JP2004219283A - Method and system for recognizing component - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a component discrimination method and apparatus which can accurately discriminate components, having terminals which are not necessary for centering at their corners, and increase component discrimination accuracy. <P>SOLUTION: A photographed image of an sucked component is processed, and using this image processing, corner coordinates C1-C4 are acquired (step S5). When it is determined that corner terminals L1-L4 exist, based on these acquired corner coordinates, the corner terminals are deleted (step S6); and the center and the inclination of the component are calculated, based on terminals, after the deletion of the corner terminals (step S9). The accuracy of component discrimination can be enhanced, and the precision of component mounting can be increased in formation such as this, since the component discrimination is performed, after deleting the corner terminals which do not have any relation to centering and thus are not necessary for recognized discrimination. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a method and apparatus for recognizing a component, and more particularly, processes a captured image of a component, and recognizes the orientation of the component, such as the component center and the component inclination, based on the terminal of the component detected by the image processing. The present invention relates to a component recognition method and device.
[0002]
[Prior art]
Conventionally, in order to accurately mount electronic components (hereinafter simply referred to as components) on a printed circuit board in a component mounter, the components picked up by the suction nozzle are imaged by an imaging device such as a CCD camera before being mounted on the printed circuit board. The component is accurately mounted on the printed circuit board by calculating the suction attitude at the time of suction, that is, the suction shift (the shift between the suction position and the center position of the component, the shift of the suction angle), and correcting it.
[0003]
When a component is imaged and its suction attitude is recognized, the component has a lead terminal on each side. Therefore, as in Patent Document 1, a terminal is detected for each side where a lead terminal is present, and the detection result is obtained. Is performed based on the component recognition, and the suction deviation is calculated. In addition, a process of deleting items other than terminals from the detection result and inspections such as detection of a defective terminal are also performed for each side to improve detection accuracy.
[0004]
[Patent Document 1]
JP-A-6-223185 (Claim 1)
[0005]
[Problems to be solved by the invention]
However, since the detection lead inspection by the conventional lead recognition method is performed only for each side, the items that can be inspected are limited. In particular, the only processing to delete items other than lead terminals from the detection results is processing to delete terminals that are not on the same straight line, and processing to delete leads at the ends when the pitch at both ends is abnormal. Therefore, when something other than the lead exists on the same straight line at the same interval as the lead pitch, it is impossible to determine whether the lead terminal is a lead terminal or not based on only the detection result of one side. Therefore, there is a problem that recognition of a component such as QFN in which a terminal having no relation to various types of centering exists at a corner portion of the component is not stabilized by the conventional method.
[0006]
Accordingly, an object of the present invention is to provide a component recognition method and apparatus capable of improving the recognition accuracy of a component having a terminal irrelevant to centering at a component corner.
[0007]
[Means for Solving the Problems]
In order to solve the above problems, the present invention provides:
In a component recognition method of processing a captured image of a component and recognizing a posture of the component based on a terminal of the component detected by the image processing,
Processing the image of the captured part to obtain corner coordinates,
Determine whether there is a corner terminal based on the acquired corner coordinates,
If it is determined that a corner terminal exists, delete the corner terminal,
A configuration is adopted in which the posture of the component is recognized based on the terminal after the corner terminal has been deleted.
[0008]
In the present invention,
In a component recognition device that processes a captured image of a component and recognizes the orientation of the component based on the terminal of the component detected by the image processing,
An image processing device that processes an image of the captured component;
Means for deleting the corner terminal when it is determined that there is a corner terminal based on the corner coordinates obtained from the image processing,
There is also employed a configuration having means for calculating the center of the component and its inclination based on the terminal after the corner terminal has been deleted.
[0009]
In such a configuration, corner recognition unnecessary for component recognition not related to centering is deleted, and component recognition is performed. Therefore, component recognition accuracy can be improved, and component mounting accuracy can be improved.
[0010]
Deletion of a corner terminal can be performed by determining whether a corner terminal exists based on the terminal distance from the corner coordinates. If the corner terminal is close to the adjacent terminal depending on the part type, the corner terminal may not be recognized, so the image is scanned along the diagonal obtained from the corner coordinates and the corner terminal is detected by this scan I do. The corner terminals detected in this way are removed from the image by filling them.
