain 14 image portions in a column-by-column manner; or, like the same region imager, to successively obtain a second-degree spatial region of each portion corresponding pattern 16 Image portion® As used herein, the term "circuit" means any suitable circuit, including (but not limited to) a printed circuit board, display screen, integrated circuit, multi-chip module, ball grid array substrate, A device that connects between a printed circuit board and an electronic component, or any other fully or partially formed electrical or electronic circuit. It should be understood that although the following invention is described herein in the context of a printed circuit board, it can be used for inspection or detection of any suitable circuit, article or pattern. A series of image portions 14 are provided by the detector 1 2 to an image processor unit 2 6. The image processor unit 26 preferably includes an image portion comparison function and a defect determination function. The image processor unit 26 also receives the reference image portion 24, wherein the reference image portion 24 is from a conventional CAM device (not

12732Μ is shown in the figure) output. Image processor unit 26 is preferably a commercial processor such as a SPARC (R) processor from Sun Microsystems. The function of the image processor unit 26 will now be described. A feature of the invention is that the operation of the image portion comparison function begins by examining a plurality of locations on the pattern 16 in order to find a first plurality of locations having a first defect likelihood among the plurality of locations. This initial inspection is performed, for example, by a first algorithm set at a plurality of locations representing a transition profile between different regions of the pattern, such as a transition between the conductor and the substrate. The inventors found in the circuit inspection that the outline representation of the circuit image is particularly suitable for defect inspection because they contain enough information to represent the geometry and location of the components defining the circuit. This information is an easy-to-handle format and is substantially less informative than the corresponding bit-mapped image. It should be understood that a typical inspection image of a particular circuit board includes a plurality of image portions 14 as depicted in FIG. It should also be understood that a plurality of locations are typically included in many locations in each image portion 14. Moreover, it should be understood that the operation of the present invention preferably checks a plurality of circuit patterns 16 . In the particular embodiment depicted in Figure 1, although it is clear that this is not the case, each of the plurality of locations corresponds to a single image portion 14. A suitable first algorithm, or a first set of algorithms, in accordance with an embodiment of the present invention includes a comparison algorithm 24 for comparing an image portion 14 with a corresponding reference image portion, and further considering, for example, The at least nearly identical image portions 14 and the corresponding reference image portion 24 are discarded by a high quality critical measurement. At least nearly the same image and reference part 1 4

1273216 (6) and 24 are overlapped descriptions and are indicated by reference numeral 30. As shown in Fig. 1, image portion 14 is preferably each contoured representation corresponding to a transition position between conductor member 18 and surface 20 of substrate 22. It can be seen that the plurality of image portions 14 and their corresponding reference image portions 24 are not at least nearly identical. The parent image portion 14 will present at least a portion of the outline, and the reduced portion of the contour is different from the corresponding contour portion of a corresponding reference image portion 24, that is, generally does not overlap. The three pairs of non-identical corresponding image portions depicted in Figure 1 overlap with the reference portion and are referenced by reference numerals 3 2 , 3 4 and 3 6 , respectively. It should be understood that although the system 1 is interspersed in the preferred text for checking the outline representation of the representation, any other suitable pattern representation such as a one-bit map can be used. Contour representations may be used in the preferred embodiment of the present invention because the contours constitute appropriate descriptors for pattern 16 to aid in the occurrence of, for example, scoring, ridges, open and short circuits of typical circuits. ^ ' A special feature of the present invention. When performing the step of initially comparing the image portion i 4 with reference to the portion 24, it is only necessary to identify, for example, the pair 32 being at least a pair of the same image and the reference portion, initially comparing The second stage evaluation of the reference number 4〇 is executed. The purpose of the review is to determine that the first plurality of positions constitute a second position, and the probability of a second defect is greater than the probability that the sign is greater than the first defect. The location includes zero, - or a plurality of locations with a second defect. The comment 40 is preferably a fourth set of algorithms that are the same as the group algorithm, and the typical 2 is more accurate and/or sound than the first set of calculations, so that it can be decided to add the shape after the first cup. A number of positions of 1273216 - (7) those locations are more likely to actually correspond to a defect. Typically, the second set of algorithms typically requires more computer resources and/or consumes more time than the first set of algorithms. It should be understood that although only two algorithms are shown in Figure 1, additional processing stages may be used each using a continuous more accurate, and/or robust algorithm, and/or more dense resource algorithms. Particular attention is paid to the timing obtained for a series of image portions 14 and the initial comparison between the corresponding reference portion 24 and the second stage evaluation 40. It can be seen that the initial comparison of each image portion takes less time than the time slot display as shown by reference numeral 42. At the same time, it can be seen that as soon as the second stage evaluation starts, the initial comparison of the image portion 14 is temporarily interrupted, and is continuously obtained at the same speed as before. Usually, as indicated by arrow 44, since time slot 42 can be dispensed with, as long as a second stage evaluation is completed, the accumulated unprocessed work of image portion 14 is processed at the same speed as before, but at a faster rate. rate. As soon as the unprocessed work is waived, the initial comparison rate will return to the original rate, including time slot 42. It will be appreciated that the above timing can be equally applied to any other suitable embodiment in which a plurality of circuit patterns, image portions, or boards of corresponding locations perform an initial comparison and are evaluated as an appropriate second stage. Thus, as shown in FIG. 1, and as further understood from FIG. 2, image portions 14 representing a plurality of locations are continuously acquired and provided to image processor unit 26 in a similar stream. Preferably, the operation of the first algorithmic group quickly discards the image portion 14 that is similar in height to the corresponding reference image portion 24. Only when a pair of non-identical corresponding images and reference parts such as pairs 3, 3 4 and 3 6 will encounter an image portion comparison circuit whether or not to use the second stage 1273216

