JP4744234B2 - Component positioning method and apparatus - Google Patents

Description

  The present invention relates to a component positioning method and apparatus, and in particular, processes an image of an imaged component suitable for use in an electronic component mounter, and efficiently processes the component from features such as a side and a corner of the component. The present invention relates to a component positioning method and apparatus that can be well positioned.

  In an electronic component mounting machine, an electronic component supplied from a feeder is sucked by a suction nozzle, and the suction nozzle is moved to a predetermined position on a circuit board to mount the electronic component on the substrate. In this case, since the electronic component is not necessarily picked up in a correct posture, when mounting the electronic component on the circuit board, the image recognition unit acquires image data of the component with an imaging device such as a CCD camera immediately before mounting. Then, image processing is performed to detect the deviation between the suction center and the component center, and the suction inclination, thereby positioning the components and mounting the electronic components with high accuracy.

  Here, since the conventional part positioning device has a recognition algorithm specialized for each part shape, the user selects the type of the part, and the part data (part, lead, ball terminal, If the size, position, number, inspection conditions, etc. of features such as corners, sides, marks, etc.) are input, the parts can be positioned correctly.

  In addition, as for a deformed part which is not a fixed part of a fixed shape such as QFP or BGA, as proposed in Patent Document 1, a plurality of feature parts are combined to define positioning data of one part. And correctly position the part.

  At this time, information on the order in which the feature defining the part is recognized is incorporated in each recognition algorithm, and it is not necessary for the user to be aware of this when performing the part recognition process. There was no.

JP 2001-209792 A

  However, as in the conventional component positioning apparatus, if the recognition order of component features is fixed and cannot be specified by the user, there are the following problems.

(1) Occurrence of recognition error or misrecognition As shown in FIG. 1A, for example, the material of the lead terminal 11A to 11D on the four sides of the component 10 is different from that of the specific terminal 11A, and the image becomes dark. A recognition error occurs when the reflection of a feature object is unclear, or when there is a lot of noise reflection due to the mark 12 printed on the body of the component 10 or the jumping out of the wiring 14 as shown in FIG. In addition, as shown in FIG. 1C, when a shape (also referred to as a similar object) 16 similar to the characteristic object (terminal 11A) is adjacent at the same pitch, the characteristic object and the similar object are correctly determined. Cannot be recognized and recognition is not stable.

  This is because, in the component positioning function in the mounting machine, in order to avoid excessive tact, if an error occurs during component recognition, the error ends as shown in FIG. For this reason, if the image quality of the feature to be recognized first is poor as shown in FIGS. 1A and 1B or a confusing shape as shown in FIG. It is easy to generate a phenomenon that the feature cannot be recognized normally, and as a result, the positioning of the component fails.

  In this case, as shown in FIG. 2 (B), if the recognition order of the feature having a poor image quality or a confusing shape can be postponed, the position information of other feature that can be recognized correctly is used. There is also a possibility that positioning can be performed correctly.

  However, information such as the poor image quality and the presence / absence of a similar shape as shown in FIG. 1 is difficult to determine in advance by the component positioning device from the component data specified by the user. I couldn't cut it.

(2) Difficulty in identifying the cause of the error The component positioning device has a function of outputting the feature that caused the error and the cause of the error when a recognition error occurs. However, there are many cases where the direct cause of the recognition error is not always the true cause of the error. In order to determine the true error cause at the user level and remove it, it is necessary for the user to be able to grasp the process of recognition processing that led to the error and the factor that the detection process cannot be performed normally.

  When the recognition order of component features is fixed as in the conventional component positioning device, the user does not know in which order the feature of the component is recognized. I couldn't figure out what happened. In addition, since the conventional component positioning device does not have a function of notifying the user of factors that are likely to cause errors, the user cannot grasp specific error factors.

According to the first aspect of the present invention, in the method for positioning a part having a plurality of features, a user designates an order of recognizing the features, recognizes the features according to the order designated by the user, and performs image processing. Thus, the problem (1) is solved. In addition, when the user cannot specify the recognition order, the recognition state when a feature group that is a set of the same feature is individually recognized and executed in advance by the same algorithm as the part recognition process, and the presence of similar items An evaluation value is calculated from the presence / absence, feature density, background density, presence / absence of noise components, and recognition speed, and the optimal recognition order is obtained by assigning the recognition order in descending order of the evaluation value. By creating component data and recognizing components in the optimum recognition order, even a user who does not know the optimum recognition order can efficiently recognize the components. Further, the problem (2) is solved by notifying the user of the optimum recognition order obtained by the means and the evaluation value of each item.