[0011]
By performing the terminal filling process in this manner, terminals that are not desired to be detected can be deleted from the image, so that not only unnecessary terminals themselves are not detected, but also detection of surrounding terminals is not affected. Only necessary terminals can be accurately detected, and accurate component recognition can be performed.
[0012]
In addition, since the corner terminals can be reliably detected by diagonal scan, different component inspection, that is, inspection as to whether or not component recognition has been performed on the designated component, can be performed.
[0013]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, the present invention will be described in detail based on embodiments shown in the drawings.
[0014]
<Configuration of component mounting device and image processing device>
FIG. 1 is a block diagram showing a control configuration of a component mounting apparatus (component mounting machine) to which the present invention is applied. The component mounting apparatus includes a component (electronic component) 20 supplied from a component supply device (not shown). Has a head unit 3 provided with a suction nozzle 3a for sucking. The head unit 3 is configured to be movable in the XY-axis directions by being driven by an X-axis motor 11 and a Y-axis motor 12 driven by a controller 10 composed of a CPU. The head unit 3 is configured to be driven up and down by being driven in the Z-axis direction by the Z-axis motor 13 during component suction and component mounting, and the suction nozzle 3 a to be rotated about the nozzle axis by the θ-axis motor 14.
[0015]
Further, the component mounting apparatus is provided with an imaging device 5 having a lens 5a such as a CCD camera for imaging the component 20, and the head unit 3 is moved to the imaging device 5 after the component is sucked, and is illuminated by the illumination device 15. The illuminated component 20 is imaged by the imaging device 5. The captured image of the component is input to an image processing device 7 including a CPU 7a, an image memory 7b, an A / D converter 7c, and a component data memory 7d. The image processing device 7 fetches an image of the set area, processes the image according to a known algorithm, recognizes the suction posture of the component, and calculates a positional shift between the component center and the suction center and a suction angle shift. The controller 10 corrects this displacement when driving the X-axis, Y-axis, and θ-axis motors 11, 12, and 14, and mounts the component 20 on a board (not shown) that is being conveyed.
[0016]
The component mounting apparatus is provided with input devices such as a keyboard 21 and a mouse 22 for inputting component data. The generated component data is stored in a storage device 23 such as a hard disk, a flash memory, or the like. It can be stored in the component data memory 7d in the device 7. As the component data, data supplied from a host computer (not shown) connected to the component mounting apparatus may be stored in the storage device 23 or the component data memory 7d without passing through the input device. . In addition, a monitor (display) 24 is provided, and on this screen, component data, calculation data, and component images captured by the imaging device 5 can be displayed.
[0017]
FIG. 2 illustrates a state in which the component 20 a sucked by the suction nozzle 3 a is illuminated by the illumination device 15 and is imaged by the imaging device 5. The imaging device 5 captures an image of the set area by the control line 7d from the CPU 7a. The captured image is input to the image processing device 7 through a signal line 7e, converted into a digital signal through an A / D converter 7c, stored in an image memory 7b, subjected to image processing, and processed as a component. The posture is recognized. The image to be processed can be displayed on the monitor 24.
[0018]
<Part recognition flow>
FIG. 3 illustrates a flow of calculating a suction deviation by recognizing a component posture when the component is a QFN component, for example.
[0019]
First, as shown in FIG. 4, there are various types of QFN components. Therefore, the QFN components are classified according to the shape of the corner terminals, and are classified into types 0 to 3 according to the shapes of the corner terminals. In the future, if other corner terminal-shaped parts come out, we will respond by adding more types. In each of the following descriptions, a description will be given of component types (type, number of terminals, outer shape) suitable for the description.
[0020]
The component is represented by data describing the vertical / horizontal dimensions of the component, the location of the corner terminal, the shape of the corner terminal existing there, and the terminal information of each of the four sides. The terminal information includes the number of terminals, pitch, terminal length, terminal width, chipping information, terminal bending judgment rate, and the like. Using the above expression method, the position and shape of the corner terminal and the terminal information of each side are input using an input device such as the keyboard 21 or obtained from the host computer, and are obtained as part data. It is transmitted to the component data memory 7d in the image processing apparatus 7 in advance, and is managed for each component identification number (ID).