(8) w method, 'and perform a further evaluation of 4 Ο, where the corresponding image and reference 4 injuries are not nearly the same. Although the operation of the unit 26 for processing the second algorithm group is interrupted from the ratio of the obtained image portion 14 to the reference image portion 24, the image portion 14 continues. It is taken from the debt detector 1 2 and stored, for example, in a buffer (not shown), thereby forming a temporary unprocessed work 44 that is initially part of the jersey. In accordance with an embodiment of the present invention, it will be appreciated that the above-described inspection sequence can be facilitated by a faster first algorithm than the algorithm used in the evaluation. At its speed, the algorithm group can effectively filter out a large number of non-defective portions of the pattern 16 from the previous one. In the example shown above, the non-defective portion of pattern 16 is the two types of portions that are located at approximately the same position as the portion of the pattern corresponding to the reference. In the above example, the first algorithm group operates on the pattern 16 at a rate that is at least the same multiple or faster than the rate at which the pattern image is taken and in accordance with an embodiment of the present invention. The first algorithm group compares the similarity of the pattern 16 with the reference according to a first simplified standard set, and uses it to determine whether the similarity meets or exceeds a very high quality threshold. This can be that all or some of the actual defects can be identified. The possible disadvantage of the algorithm group is that the error indicates that a considerable amount of error defects will be generated. The shirt is like a partial defect. When further evaluation, it may be found that there are no defects. They present some different facts. Even when compared with the corresponding reference, a specific implementation algorithm group according to the present invention is typical than the image portion. The second 14 used in the evaluation of 40 t is larger than the corresponding reference portion 24 '12-1273216 (9) and the processing time and/or processing resources are larger than the first algorithm group used. According to some embodiments of the present invention, the second set of algorithms used in the evaluation 40 only temporarily stores the image portion 14 in a more precise manner associated with the corresponding reference portion 24; Compare them to determine if they are equally acceptable. In accordance with an embodiment of the present invention, evaluation 40 is performed after an entire image portion 14 and the corresponding reference image portion 24 is processed using the first algorithm group. Note, however, that the first algorithm set is used to perform the initial check of the evaluation 40 - the interrupt can occur in a variety of other sequences. For example, the operation of the first algorithm group can be immediately interrupted as long as any position found in the image portion 14 different from a reference portion 24 is found, for example, as seen at any position indicated by reference numeral 48. Then, at position 4, the pattern portion, the pattern in the vicinity of the close vicinity corresponding to the different portions except the image portion 14 and the reference portion 24 is isolated, temporarily stored in a precise manner, and then Use the second algorithm group to compare to determine if they can be accepted or different and not accepted. According to other embodiments of the present invention, in order to determine whether a local region of the pattern 16 is found to be defective in the first algorithm group, the second algorithm group preferably uses the standard additional inspection standard used by the first algorithm. And/or different from the standard used in the first algorithm. Although a partial region can be considered to be defective by using the simplified criteria in the first algorithm, further analysis using additional and/or different criteria in the second algorithm group can indicate that the local region of the evaluation is No defects. 1273216 (10) In accordance with a preferred embodiment of the present invention, for example, the first comparison between the first algorithm set and the second stage evaluation 40 is preferably performed in an interleaved manner on the same processor. It should be understood that although the process shown in Figure 1 is performed on a first in, first out basis, each image portion 14 can be evaluated by a first algorithm and then, if desired, an entire different image portion is evaluated. A second algorithm group of 40 evaluates, but some changes to the above methods can be implemented prior to processing an initial comparison of a subsequent image portion 14. Thus, in another embodiment of the invention, an image stream is available for storage in a first in first out manner. As long as one or more completed image portions are collected, a complete image portion can be processed using the first algorithm group. In another mode of operation, the processing using the first set of algorithms can be interrupted immediately, in order to evaluate any different locations in the evaluation 40 4 8 whenever an interrupt is encountered. In another mode of operation, only the different positions 4 8 are in the evaluation 40 processing; however, all positions 4 8 are encountered during processing using an image portion 14 of the first algorithm group, and only the position 4 8 Can be saved. Any storage location 4 8 will then be processed in evaluation 40 before processing a subsequent image portion 14. It is noted that a feature of the present invention is the use of at least two different algorithms, for example, a first algorithm group and a second algorithm group, each algorithm group having different resource and/or time requirements on the same processor, and preferably It is in adjacent time slots. It is a further feature of the present invention that at least two different algorithms can be used to process an image in an interleaved manner, and one of the algorithms uses less emphasis on computing resources than another algorithm. Less emphasis on resource algorithms identifies portions of the image that are further processed by more resource algorithms. When using more -14-

TOUO 1 A οι) When multiple resources emphasize the algorithm, fewer resources emphasize that the operation of the algorithm can be interrupted, and the unprocessed work of the image portion can be established by the less resource-intensive algorithm. This time multiplex can be achieved by a first algorithm that can operate at a rate that is faster than the portion of the image that inspects the pattern. Moreover, by interleaving the foregoing initial comparison with a subsequent second stage evaluation, the total comparison of the cumulative elapsed time of the corresponding image and reference portion stages and the cumulative elapsed time of subsequent execution of the various second stage evaluation processes does not significantly exceed Get the total time required to view the image. Thus, as previously described, in the context of circuit inspection, the same processor can analyze the image using at least two different algorithms to determine the occurrence of defects in the circuit for approximately the same period of time required to scan an entire inspection circuit. Please refer to FIG. 3, which is a flow chart showing the initial inspection function of FIG. 1 according to an embodiment of the present invention, and with reference to FIGS. 4A-4F, which are described in accordance with a preferred embodiment of the present invention. A simplified diagram of the results of the steps of Figure 3. The initial inspection function is to obtain an image of the inspection pattern portion (step 50), such as a circuit pattern. The image can be an entire or partial inspection pattern. Figure 4A shows the image portion 52 of a portion of the circuit. The image portion 52 is an enlarged image portion 14 of position 5 C in Fig. 1. The image portion 52 is an image showing a portion of the conductor member 18. A designated conductor image 504, a portion of the surface 20 opposite the background, and a designated surface image 56. A feature of an embodiment of the present invention is the use of an inspection pattern representation that is preferably in a compressed form and that contains informational descriptions of the image characteristics, in order to examine Figure -15-1273216 (12). The inventors have found that the profile is a suitable property descriptor for an inspection circuit, and in the context of circuit optical inspection, the profile contains the necessary information for optical inspection circuit defects. The outline of a circuit image is indicative of a transition between the image portions representing the conductor, such as conductor 18 of Figure 1, and the image portion is representative of a surface, such as surface 20. At step 60, the pattern property descriptor is retrieved from the image portion of the inspection to form a pattern representation suitable for automatic computer inspection. Figure 4B is a contour 62 showing the appropriate pattern characteristic descriptors in the outline representation 64 of an image portion 52. As shown in Figure-4B, contour 62 is a finite line defining the edge of conductor image 54. It will be appreciated that in, for example, a digital image, the outline may be the entire pixel representing the edge of the conductor image 54. Alternatively, contour 62 may be a sub-pixel size line rust representation of the edge of conductor image 54 known as CELs or profile elements, or any other suitable representation of transitions between different pattern portions, such as conductor 18 in circuit 16. Between the surface 20. At step 70, a reference to the corresponding property descriptor is retrieved. The reference is a feature descriptor structure known as a non-defective pattern compared to a defect check of a check pattern. Figure 4C is a set of reference profiles 72 depicted in a reference contour portion 74. The reference profile portion 72 contains the rims expected to be in the outline representation 64 of the image portion 52 of a non-defective circuit. In a preferred embodiment of the invention, image portion 52 is available and processed in an online manner that is generally simultaneous with image acquisition. When an inspection circuit is obtained, its image is maintained by dynamic registration and has a corresponding reference using, for example, the method generally described in U.S. Patent No. 5,459,535, the entire disclosure of which is incorporated by reference. Incorporated herein. In accordance with a specific embodiment of the present invention, -16-1273216 (13) reference portion 74 is, for example, traversed from a CAM image that is known to have no defects, is used for inspection, and is used for offline processing. ready. The dimension surname + — is maintained in the outline representation of the arrangement between the 64 and the parameters in order to ^ ^ to the hip to ensure that an appropriate correspondence is selected for each contour representation of the processing. In step 80, the check pattern and the reference pattern are overlapped with each other to help the ratio between the two representations (the solid line) and the reference contour, The brothers used 幵y to form an overlapping image 8 4. In accordance with the present invention, step 80 includes an ith embodiment of the present invention in which the contour table is not precisely aligned with the reference contour portion by a micro-registered one. The micro-registration process is a detailed description. In step 90, the difference between a check pattern and a corresponding reference descriptor can be achieved by using a very high similarity: the feature descriptors similar to the reference corresponding descriptors represent a non-defective pattern portion. And can be discarded from 0