In the invention of claim 2 , when the recognition order designated by the user is actually difficult to recognize, the recognition order is automatically acquired by the acquisition method of the optimum recognition order, and at the time of subsequent recognition The problem (1) is solved by reflecting.

According to a third aspect of the present invention, in a part positioning apparatus having a plurality of features, the user designates the order of recognizing the features, and recognizes the features and performs image processing according to the order designated by the user. And a recognition state when a feature group that is a set of the same feature is individually recognized and executed in advance by the same algorithm as the component recognition process. The evaluation value is calculated from the presence / absence of the image, the density of the feature object, the background density, the presence / absence of the noise component, and the recognition speed, and the optimum recognition order obtained according to the highest evaluation value and the optimum recognition obtained by the function The present invention provides a component positioning apparatus provided with a function of recognizing parts in order and a function of notifying a user of an optimal recognition order and evaluation values acquired by the function.

According to the first and second aspects of the invention, since the recognition order of the feature of the part can be specified at the user level, (1) it is difficult to recognize due to noise, unclearness of the reflection, feature position and size variation, Improve recognition rate by lowering the recognition order of difficult-to-recognize features by creating part data that reflects information that only the user can grasp, such as the possibility of false detection by similar objects be able to. In addition, (2) when an error occurs, the user can determine which error occurred when what was recognized in what order. Furthermore, (3) when an error occurs, component recognition may be possible if the recognition order of the feature group along the cause of the error is lowered, but this can be determined and operated by the user. Become.

In addition, by using the optimum recognition order acquisition function according to claims 1 and 3 , recognition is performed when individual feature groups that are a set of the same feature are recognized and executed in advance by the same algorithm as the part recognition processing. Since the optimal recognition order can be acquired from the state (normal or abnormal), presence / absence of similar objects, feature density, background density, presence / absence of noise components, and recognition speed, it can be tacted in terms of recognition rate in a short time. The most suitable part recognition data can be created. By using this optimal recognition order acquisition function for parts of various shapes, the user can learn the pros and cons of the component recognition device (factors that make the recognition unstable), and as a result, rely on this function. The optimum recognition order can be independently determined without the need, and the function (the order in which the features are recognized is specified by the user without relying on this function). The function of recognizing the feature object and performing image processing to position the component in accordance with the sequence described above can be used.

Further, according to the second aspect , when it is difficult to recognize in the recognition order set by the user, the recognition order can be automatically changed to the optimum recognition order. Even when the optimum recognition order does not match, it can be recognized correctly.

  As parts with new shapes are always on the market, how quickly new parts can be mounted greatly affects the time required for board production. In this sense, the acquisition function of the optimal recognition order in claim 2 not only increases the user's operation efficiency, but also helps to acquire image processing habits, so that the direct specification function of the recognition order of claim 1 can be used. Therefore, it leads to further efficiency. In addition, even if it is difficult to recognize in the recognition order specified by the user as in claim 3, since it is automatically replaced in the optimal recognition order, return work does not occur and efficient production can be performed.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 3 shows an embodiment of an electronic component position detection apparatus for carrying out the present invention.

  In FIG. 3, 10 is an electronic component, 28 is a suction nozzle for sucking the electronic component 10, and 30 is an illumination device 32 provided with a number of illumination lamps, a lens 34, for example, an imaging device 36 comprising a CCD camera. A component detection unit 40 includes an image processing apparatus 40 including a CPU 42, an A / D converter 44, a component data storage memory 46, an image memory 48, and an error image memory 50. A monitor 60 displays an input image and a processed image. It is.

  The operation will be described below.

  The user inputs data of the component 10 to be recognized to the mounting machine body. At this time, the user can specify the recognition order of the feature of the component 10.