[0021]
At the time of execution of the recognition, the picked-up component is picked up by the image pickup device 5, the image is stored in the image memory 7b, the ID number of the component is specified, and the component data relating to the component is read from the component data memory 7d. It is read out and the recognition process shown in FIG. 3 is started.
[0022]
First, in step S1, scanning is performed from the outside to the inside of the image to detect external contacts of the component, and a circumscribing window W indicating a rough position of the component is set from the result. Subsequently, the lead row windows W1 to W4 are applied to each side (step S2), and an inspection area is set in an area where a lead exists. Then, for each of the four sides, a DOG filter (Marr &Posio's Difference-Of-Gaussian filter) is applied to the lead row window to detect the edge of the lead terminal, and using that, the lead end (“+” mark in the figure) is used. The coordinates are detected (steps S3 and S4). At this time, if the number of detected lines is less than the specified number, the recognition process is terminated at that point as an undernumber error.
[0023]
Subsequently, a lead line straight line connecting the lead tips is obtained from the detected lead tip coordinates. As shown in FIG. 5A, when a straight line is obtained by using all the detected leads, a shifted straight line H ′ occurs when a corner terminal or a lead on another side is detected. Therefore, in order to obtain an accurate lead line straight line, points that are not on the same straight line are deleted from the detected lead line coordinates.
[0024]
This method is illustrated in FIG. 6. First, as shown in FIG. 6A, the detection lead row is divided into two groups G1 and G2 on the left and right sides, and the lead longitudinal coordinates (y and Create a histogram of Then, in each group, an intermediate value is obtained from the obtained histogram, and a detection terminal having an intermediate value at the y coordinate is extracted. For example, assuming that the intermediate value is y4 in the group G1 and the intermediate value is y2 in the group G2, all points having the respective intermediate values are obtained. The obtained point is indicated by “+” in FIG. 6B, and then, as shown in FIG. 6C, the average value of the extracted terminals is obtained, and two points having the average value are obtained. R1 and R2 are determined, and a straight line passing through the two points R1 and R2 is defined as a reference straight line R as shown in FIG. Next, as shown in FIG. 6E, the detection terminal R3 that is separated from the reference straight line R by a certain distance or more is deleted.
[0025]
As described above, after eliminating points that are not on the same straight line, a least-squares straight line is obtained for the coordinates Rn of the remaining detection terminals as shown in FIG.
[0026]
Subsequently, this lead line straight line is obtained for each side, and the intersection of each lead line straight line is obtained as component corner coordinates C1 to C4. This corresponds to step S5 in FIG.
[0027]
Since there are terminals of various shapes at the component corners and such corner terminals L1 to L4 are not necessary for centering (positioning), when a corner terminal is detected at the component corner, it will be described in detail below. As described above, the corner terminal is deleted based on the corner coordinates C1 to C4 (step S6).
[0028]
Since the method of deleting the corner terminals differs for each type of component or for each terminal shape shown in FIG. 4, a different deletion process is performed for each type as described later. For example, in the case of a component belonging to type 3 as shown in FIG. 7A, when the leading end of the lead terminal L11 adjacent to the corner terminal L10 is too close, a lead edge is detected by applying a DOG filter. The two terminals are detected as one, and the lead end E10 may be detected with respect to the two terminals L10 and L11 as shown in FIG. 7B. In this case, the presence of the corner terminal L10 affects the detection of the lead terminal. On the other hand, in the case of a component belonging to type 1 as shown in FIG. 7C, if the tip of the lead terminal L21 adjacent to the corner terminal L20 has a certain distance, even if a DOG filter is applied. The two terminals are not detected as one. As shown in FIG. 7D, the lead tips E20, E21,. . . . . Is detected, and the presence of the corner terminal L20 does not affect the detection of the lead terminal.