Figure 4 is a representation of a profile 62 in a superimposed representation of a threshold value 91 in accordance with a preferred embodiment of the present invention. In an overlapping representation with a similar threshold value of 9, the measurement is represented by a quality threshold indicator 92 (dashed line). A preferred embodiment of the similar evaluation is based on the spatial position of the weight corresponding contour 62 and the reference contour 72. Thus - the circuit image, or take the reference contour and the test wheel 靡 indicates that the 7 7 can be used to detect defects corresponding to the pattern description, and different recognition. Part of Embodiment 74 (Dash Line) Step 8 0 includes 7 steps to be evaluated in the amount of overlap characteristic of Figure 5-6 below. A part of the high is further evaluated and is described between the class profile 7 2 and a constant height similar to the image in the superimposed image 8 4 according to the present invention.

I2732164) The reference character 9 2 is a threshold value indicating the difference between the contour 6 2 segment which is quite similar to the reference contour 7 2 and the different contour 62 segment which is considered to be defective by the second evaluation. All segments of the rim 6 2 between the reference contour 72 and a threshold indicator 92 at representation 91 are considered to be similar to the reference segment 7 of the reference profile 7 and need no further evaluation. The segment of the contour 62 outside the Converse indicator 92 is a pair of segments that are considered to be similar to the reference profile 72, thus indicating a possible defect requiring a second evaluation. The distance from the quality threshold indicator 92 of the reference profile 72 can be adjusted, for example, by setting a - check sensitivity parameter so that a greater or lesser sensitivity of the suspected defect can be found in the beginning of the inspection phase. In a preferred embodiment of the invention, it is desirable to set thresholds 96 and 98 and 96 and 98 very close to segment 72. Making the distance from the threshold 92 of the reference profile 72 smaller increases the sensitivity of detecting suspected defects. As a result, a fairly high rate of detection errors clearly defects, and a relatively low rate misses actual defects. Suspicious defects are then further evaluated in the second evaluation phase 40 (Figure 1B). The appropriate distance selection from the critical value 9 2 of the reference profile 7 2 is a design consideration to, for example, achieve the ability to detect sensitivity and the ability of the second evaluation phase 4 to process and evaluate a smaller or larger amount. Suspicious defects. The four segments of the profile 62 specifying reference numerals 93, 94, 96, 98 and 99 are shown in Figure 4E, which shows a visually discernible relative difference of the corresponding reference profile 72. An allowable segment 93 is different from the corresponding segment at the reference contour 72. However, it is entirely located within a bounded region of a pair of thresholds 92. Each of the very different segments 9 4 , 9 6 , and 9 8 is at least partially expanded outside of the region bounded by the relative pair of thresholds 92 . -18- l2732l6 (15) ^Step 1 0 0, different from the corresponding reference contour 7 2 is larger than the allowable 43⁄4 sinking, the 4th segment can be captured, and then supplied to the second evaluation 4〇 (Figure 1 盥 Reference 4, • All descriptor fragments, such as fragment 9 1 that are different or different from the permutations defined by very high similar measurements, can be discarded from further evaluation. It should be understood that a descriptor fragment of the second +, outer estimate is extracted. Sufficient space is required, so it can be properly evaluated in the evaluation process. ® 4F is described in the steps 丨〇〇, according to a specific embodiment of the present invention, the pair of very different segments 104, 106 and 108, and the weighted +, reference - contour 72. Other snippets including other segments that allow different segments 93 are all discarded and are not further processed. Note that each pair of 104 and Only the portion of the contour 62 that is actually outside the threshold value 92 is included. Preferably, the pairs i 〇 4 and 1 〇 6 include at least a portion of the contour 6 2 ' and the contour 62 is at least partially located at a Outside the area between the critical value 92 and the reference contour 72 In addition, the pairs 1 〇 4 and 1 0 6 are bounded by a portion of the contour 62 at either end until a position where it intersects the reference contour 72. It should be understood that other suitable methods for selecting a segment capture may be used. For example, it can be seen from the different segments 1 0 8 that a suitable segment is selected as an arbitrary distance function for delimiting any partial segment 6 2 to expand to a critical value. The area between the 92 and a reference contour 72 is external to the second image. The appropriate image portion selection type is not a limitation and any other suitable selection of the image portion or segment is provided to the second inspection. Therefore, via an additional example Whenever a very different segment is encountered, a representation containing two adjacent contour segments can be selected and provided to the second evaluation-19-

1273216 Display. However, the arrangement points are preferably as small as possible. Fig. 6B generally shows the overlapping arrangement after the general arrangement, but outside the microarray of the reference 1 1 4 and an inspection image portion 1 1 7 . The process described in Fig. 5 is more precisely arranged than the reference portion Π 4 and the check portion i 1 7 than the display of Fig. 6B. In the example seen in Figure 6A, all of the alignment points 11 2 are placed along the contour 1 1 8 . However, the alignment points can be assigned to the reference in any suitable manner, where the corresponding points of the inspection image can be found. In various embodiments of the present invention, the arrangement point 112 may be selected or assigned in an offline process during a reference preparation period. The arrangement points are chosen such that the combination of alignment points has a uniform chromaticity uniform direction distribution desired, such as found in a chromaticity along a circumference. It should be understood that the amount of alignment points selected is a system parameter determined, for example, empirically. A large number of alignment points can improve the accuracy of the alignment; however, a large amount of computational resources need to process a relatively large number of alignment points compared to a relatively small number of alignment points. In step 120 of matching positions, the designated reference numeral 122 of Figure 6B can be used to find in the portion of the inspection image of at least some of the arrangement points 1 1 2 of the reference 1 1 4 . In accordance with an embodiment of the present invention, a matching position 1 22 is found by expanding a radiation 1 24 from each of the arrangement points 1 1 2 in an arrow 1 16 direction associated with each of the arrangement points. Thus, the matching position 1 2 2 of an arrangement point is the intersection of a radiation 124 and a contour 126 located in the inspection image portion. Typically, not all matching locations are suitable for use in a micro-registration process. Therefore, in accordance with an embodiment of the present invention, in step 130, a portion of the arrangement point 1 1 2 with a good match and a matching position 1 2 2 is optional -21 - 1273216