  For example, let the component 10 as shown in FIG. 4 (here, an example of a so-called irregular component, not a regular component such as QFP or BGA) recognize the upper side lead row L0, the lower left corner C1, and the lower right corner C2, Assume that the component 10 is to be positioned. At this time, the user recognizes that the mark 12 is printed on the inside of the C2 corner, and therefore the interior of the C2 corner appears dark when the part 10 is imaged, and the recognition rate is poor when the image quality is poor. I often know that. Therefore, in this case, it is possible to create a part data by defining a recognition order for recognizing the C2 corner last.

  The mounter body that has received the component data transmits it to the image processing apparatus 40. The image processing apparatus 40 checks the consistency of the received data. If the recognition order is designated, the feature processing data is linked in the recognition order designated by the user and stored in the component data storage memory 46.

  For example, in the case of the components shown in FIG. 4, when the user defines that the recognition is performed in the order of L0, C1, and C2, the link structure as shown in FIG. 5 can be used for management. Here, the feature object group information and the feature object information in FIG. 5 correspond to the element group information and element information proposed in Patent Document 1, respectively, and detailed description thereof will be omitted.

  The component recognition process starts when the mounter body transmits a recognition execution command to the image processing apparatus 40. The recognition execution command includes information such as a component ID, a component type (QFP, BGA, irregularly shaped component, etc.), a camera channel, and the like.

  The image processing apparatus 40 captures the part 10 with the camera 36 of the designated channel and stores the image in the image memory 48. Also, the corresponding part data is taken out from the part data storage memory 46 and transferred to the recognition execution routine.

  Here, FIG. 6 shows a recognition processing procedure in the case where the recognition order is specified for the odd-shaped parts as shown in FIG. As shown in FIG. 5, the recognition is performed in the order linked to the component data, and when the recognition of all the feature objects is completed, the component center and the inclination are calculated. The image processing apparatus 40 transmits the component center and inclination to the mounter body, and ends the recognition process.

  The function that allows the user to specify the recognition order as in the present invention improves both operability and functionality for a familiar user, but it can be used conversely for an unfamiliar user or a user who refrains from troublesome settings. It becomes a troublesome function.

  Therefore, an optimum recognition order acquisition function can be provided as a component recognition test function. This is because the image processing apparatus 40 determines in which order the feature group should be recognized in the component data designated by the user, and in order to reflect this recognition order in the actual recognition processing, This is a function to rewrite the link structure in the recognition order.

  This function operates at the following two timings. First, as in normal component recognition, operation is performed by transmitting component data to the image processing apparatus 40 and then transmitting an optimum recognition order acquisition command. At this time, the user does not need to specify the recognition order in the part data, but can specify the order of a certain feature group partly. Like the recognition execution command, the optimum recognition order acquisition command includes information such as the component ID, component type, and camera channel. In addition, the number of executions and the operation mode (automatically acquired optimal recognition sequence is simply notified to the user). It is possible to specify whether or not to reflect it in actual part recognition).

  When receiving the optimum recognition order notification command, the image processing apparatus 40 acquires the optimum recognition order and reflects this in the component data structure. This result is notified to the user.

  This optimum recognition order acquisition command can be executed on a captured image. For example, if an error occurs during recognition, the error image is stored in the error image memory 50. For example, if an error occurs frequently, the optimum recognition order acquisition function is operated on the error image. It is also possible.

  Secondly, after the component data in which the user designates the recognition order is transmitted to the image processing apparatus, when the first recognition results in an error, it automatically operates using the error image. If a recognition error occurs for the first time, the recognition order may have been specified incorrectly. Therefore, the correct recognition order is acquired by using the optimum recognition order acquisition function, and the recognition order automatically acquired is changed to be valid at the time of subsequent recognition. However, when this process is not necessary, it is possible to set so as not to operate with the component data.

  Depending on the positioning algorithm, if recognition is performed according to a user's command, recognition processing may take extra time or a recognition error may occur. For this reason, the present embodiment also has a function of automatically correcting the order designated by the user.

  That is, as illustrated in FIG. 7, when a plurality of the same feature group is defined, if recognition is performed in units of individual feature groups, a recognition error may occur. For example, assume that the user defines the order of recognition of the lead terminals of the component 10 shown in FIG. 7 in the order of L0, L1, and L2. In this case, another group of lead terminals L1 and L2 are adjacent to both sides of L0 that is initially designated for recognition. Here, if the electrode detection algorithm proposed in Japanese Patent Laid-Open No. 2000-232299 is used for the lead detection process, it is necessary that a sufficient background exists on either the left or right side of the lead group to be detected. That is, there may be a case where L0 cannot be recognized first on the detection algorithm. In this case, it changes so that L2 designated 2nd may be detected first. Then, L0 can be predicted and positioned using the result of detecting L1, and the components can be correctly positioned.