[0029]
Therefore, when the corner terminal has a shape that does not affect the detection of the lead terminal as in types 1 and 2, it is determined whether or not the detection result includes the corner terminal by looking at the distance from the corner. Since the corner terminal exists at the corner portion of the component, when the terminal closest to the corner exists within a predetermined distance from the corner, it is determined that the terminal is a corner terminal. Since the corner terminal is unnecessary for centering, when it is determined that the terminal is a corner terminal, it is deleted from the detection result. If there is no corresponding terminal, it is not deleted. After the termination of the terminal, two diagonal lines are obtained from the corner coordinates, and scanning is performed on the diagonal line to check for the presence of a corner terminal. When inputting the component data, if the type or position of the component type is erroneously input, a difference from the terminal detection by scanning occurs, and a different component inspection can be performed. For example, if a corner terminal cannot be found for a corner where a corner terminal should exist from the component data, component recognition has been performed for a different component, and in such a case, an error is terminated. .
[0030]
On the other hand, when the corner terminal has a shape such as type 3 which affects the detection of the lead terminal, first, a diagonal scan is performed to detect the corner terminal. The contour of the corner terminal is detected, the corner terminal is painted out from the processed image, and the corner terminal is deleted. This is performed for each of the four corners, an image without corner terminals is created, and lead detection is performed again based on the image.
[0031]
The corner terminal deletion processing and the different part inspection performed in step S6 will be described later in detail for each type.
[0032]
As described above, after the corner terminal deletion process and the different component inspection are performed, a lead inspection is performed on each side to determine whether or not a defective terminal exists with respect to the remaining lead terminals detected (steps S7 and S8). If it is normal, the center and inclination of the component are calculated based on the lead terminal (step S9).
[0033]
<Corner terminal detection by diagonal scan>
As described above, since it is necessary to perform the image processing by filling the corner terminals depending on the type of the inspection and the type of the different parts, the detection of the corner terminals by diagonal scanning will be described below.
[0034]
First, a binarization threshold for detecting a corner terminal is set. As shown in FIG. 8, a point on the lead line straight line H "of each side and points in the vicinity thereof are sampled to be sample points, a density histogram of the sample points is created, and a discriminant analysis method is used. In the discriminant analysis method, assuming that a set of density values is divided into two classes (at least t and less than t) by a threshold value t, the discriminant analysis method This is a general threshold value selection method based on the idea that the parameter t is determined so as to obtain the best separation.For example, the threshold value for the upper left corner is obtained for two sides forming the corner, that is, the upper side and the left side. And the average value of the threshold values.
[0035]
Subsequently, the component image is processed according to the flow shown in FIG. 9 to check whether or not a corner terminal exists.
[0036]
First, since the corner coordinates of the component image have been obtained (step S5 in FIG. 3), a diagonal line connecting the diagonal corners is obtained (step S11), and the scan direction and range for detecting the terminal are determined. It is set (step S12). As shown in FIG. 10A, the scanning direction is made different depending on the diagonal inclination and the corner position in order to increase the detection sensitivity. The scan range is defined by ranges M1 and M2 each having a length of 1/4 of the distance between corners before and after the corner coordinates, and a scan start point m1 and an end point m2 are set.
[0037]
After setting the scan direction, the scan start point, and the end point, the scan is started (step S13). As shown in FIG. 10C, when the diagonal slope is gentler than 45 °, points on the diagonal and upper and lower points are scanned from the scan start point to the end point (y-direction scan), and the diagonal slope is determined. If the angle is steeper than 45 °, the point on the diagonal line and the left and right points are scanned in the x-axis direction.
[0038]
In step S14, a scan result is determined. When a point equal to or larger than the binarization threshold value for the corner terminal is detected, the diagonal scan ends, and the process proceeds to step S15 for acquiring a terminal contour. Conversely, if no point equal to or higher than the binarization threshold is found even after scanning to the scan end point, it is determined that there is no corner terminal.
[0039]
When the terminal is detected in step S14, the contour is detected from the detected point (step S15). FIG. 11 shows a state in which this contour is obtained. In FIG. 11A, Q1 is a point detected by the diagonal scan, and eight neighboring areas excluding the detection point Q1 are investigated in numerical order. Since the contour is detected at Q2 or more because the threshold is exceeded, this is set as the next contour point. Subsequently, as shown in FIG. 11 (B), the investigation around 8 around Q2 is started, and the same processing is repeated until returning to the first start point. Note that the investigation direction near 8 is determined by the diagonal scan direction.