(19) or parameter group. In accordance with a preferred embodiment of the present invention, the calculation of the revolution may reduce the least squared distance between the pair of alignments of each of the desired parameters and the matching pair of matching locations. At 歩骒170, the conversion parameter is applied to the inspection image portion η?, and the inspection image portion i 17 is enlarged (or reduced), rotated, and converted according to the conversion parameters calculated at step 15 。. After the conversion parameter application of step 丨7 ’, the micro-registration process will wrap back to the step! Further, it is repeated until the proper alignment of the reference and image obtained as shown in the example of Fig. 4D is to be understood that in each iteration of the micro-registration process, the corresponding position of a new group can be identified and selected with a good match. Referring to Figure 7, it is a simplified flow chart of a process 200 for performing suspicious defect evaluation, which is a very unacceptable bottle. It is preferable to use the second aspect of the evaluation shown in FIG. 1B to evaluate the present invention. The preferred embodiment of the present invention is as follows: μ 21 — — 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查 检查'Equipped with the same corresponding reference processing. - Suspicious defects and stereotypes, the paired pattern property descriptor example is shown in Fig. 4F, which is a pair of 106 and 108 corresponding to the reference corridor 72. J θ Only 1U4 Guangkang: a preferred embodiment of the invention, in step 22. A suspicious map, the hold descriptor is the A# I call registration with the corresponding reference local 檄, the step W 主The book processing is similar to that shown in Figure 5_6b, which is the same as that of the micro-volume. The book is processed, but the trapped figure is not different in that it is only in the - suspicious pattern feature descriptor and its corresponding reference to the entire image # & P 4 is bound, not on a ° ° micro-registration is to make ^ Figure 5 to use the towel -23-

ι2Ί32% The same arrangement point. Alternatively, a new set of alignment points may provide alignment points that are consistent or assigned, for example, in a reference. After a pattern characteristic descriptor of a suspect defect and a corresponding reference micro-registration, in step 203, the difference between the pattern characteristic descriptor and the reference is evaluated in relation to a predetermined similar measurement. In some embodiments of the invention, the pattern property descriptor is an outline of the portion of the inspection image, and a similar measurement is a geometric distance between the contour segment associated with a suspect defect and its corresponding reference portion. . When used in step 90 (Fig. 3), similar measurements can be similar to the same measurement and represented by quality threshold indicator 92 (Fig. 4E). Alternatively, a larger or smaller similar measurement can be used; alternatively, several similar measurements can be used according to predetermined logic. Finally, in step 240, the pattern characteristic descriptors will be reported as actual defects from the difference in their corresponding reference presentations that are greater than the predetermined similar measurements. Actual defects are included in a defect report. Alternatively, the actual defect can be further processed to classify the defect or identify a portion of the actual defect in an offline or post-processing. In addition, the second stage evaluation is used as a second stage of filtering, and in addition to evaluating 4 ,, a third set of such algorithms that form a further evaluation level is only applied to these pattern characterizations that specify actual defects. Symbol to identify a portion of actual defects.凊 参考 参考 参考 参考 参考 参考 参考 参考 , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , An evaluation process 2〇0 simplified diagram. In Fig. 8A, a second pair of concatenations comprising a non-hanging different segment 1 〇4 of the evaluation segment 262 and a reference segment 264 is shown, as well as the high-24-1273216 (21)::> threshold used in the processing of Figure 3. The indicator 92" means that the segment 262 is partially extended outside the region bounded by the indication Y 92, and it is such a suspicious defect. The steps \ segment 262 and the reference 264 are the local micro-registration. The result of the local micro-registration associated with reference 264 is shown in Figure 8B: Please display the allowable distance indicator 2" used in step 23" to: estimate whether segment 262 is sufficiently similar to reference 2", so that the reservation is met Similar to the measurement. Note that after local micro-registration, the fragment plus partial expansion is external to the area bounded by the allowable distance indicator 266 at position 268. Because the slice & 262 is a partial extension determined by the allowable distance indicator 266 Outside the bounding area, so its evaluation in step 230 does not meet the predetermined similar measurements, so in step 240 (Fig. 5), the segment 262 will be reported as an actual defect. In Figure 9A, the display includes an evaluation segment. 252 and one ginseng A first pair of very different segments 1 & 6 of segment 2 > 54, along with the UI quality threshold indicator 92 processed in Figure 3. Note that segment 252 is partially extended outside of the region bounded by indicator 92, such Is a suspicious defect. In step 22, segment 252 and reference 254 are local micro-registrations. The conversion and rotation of segment 252 is the local micro-registration result associated with reference 254 is shown in Figure 9b. Figure 9B also shows the use in step 23 The allowable distance indicator 256 is used to evaluate whether the segment 2 52 is sufficiently similar to the reference 254, thus conforming to the predetermined similar measure $. Note that the allowable distance indicator 256 is located at a reference segment 254 that is further from the indicator 92 to indicate After local micro-registration, a large degree of difference is allowed between segment 252 and its corresponding reference, without segment 254 being treated as an actual defect representation. After local micro-registration, segment 2 5 2 is still The upper portion of position 2 5 8 is expanded outside the area bounded by the high quality threshold indicator 92 -25-1273216 (22). Also note that the segment 2 5 2 is the entire bit delimited by the allowable distance indicator 2 5 6 In the region, since the segment 2 5 2 is the entire bit within the region bounded by the allowable distance indicator 256, in step 230 it can be evaluated as conforming to a predetermined similar measurement, therefore, in step 2404, the segment 252 does not In a practical defect report, in Fig. 10A, a first pair of very different segments 108 including an evaluation segment 272 and a reference segment 274 is shown, along with the high quality threshold indicator 92 used in Fig. 3. Note that Although similar to the shape of reference 274, segment 272 is set apart from reference 274. In accordance with an embodiment of the present invention, an evaluation segment evaluates whether one of the measurements resembling a reference is a distance, wherein it is converted in a micro-registration process. Since segment 272 is at least partially external to the area bounded by indicator 92, it is a suspicious defect. In step 2 2 0, segment 2 7 2 and reference 274 are local micro-registrations. The side transition of segment 272 is a partial micro-registration result associated with reference 274 which is shown in Figure 10B. Figure 10B also shows a high quality threshold indicator 92. It can be seen that since segment 272 is the entire bit inside indicator 92, its shape is height similar to reference 274; however, it has converted the distance represented by an arrow 2 7 6 . According to an embodiment of the invention, if the converted distance exceeds a predetermined typical parameter value, then the segment is considered to be an actual defect, even though its shape is nearly identical to the reference. Although formed correctly, this conversion means that a conductor represented by the segment is not located at the position expected to be in the circuit pattern. It should be understood that this threshold can be applied; in addition, other points of view such as rotation or scaled micro-registration can be applied. -26-