  The restriction of the recognition order due to the convenience of the detection algorithm changes with the improvement of the detection algorithm. That is, even if the same part data is specified by the user, the optimal correction method may differ depending on the version of the program, so there is no need to notify the user of the recognition order corrected by this recognition order automatic correction function.

  Next, the optimum recognition order acquisition function is executed according to a procedure as shown in FIG.

  That is, first, in step 100, relevant part data is extracted.

  Next, in step 102, the component 10 is imaged by the imaging device 36 and the image is stored in the image memory 48.

  Next, in step 104, each individual feature group is recognized alone. This is for determining whether or not recognition is possible without relying on the positional relationship between the groups, so there is absolutely no information on what feature groups exist in addition to the feature groups to be recognized. Recognize without using. The recognition algorithm used at this time is the same as that used for component recognition.

  If it is determined in step 106 that the feature has been detected normally, the process proceeds to step 108, and it is confirmed whether there is a similar object having the same shape as the feature to be detected around the detected feature. . The state having a similar shape is a state as shown in FIG. 9, for example. This can also be realized by using the same algorithm as the recognition process.

  Based on the above recognition and the result of the presence / absence of the similar object, in step 110, the ease of recognizing the feature object group is determined. The ease of recognition is evaluated for each of the presence of similar objects in the vicinity, the density state of the feature object and the background, the presence / absence of noise, and the recognition speed, and scored as shown in FIG.

  Here, the presence / absence of a similar substance is determined to be difficult to recognize because the closer the similar thing is, the higher the risk of detection, as shown in FIGS. 9 (A) to 9 (C). In addition, for example, it is determined that there are similar parts having shapes as shown in FIGS.

  Further, it can be determined that the higher the feature density, the easier it is to recognize. Therefore, the average density value of the feature can be obtained and divided into three groups based on the default binarization threshold as illustrated in FIG.

  Further, it can be determined that the lower the background density is, the easier it is to recognize. Therefore, the average density of the background around the feature object group can be obtained and divided into three groups based on the default binarization threshold as illustrated in FIG.

  Further, it can be determined that the presence or absence of noise is more easily recognized as the noise is smaller. Depending on the shape of the feature, it may be possible to determine the presence or absence of noise during the detection process. In this case, the presence or absence of noise is recorded. On the other hand, in the case of a feature group in which the presence / absence of noise cannot be determined during the detection process, the presence / absence of noise is unknown.

  The recognition speed can be determined to be easier to recognize as the recognition time is shorter. Therefore, it is possible to measure the time required for recognition for each recognition of each feature group, and after completing the recognition of all the feature groups, it is possible to give high scores in order from the earliest recognition time.

  After confirming the recognition of all the feature object groups in step 112, if all are recognized normally, in step 114, the validity of the detected position is checked. Here, the mutual positional relationship of the detection results is examined. At this time, as shown in FIG. 13, if there is an object detected with a deviation from the designated coordinates, this feature group has not been detected correctly, and the recognition ease evaluation value is reset to zero. It has already been investigated whether there is a similar object around the feature group that has been detected in a shifted manner, and if there is a similar object nearby, it is due to false detection of the similar object. If it does not exist in the vicinity, it can be determined that this is due to an error in the coordinates designated by the user. If it is determined that there is a possibility of a coordinate error, a warning message to that effect is added when the final result is output.

  The processing from the imaging in step 102 to the mutual position inspection in step 114 is repeated in step 116 for the number of executions designated by the user.

  Finally, in step 118, the part data structure is changed so that recognition can be performed in descending order of evaluation value. In step 120, the order is notified to the user.

  At this time, the user is also notified of the evaluation value for ease of recognition and the evaluation points used to calculate the recognition value (presence / absence of similar substances, feature density, background density, presence / absence of noise, detection time). By repeatedly acquiring this data, it becomes easier for the user to learn the conditions that make it easy to detect a feature and, conversely, factors that hinder the detection of the feature.