[0040]
The contour points detected in this way are shown in FIG. 11C, and the information is stored in an array in the order of detection. Subsequently, in steps S16 and S17, edge pair coordinates in the X direction and edge pair coordinates in the y direction are obtained from the contour point array, respectively (FIGS. 11D and 11E). In step S18, the center of gravity Xg, Yg of the detection terminal is determined from the edge pair coordinates (FIGS. 11F and 11G). The calculation of the center of gravity is obtained according to the following equation.
[0041]
Lsum (y) = Edge2 (y) -Edge1 (y) +1
Ladr_sum (y) = Lsum (y) × (Edge1 (y) + Edge2 (y)) / 2
SUM = ΣLsum (y)
Ymom = Σ (Lsum (y) × y)
Xmom = @ Ladr_sum (y)
Xg = Ymom / Sum
Yg = Xmom / Sum
If the obtained center of gravity of the detection terminal is near the diagonal line, the detected terminal can be regarded as a corner terminal. Conversely, if the center of gravity is away from the diagonal, the lead terminal may have been erroneously detected, and it is determined that there is no corner terminal.
[0042]
The detection of a corner terminal by such a diagonal scan can be used for the inspection of a different part and the filling of the corner terminal performed according to the type as described below.
[0043]
<Corner terminal deletion and inspection of different parts>
Here, the details of the corner terminal deletion and the different part inspection performed in step S6 of FIG. 3 will be described. As described above, this processing differs depending on the component type as shown in FIG. 4, and is therefore performed for each type.
[0044]
Since the type 0 component has no corner terminal, the corner terminal deletion processing is not performed. However, in order to inspect different parts, the above-described diagonal scan is performed to check for the presence of a corner terminal. If no corner terminal is detected in all four corners, it is determined to be normal. If a corner terminal is detected even in one corner, there is a possibility that a different part is present. Therefore, the different part inspection is performed again after the lead inspection.
[0045]
FIG. 12 illustrates the flow of the corner terminal deletion and the different part inspection processing for the type 1 and type 2 parts.
[0046]
First, since the lead terminals have been detected, the pitch is obtained (step S21), and the distance between the detection leads is obtained. In the case of a component having no lead chipping as shown in FIG. 13A, the lead pitch Pn is obtained by averaging the distance between the lead terminals other than the lead terminals L30 and L31 at both ends. On the other hand, as shown in FIG. 13 (B), when there are chipped leads L32 and L33, the minimum inter-lead distance is obtained from other than the terminal pitch at both ends, and the other inter-lead distance is divided by the minimum inter-lead distance. When the minimum distance between leads is P1, the distance between leads is represented by an integer. From the lead missing information of the component data, a lead distance sequence in which one pitch is P1 is determined.
[0047]
If there is an inconsistency between the detected state of the distance between leads and the lead information obtained from the missing lead information, it is determined that the parts are different and error processing is performed. In the normal case, the average value is obtained by taking into account the number of chips at the distance between leads other than the distance between the leads at both ends, and this is used as the lead pitch.
[0048]
Next, when the distance between the leads at both ends is not within ± 30% of the detection pitch, the leads at the ends are regarded as corner terminals and deleted. However, when the lead bending judgment rate is set to be less than ± 30%, the detection pitch is set within ± lead bending judgment rate [%] of the detection pitch. For example, as shown in FIG. 14A, the distance P3 between the left-most lead terminal L30 and its adjacent lead terminal is larger than the detection pitch + 30%, so the terminal L30 is deleted as a corner terminal. On the other hand, since the distance P4 from the lead terminal L31 at the right end is within the detection pitch + 30%, this terminal is not deleted. This state is shown in FIG. As described above, when the pitches at both ends are mismatched and do not match, an abnormal process is performed (step S22), and one terminal is deleted.
[0049]
Next, if the detection terminal exists within one pitch from the corner coordinates, it is regarded as a corner terminal and deleted (step S23). However, this processing is not performed when the deletion is performed due to the abnormal distance between the leads at both ends. Therefore, this processing is not performed on the left end L30 that has already been deleted. If there are a plurality of detection terminals within one pitch, the detection terminal closest to the corner coordinates is deleted.