1273216 (23) Please refer to the figure, which is a simplified flowchart of a preferred evaluation method 300 for performing an evaluation step in the process of FIG. The evaluation begins at step 3 1 0 and receives an evaluation segment and a reference segment in the micro-registration arrangement, along with data representing one or more of the following transition ranges: X-conversion, Υ conversion, rotation proportional, or any Others want to convert the data. In a particular embodiment of the invention, at step 320, the conversion data can be evaluated to ensure that the conversion exceeds one or more threshold values associated with, for example, conversion, rotation, or a proportional decision. The evaluation is based on weighted values, such as more or more, than the rotation or proportional decision. In addition, this evaluation includes the sum of several conversion parameters. If the result of the conversion assessment exceeds a threshold, the actual defect is reported. If the conversion is deemed acceptable, then in step 303, the segment may be further evaluated to determine if it is within the allowed distance from the reference, for example, as allowed distance indicators 25 6 (Fig. 8) and 276 (Fig. 9) The representation of ). If the clip is partially outside the allowable distance from the reference, an actual defect will be reported. However, if the segment is the entire bit within the allowed distance, then it can be considered acceptable for a similar reference. It can be discarded from further evaluation, and the other inspection steps are performed on the other part of the inspection pattern image. Note that the relative evaluation of each of these steps 3 2 0 and 3 3 0 occurs in the reverse order. Moreover, it is noted that in the context of circuit inspection, each of these steps 3 20 and 3 30 can effectively analyze different types of defects. Thus, for example, the evaluation of step 320 can effectively detect defects where features such as padding and conductor ends in a circuit are rendered and shaped in the correct geometry, however they are placed in an incorrect position. It can also effectively evaluate whether the width of a -27-1273216 (24) body is too wide or too narrow. The actual position of a feature can be close to or away from an absolute position as indicated by an inspection parameter. The actual width of the conductor can be close to or away from the desired width as shown by an inspection parameter. This incorrect position or width may be the result of pattern exposure and etch processing offsets and other factors. On the other hand, the evaluation at step 303 can effectively detect defects where, for example, the features of the fill, conductor and conductor ends are typically placed at the desired location; however, their geometry is incorrectly formed. An example of an incorrect geometry includes a notch and a convex along the length of the conductor formed by a generally smooth edge. According to another embodiment of the present invention, the evaluation 40 (Fig. 1B) uses a different process to analyze and evaluate very different segments. A very different contour segment is, for example, an abstract representation of a polygon, and the abstract representation is compared to an abstract representation of a corresponding reference to detect actual defects. One method of abstracting a contour segment is to treat it as a polygon, where the shape of the polygon is a variety of different features and characteristics derived from a contour segment. Appropriate features and characteristics are, for example, those features and features that allow an inspection system to distinguish between good and bad contour segments. In accordance with an embodiment of the present invention, a set of features and features that are useful in the detection of defective contour segments of circuit inspections are statistical timing collections that are distributed along a point plane of a inspection profile. Appropriate timing includes, for example, an XY coordinate timing corresponding to a point of gravity along a contour, a contour rotation angle, a first Han point distribution along the rotation angle direction, and a normal second axis along the first axis. Point allocation. For conceptual purposes, a profile of -28-1273216 (25) can be a pattern such as a polygon or a rectangle. Thus, in accordance with a particular embodiment of the present invention, the evaluation of the segment can be achieved by comparing the polygon to a polygon representing a reference segment. Note that a contour segment is abstractly represented by the manner in which the represented polygons are assigned by the contoured points, and various local deviations from the contours of a corresponding reference contour may become flat or completely ignored. As a result, the evaluation of a contour segment is focused on representing the characteristics of an entire segment, while the ignoring error indicates a local deviation of a defect. Moreover, by representing the "extraordinary" of features and characteristics, the various features and characteristics of the segments can be weighted to ensure that various features are desired to determine similar levels between an inspection profile and its corresponding reference. One of the proper balances. Referring now to Figures 12A and 12B, a simplified diagram showing an abstract representation of a very different segment 252, a corresponding reference 254, and a different segment 252 and reference 254, respectively, of reference numerals 262 and 264 is shown. Referring now more to Figure 13, an overlapping simplified diagram of abstract representations 262 and 264 is depicted in accordance with an embodiment of the present invention. The abstract representation overlap of a contour segment and its reference is used to evaluate whether a very different segment is an actual defect. In Figures 12A and 12B, it can be seen that the abstract representations 262 and 264 are rectangular polygons each having a center point 266, a rotation angle Θ, a length, and a width. In accordance with an embodiment of the present invention, a center point 266 is an average coordinate representing a point selected in an evaluation profile segment. The rotation angle Θ is the distribution of the corner of the selection point of the contour segment. The length of the rectangle is the distribution of the selected points in the axis profile along the direction of the rotation angle, and the width of the rectangle is the axis representing the axis along the direction of the rotation angle in the profile -29-1273216

(26) Select point assignment. Referring now to Figures 14A-14E, a simplified diagram of a computational method suitable for obtaining a polygon representing an abstraction profile is depicted in accordance with an embodiment of the present invention. A polygon calculation using the preferred formula is related to a very different segment 2 5 2 (Fig. 14A). As shown in Fig. 14B, the center point of the contour segment 252 is represented by the coordinate Xav-Yav' and is calculated as the relative χ-γ coordinate average of the selected sample points of the segment 252. For simplicity of explanation, only the sampling points i, j, k, 1, m, and n are displayed. It should be understood that the precise increase in the position of the center point is related to the number of sampling points used in the calculation. According to an embodiment of the invention, the distribution angle of the segment 2 5 2 , the length L of the polygon 262, and the height Η of the polygon 262 are calculated using the following values: the X coordinate of each sampling point