  In addition, when the recognition order is not designated by the user, it can be recognized by the method described in Patent Document 1, for example.

  In addition, it is also possible to specify the recognition order for each of the feature parts (for example, in the case of QFP parts, the lead row on each side) for the fixed parts not illustrated. When recognizing, if recognition is performed after obtaining the recognition position information in the recognition order, it can be easily recognized as specified by the user.

  In this embodiment, if the recognition order specified by the user is difficult to recognize in the recognition algorithm, or there is a possibility of erroneous detection or an error, the recognition order and the structure of the feature group are changed and recognized internally. As a result, the user can specify the recognition order without being aware of the difficult internal recognition algorithm. In addition, since the recognition order can be designated externally, the recognition rate is prevented from decreasing.

  Also, if you want to recognize parts that cannot be determined from which feature to recognize, or if it is troublesome to determine, automatically obtain the optimal recognition order using the optimal recognition order acquisition command. be able to.

  In addition, when a component recognition error occurs, it may be possible to avoid an error by changing the recognition order using an optimum recognition order acquisition command for the error image. This improves the error countermeasure function that can be performed by the user.

  In addition, since the optimum recognition order acquisition command executes recognition of the feature group designated by the user as a whole, it judges the feature feature coordinate designation error of the user from the inconsistency relationship of the detected coordinates and points out to the user. be able to.

Plan view showing examples of parts that cause recognition errors and misrecognitions The figure which compares and shows the (A) detection failure example by the recognition order, and the (B) detection success example 1 is a block diagram including a partial perspective view showing an embodiment of an electronic component position detecting device for carrying out the present invention. The top view which shows the example of the odd-shaped component used for the recognition in the said embodiment The figure which shows the example of the link structure managed according to the order which the user defined with the components of FIG. The flowchart which shows the procedure of the recognition process in case the recognition order is designated by the odd shape components like FIG. Plan view of parts for explaining the automatic correction function of the recognition order in the embodiment The flowchart which shows the process sequence of the optimal recognition order acquisition function in the said embodiment. The figure for demonstrating the similarity used with the said optimal recognition order acquisition function The figure which similarly shows the example of scoring The figure which shows the example of the case classification of the feature object density similarly The figure which also shows the example of the case classification of the background density The figure which shows the example where designation of the detection result differs similarly

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Components 11A-11D ... Terminal 12 ... Mark 14 ... Wiring 16 ... Similarity 18 ... Hole 28 ... Adsorption nozzle 30 ... Component detection part 32 ... Illumination device 34 ... Lens 36 ... Imaging device 40 ... Image processing device 42 ... CPU
46 ... Part data storage memory 48 ... Image memory 50 ... Error image memory 52 ... Monitor

Claims (3)

In a method for positioning a component having a plurality of features,
The user specifies the order in which features are recognized,
A method for positioning a component for recognizing a feature and performing image processing according to an order designated by the user,
The recognition state, presence / absence of similar objects, feature density, background density, and noise component when individual feature groups that are the same feature set are recognized and executed in advance using the same algorithm as the part recognition process. A function that calculates an evaluation value from the presence / absence and recognition speed, acquires an optimal recognition order in descending order of the evaluation value, a function that recognizes components in the optimal recognition order acquired by the function, and an optimal acquired by the function A component positioning method having a function of notifying a user of a recognition order and an evaluation value.   If the recognition order designated by the user is difficult to recognize in an actual recognition image, a function for positioning the component by changing to an optimal recognition order internally using the optimum recognition order acquisition function according to claim 1 is provided. The component positioning method according to claim 1, further comprising: a component positioning method according to claim 1. In a positioning device for parts having a plurality of features,
Means for the user to specify the order in which features are recognized;
Means for recognizing features and processing them according to the order specified by the user;
A component positioning device comprising:
The recognition state, presence / absence of similar objects, feature density, background density, and noise component when individual feature groups that are the same feature set are recognized and executed in advance using the same algorithm as the part recognition process. A function that calculates an evaluation value from the presence / absence and recognition speed, acquires an optimal recognition order in descending order of the evaluation value, a function that recognizes components in the optimal recognition order acquired by the function, and an optimal acquired by the function A component positioning apparatus having a function of notifying a user of a recognition order and an evaluation value.

Images