Σ (Ζ coordinates of each sampling point) 2 Number of i-samples (2) Σ [z coordinate of each sampling point * coordinate number of each sampling point 1 number of sampling points '~ ^ αν ^αν

Σ (m-sample per sample) 2 V2 (3) c=~~ number of sampling points - results, according to an embodiment of the present invention, along a configuration of a segment 2 5 2 of FIG. A segment distribution angle represented by one axis 2 6 0 is calculated using the following formula: -30-

1273216 (27) θ = (4) arctan(^, ^c) 2 represents a rectangular polygon length L assigned at a contour selection point along an angular distribution axis of the axis 260 in the contour 252 of Fig. 14D. Calculated according to a specific embodiment of the invention using the following formula: Z = + c- + c) 2 j

(5) VJ Finally, a rectangular polygon height 分配 assigned to a contour selection point along an axis of a normal angle distribution axis of a shaft 260 of the contour 2 5 2 of the contour of FIG. 14E, for example, is based on the following formula. A specific embodiment of the invention calculates:

(6) Referring now to Figure 13, it can be seen that as long as the polygon representation of the evaluation profile is 2 62, it will be represented by a polygon corresponding to the overlap of rectangular polygons 262 and 264 as shown. 2 6 4 Compared. Whether a contour segment is an actual defect is judged to be based on a similar or different degree of distance between polygons. Matches like shape polygons are known, and are usually in H. Alt, Β · Behrends and J. Blomer, under the name "Approximate Matching of

Polygonal Shapes, in Proceedings of the 7th Annual Symposium on Computational Geometry", pp. 186-193, -31 - 1273216 (28) ACM Press, New York, NY. 1991, the contents of which are incorporated herein by reference. It should be understood that the polygons representing an evaluation profile and their corresponding references are not required, and are typically not a correct match. According to a particular embodiment of the invention, the distance between the polygons 2 6 2 and 6 4 is a description of the polygons, respectively. Evaluation of each parameter, such as center point position, rotation, length and width, in addition to summing the relative distances of various parameters, and evaluating the total distance associated with an entire acceptable distance threshold. If any individual parameters exceed A predetermined maximum difference, such as a center point is converted by a distance greater than a reference center point; or if all of the various totals exceed an acceptable distance threshold, a polygon can be considered to represent an actual defect. The distance between the polygon and its corresponding reference polygon can be calculated as a multi-dimensional space interleaving. Therefore, for example, If the angle of rotation of an evaluation polygon is quite large, the distance can be compensated for by a center point, a height, and other parameters related to the length of a desired value. Please refer to 15Α and 15Β, according to the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT A portion of the inspection circuit and a simplified representation of a digital image used in the outline of the inspection process are described. Note that the present invention is described in terms of an inspection process using a pattern formed by a generally continuous profile. According to some embodiments of the present invention, the digital image is the obtained inspection pattern, and the rim is represented by pixel collection. FIG. 1 Α is a digital image portion 3 5 2 corresponding to the image portion 52 of FIG. Simplified. It can be seen that the digital image portion 3 5 2 includes a plurality of pixels 3 5 4 . The conductor pixels 354 shown by the parent and oblique lines are the substrates shown in the position of the conductor 18 of FIG. 1 instead of the oblique lines. Pixel 3 56 is represented in Table -32- of Figure i

1273216 (29) Face 20 position. The definition of the contour as a whole contour pixel 3 60 of a digital contour image 3 5 8 is shown in Figure 15B. In accordance with an embodiment of the present invention, the contour pixels 358 are the conductor pixels 354 in the digital image portion 352, and at least one end is bounded by a substrate pixel 356. The contoured pixels 360 are for example extracted from any suitable grayscale or binary digit image portion 352 of an inspection pattern 16 in a manufacturing circuit. According to another embodiment of the present invention, whether a contour segment is very different from the various decisions it refers to is between checking a digital contour image of the pattern portion and its contour pixels of a digital contour image to be referenced. The distance is achieved. Thus, according to an embodiment of the present invention, if the position of one or more contour pixels is different from the corresponding contour pixel position of the reference digital image in the reference, a contour segment is visible at the beginning of the inspection phase. For a very different silhouette segment. During the second phase of the inspection, as depicted in Figure 7, if a very different segment is converted by more than 3 pixels during a micro-registration process, or if a pixel of a contour segment is at a corresponding pixel from a reference profile A distance greater than 2 pixels, a very different segment can be considered an actual defect. It should be understood that these values are example values and they can be modified to image resolution and/or a function to check sensitivity. Referring to FIG. 1, a simplified image of a digital image of a portion of the inspection circuit 425 defined by the sub-pixel resolution profile element 454 is shown. The sub-pixel resolution profile element as indicated by the arrow is a sub-pixel position representing a contour in a digital image. Including generation as defined by sub-pixel outline elements

1273216 (30) Sub-pixel information capture from gray level images of pattern images is well known. Reference is made to U.S. Patent Nos. 5,774,572, 5,774, 5,73, and U.S. Patent Application Serial No. 09/63, the entire disclosure of which is incorporated herein by reference. In accordance with a preferred embodiment of the present invention, the sub-pixel resolution profile elements in the inspection pattern representation are compared and evaluated with a reference image, typically forming a vector representing the contour, as previously described to find the actual pattern in the pattern. defect. It will be appreciated by those skilled in the art that the present invention is not limited to the particular description and description herein. Rather, the scope of the present invention includes the various features and combinations described above, and can be modified and added by those skilled in the art as long as the reading is not in the foregoing description of the prior art. BRIEF DESCRIPTION OF THE DRAWINGS The invention will be better understood and appreciated from the following detailed description of the drawings, in which: FIG. 1 is a simplified diagram depicting a pattern inspection system constructed and operative in accordance with an embodiment of the present invention; FIG. 3 is a flow chart showing the initial inspection function of FIG. 1 according to an embodiment of the present invention; FIG. 4 is a flowchart of an initial inspection function according to an embodiment of the present invention; FIGS. 4A-4F are diagrams showing an embodiment of the present invention. FIG. 3 is a simplified flowchart of the result of checking the function steps of FIG. 3; FIG. 5 is a simplified flowchart of the process of obtaining the image portion and the corresponding reference portion by using the micro-registration in the specific embodiment of the present invention; -34-1273216 (31) A and 6B are selected viewpoint views describing the micro-registration process of the process of FIG. 5; FIG. 7 is a simplified flowchart for performing the suspicious defect evaluation process according to an embodiment of the present invention; FIGS. 8A-10B are diagrams according to the present invention. A simplified diagram of a different segment evaluation process of a particular embodiment; FIG. is a simplified flow diagram of a preferred method of performing the process of evaluating the process of FIG. 7;

1A and 1B are simplified diagrams of outlines and reference portions described in conjunction with abstract representations in accordance with an embodiment of the present invention; FIG. 13 is a simplified simplified representation of the abstract representation of FIGS. 12A and 12B; FIG. 4E is a simplified diagram depicting a portion of an abstract representation calculation in accordance with an embodiment of the present invention; FIG. 15A is a simplified diagram of a portion of a digital representation of a circuit pattern;

Figure 15B is a representation of a digital image pixel outline of Figure 15A; and Figure 16 is a simplified illustration of a digital image depicting a pattern defined by a sub-pixel resolution profile element. Schematic representation symbol description 10 pattern inspection system 12 detector module 14, 24, 52 image portion 16 pattern 22 circuit substrate 20 surface 18 conductor member -35 - 1273216 (32) 26 image processor unit 40 second stage Evaluation 42 Time slot 54 Designated conductor image 62, 72, 118 Profile 64, 84 Profile display 91, 92 Threshold 116 Vector 112 Arrangement point 114, 1 17 Reference section 126 Perimeter 200 Evaluation process 92 Quality threshold indicator 262 , 252, 254 fragment 266 allow distance indicator 260 axis

356, 358, 360 pixels 3 5 2 digital image portion -36-

Claims (1)

Patent application ----- Ma Li range replacement (9) years ίο月) Qiao r years, July) / Japanese repair (more) is replacing page ____-.".'.................. . . - picking up, claiming a patent range 1. A method for inspecting a circuit pattern, comprising: performing an initial inspection of a continuously acquired image of a circuit pattern to define a potential defect in the circuit pattern; Identifying a potential defect in the circuit pattern during the initial inspection, interrupting the initial inspection and performing a second evaluation of the portion of the latent image that includes the potential defect; and after completing the second evaluation, replying to the Initial inspection. 2. The method of claim 1, wherein the continuously obtaining image is obtained by continuously scanning a checked circuit pattern. 3. The method of claim 1, wherein the continuously acquiring image is obtained by continuously imaging the second spatial region of the circuit pattern. 4. The method of claim 1, wherein the initial inspection includes The continuously acquired image is detected at a rate faster than the acquisition rate of the continuously acquired image. 5. The method of claim 1, further comprising buffering the successive image portions when the second evaluation is performed to obtain the continuously acquired image portion. 6. The method of claim 4, further comprising buffering the successive image portions when the second evaluation is performed to obtain the continuously acquired image portion. 7. The method of claim 6, wherein the recovering the initial inspection comprises examining the buffer portion at a rate faster than the inspection rate, and then further checking the additional continuous 83010 of the image at the inspection rate. 951030.doc The nature of the decision is that it should be taken and taken. 1273216 Part. 8. The method of claim 1, wherein the initial inspection of the continuous acquisition of the image comprises examining a contour representation of the circuit pattern. 9. The method of claim 1, wherein the implementing a second evaluation comprises an algorithm group, wherein the algorithm group is different from the originally checked algorithm group. 10. The method of claim 9, wherein the implementing a second evaluation comprises an algorithm group, wherein the algorithm group requires more computer resources than the initial inspection. 11. A method for inspecting a circuit pattern, comprising: obtaining an image of a circuit pattern during an image acquisition time interval; performing an initial inspection of the image at a rate faster than obtaining the image; and responding to The initial inspection performs an additional evaluation of a portion of the images, wherein a time interval between the initial inspection and the additional assessment is substantially the same as the image acquisition interval. 12. The method of claim 11, wherein the initial inspection and the additional evaluation are performed using the same computer processor. 13. The method of claim 11, wherein the first check is to use a first algorithm group, and the one additional evaluation includes using a second algorithm group, wherein the second algorithm group It is different from the first algorithm group. 14. The method of claim 13, wherein the second algorithm group uses more computer resources than the first algorithm group. 15. For the method of claim 13 of the patent scope, wherein the second algorithm group is 83010-951030.doc 1273216 卜月"曰妒L: exchange for a few jΙιιιίικι V. 1. 111111"r need to be the first The algorithmic group has a longer time interval to examine a particular portion of the image. 16. The method of claim 12, wherein the initial inspection is interrupted by a potential defect in the circuit pattern, and the additional evaluation is on an image portion identifying the potential defect. The method of claim 11, wherein the obtaining an image comprises obtaining a contour representation of the circuit pattern. 18. A method for inspecting a circuit pattern, comprising: obtaining an image of the inspection circuit pattern; identifying the image portion different from a corresponding reference portion, and the difference is greater than a first different threshold; The portions are micro-registered; and a micro-registration portion is evaluated to determine a defect in the circuit pattern. 19. The method of claim 18, wherein obtaining an image comprises obtaining a contour representation comprising an outline of the inspection circuit. 20. The method of claim 19, wherein the identifying comprises extracting a contour segment that differs from a corresponding reference by a greater than the first different threshold. 21. The method of claim 18, wherein the micro-registration comprises a portion, wherein the portion is different from a pair of reference portions and the corresponding reference portion in a repetitive process. 22. The method of claim 18, wherein the micro-registration comprises a portion of the portion, wherein the portion is different from the entire arrangement point 83010-951030.doc 1273216 A corresponding reference and the corresponding reference part. The method of claim 22, wherein the arrangement point is related to a vector representing one end of a conductor profile, and the profile end is a substrate. 24. The method of claim 23, wherein the vector is used to determine a corresponding location in the image. 25. The method of claim 18, wherein the identifying portion is a first algorithm group and the evaluation is using a second algorithm group 0 26. For example, claim 25 The method wherein the second algorithm group requires more computer resources than the first algorithm group. 27. The method of claim 18, wherein the execution of the identified portion is related to a first threshold. 28. The method of claim 27, wherein the evaluating is performed in relation to a second threshold or a distance from a portion of the micro-registration. 29. A method for inspecting a circuit, comprising: forming a contour representation of an inspection circuit; initially inspecting a plurality of locations on a contour of the contour representation, in order to find one of the plurality of locations a first plurality of locations of the first defect likelihood; and during the initial inspection, as long as one of the first plurality of locations is identified, a first number of locations are brought a two-stage inspection, before the initial inspection is completed, in order to determine which of the first plurality of positions is constituted to have a second deficiency 83010-951030.doc 1273216 A second plurality of locations of likelihoods, wherein the second defect likelihood is greater than the likelihood of the first defect. 30. A method for pattern inspection, comprising the steps of: comparing a pattern property descriptor in an inspection pattern with a pattern property descriptor in a reference, and providing a comparison output; A quality threshold is applied to the comparison output to provide a threshold output that distinguishes between a portion of the pattern that meets the threshold value of the equivalent surface quality and a portion of the pattern that does not meet the relatively high quality threshold; and thereafter, further Check for portions of the pattern that do not meet the fairly high quality threshold. 31. The method of claim 30, wherein at least one of the descriptors comprises an outline of at least a portion of the pattern. 32. The method of claim 30, wherein at least one of the descriptors comprises an edge of at least a portion of the pattern. 33. The method of claim 30, wherein at least one of the descriptors comprises a color of at least a portion of the pattern. 34. The method of claim 30, wherein at least one of the descriptors comprises a geometric feature of at least a portion of the pattern. 35. The method of claim 4, wherein the pattern comprises a conductive pattern formed on a printed circuit board. 36. The method of claim 30, wherein the portion of the pattern that meets the relatively high quality threshold is a normal amount that exceeds a portion of the pattern that does not meet the relatively high quality threshold. 37. The method of claim 30, wherein the inspection pattern and reference 83010-951030.doc 1273216 yr year f6 month W曰 repair 11) comparison of the test, providing a comparison output, and applying a fairly high quality threshold to the comparison output to provide a pattern that meets the relatively high quality threshold A threshold value output that is partially different from a portion of the pattern that does not meet the relatively high quality threshold is normally implemented using less computational resources than each of the inspection pattern portions. 38. The method of claim 36, wherein the method is for comparing a pattern to a reference, providing a comparison output, and applying a relatively high quality threshold to the comparison output to provide compliance The step of outputting a threshold value that differs between the pattern portion of the high quality threshold and the portion of the pattern that does not meet the relatively high quality threshold is normally to use less computational resources than each of the inspection pattern portions than the further inspection step Implementation. 39. The method of claim 30, wherein the further is to use at least a second threshold, wherein the second threshold is less stringent than the substantially high quality threshold. 40. The method of claim 30, wherein the comparing step comprises comparing a rim element to a wheel element. 41. The method of claim 30, wherein the comparing step comprises comparing different pixels. 42. A method for inspecting a pattern defect, comprising: displaying at least two predetermined characteristics of a portion of the inspection pattern in a first polygon; at least two predetermined portions of a reference portion of the portion of the inspection pattern The characteristic is represented by a second polygon; and the comparison between the first polygon and the second polygon is determined at 83010-951030.doc 1273216 This section checks for defects in the pattern. 43. The method of claim 42, wherein the predetermined characteristic comprises a statistical opportunity relating to a spatial distribution of locations in a portion of the inspection pattern or at a reference portion. 44. A method comprising pattern inspection comparing a inspection pattern with a reference by: rendering a portion of the overall spatial characteristic of the inspection pattern in a first polygon; a portion corresponding to the inspection pattern The entire spatial characteristic of the reference is represented by a second polygon; the first polygon is compared to the second polygon to determine defects in the pattern. 45. The method of claim 44, wherein the first polygon and the second polygon are an assignment representing the spatial characteristic. 46. The method of claim 45, wherein the spatial characteristic includes a statistical opportunity. 47. A method for pattern inspection, comprising: providing a contour representation of an inspection pattern, the contour representation defining a contour in the inspection pattern, substantially without identifying a functional structure constituting the inspection pattern; Comparing a corresponding reference pattern, extracting the contour representation portions different from the reference, the system transmits a portion greater than a predetermined allowable difference threshold value and outputs the captured portion; and locally processes at least some of the captured portions to Identify defects in the pattern. 83010-951030.doc 1273216』I recorded. The method of claim 47, wherein the provision of the representation includes an outline representation provided in the pattern. 49. The method of claim 47, wherein the corresponding reference pattern comprises a contour representation of a reference pattern. 50. A method for fabricating a circuit, comprising: depositing a portion of a circuit on a substrate according to a pattern; performing an initial inspection of a continuous image of the pattern to determine in the portion of the circuit Potential defect; as long as a potential defect in the pattern is identified during the initial inspection, the initial inspection and implementation of a second assessment comprising the continuous acquisition of the potential defect is interrupted; And after completing the second evaluation, reply to the initial check. 51. A method for fabricating a circuit, comprising: depositing a portion of a circuit on a substrate according to a pattern; obtaining an image of the circuit during an image acquisition time interval; An initial inspection of the image is performed at a rate of the image; and an additional evaluation of the partial image is performed in response to the initial inspection, wherein a time interval for performing the initial inspection and the additional assessment is substantially The image acquisition time interval is the same. 52. A method for fabricating a circuit, comprising: depositing a portion of a circuit on a substrate according to a pattern; obtaining an image of the inspection circuit; and identifying the image portions different from a corresponding reference portion And the difference is greater than a first different threshold; 83010-951030.doc 1273216 will be judged. 53. The root of the root shape must be trapped in the middle of the big 54. - The root of the inspection will refer to the greater than the bureaucratic June / day to buy I. 1||-. ................................. - ·. ^ - 'Τ'τβτ These partial micro-registrations; and Estimating a micro-registration portion to determine a method for fabricating a circuit in the circuit pattern, comprising: depositing a portion of the circuit on a substrate according to a pattern; forming the pattern outline of the inspection circuit Representing; detecting a plurality of locations on the rim of the contour representation for finding a first plurality of locations at the plurality of locations having a first defect likelihood; and during the initial inspection, identifying The first plurality of positions and one of the contour positions are followed by a second stage check at the first plurality of positions, and the one of the first plurality of positions is to be determined before the initial check is completed. A second plurality of locations having a second energy deficiency is formed, wherein the second defect is likely to be the first defect likelihood. A method for fabricating a circuit, comprising: depositing a portion of a circuit on a substrate according to a pattern; for "inspecting the portion of the circuit of the circuit, not defining the contour of the portion of the circuit Substantially without identifying the functional structure constituting the circuit; the contour representation is compared to a corresponding reference pattern for extracting the contour representation portion different from the difference, wherein the difference is a predetermined allowable difference threshold And outputting the captured portion; and processing at least some of the captured portions to identify the 83010-951030.doc in the circuit The method of claim 54, wherein the corresponding reference pattern comprises a contour representation of a reference pattern. 56. A method for fabricating a circuit, comprising: depositing a portion of a circuit on a substrate according to a pattern; comparing an inspection circuit with a pattern characteristic descriptor in a reference to provide a comparison output; applying a relatively high quality threshold to the comparison output to provide a difference between a portion of the pattern that meets the relatively high quality threshold and a portion of the pattern that does not meet the relatively high quality threshold Threshold output; and further inspection of portions of the pattern that do not meet the relatively high quality threshold. 83010-951030.doc -10-