[0050]
Note that the two values used for the deletion determination, the allowable value of the distance between the leads at both ends, and the distance from the corner can also be externally specified from component data or the like. In the description, default values are used.
[0051]
Finally, the above-described diagonal scan is performed (step S24), and the presence or absence of four-sided corner terminals is checked (steps S25 and S27). If a corner terminal is detected at a corner designated as having a corner terminal, it is determined to be normal. If the presence or absence of a corner terminal is different from the designation even at one corner, there is a possibility that a different part is used. In order to inspect the different parts again after the lead inspection, the position of the corner where the corner terminal was present is remembered (step S26).
[0052]
The corner terminal deletion processing of the type 3 component is performed according to the flow shown in FIG. In the case of type 3, since the corner terminal affects the detection of the lead terminal, a diagonal scan is performed, the corner terminal is detected, the corner terminal is painted, and the corner terminal is deleted.
[0053]
Therefore, first, the above-described diagonal scan is performed (step S31), and the presence or absence of a corner terminal is checked (step S32). If the terminal does not exist, it is impossible to paint the corner terminal in the next process, so the process is skipped and the process proceeds to the next corner. However, the absence of the corner terminal at the corner designated as having the corner terminal may indicate a different part. Therefore, in order to perform the different component inspection again after the lead inspection, the position of the corner where the corner terminal is not detected is remembered (step S33).
[0054]
If it is determined by diagonal scanning that a corner terminal is present, the corner terminal is painted (step S34). This filling process is performed as shown in FIG. 16. First, scanning is performed between two points X1 and X2 which are half a pitch outside the edge point in the x direction to obtain the positions of the points X3 and X4 having the lowest density. I do. This is defined as valley coordinates. The area between the left and right valley coordinates is painted so that the density change is constant.
[0055]
When the filling with the edge pair in the x direction is completed, the same processing is performed in the y direction. Since the density of a point on the edge has already been changed, a valley coordinate is obtained by scanning between a point outside one edge of the edge and a point outside a half pitch. The density is set so that the density change between the upper and lower valley coordinates is constant. It is applied only when the density to be newly applied is lower than the density already filled. With such processing, the corner terminal L can be painted out, and the image of the corner terminal can be deleted.
[0056]
When the filling of all the detected corner terminals is completed (step S35), lead detection is performed again (step S36), and the center and inclination of the component are calculated according to the detection result.
[0057]
By performing the terminal filling process in this manner, terminals that are not desired to be detected can be deleted from the image, so that not only unnecessary terminals themselves are not detected, but also detection of surrounding terminals is not affected. Only necessary terminals can be accurately detected, and accurate component recognition can be performed.
[0058]
Note that the terminal filling process described with respect to type 3 may be performed not only on components like type 3 but also on types 1 and 2 having other corner terminals.
[0059]
Further, in the diagonal scan performed in the corner terminal deletion and the different part inspection, there is a possibility that there is a different part, and the position of the corner terminal in which an error has occurred is stored (step S26 in FIG. 12). In step S33 in FIG. 15, the stored terminal is scanned again to determine whether or not there is a corner terminal. This process is illustrated in FIG.
[0060]
After the lead inspection is completed, the coordinates of the tip of the lead are determined, and as shown in FIG. 17A, the coordinates of the corner coordinates E40 and the coordinates of the edges E41 and E42 of the two lead terminals L41 and L42 adjacent to the corner coordinates E40. I understand. A rectangle A connecting these two lead terminals and the corner coordinates is obtained. As shown in FIG. 17A, in the range of the rectangle A, some of the lead terminals L41 and L42 enter the side where the lead exists. In order to avoid the lead terminal portion, as shown in FIG. 17B, a scan for finding a point where no terminal exists from the side to the inside is performed from the lead end side toward the lead base. As shown in FIG. 17C, a rectangle A ′ in which the previously obtained range is narrowed is determined according to the point where the innermost lead terminal does not exist.
[0061]
Then, scanning is performed from the corner coordinate side within the range A ′ set in FIG. 17C, and if a point equal to or greater than the binarization threshold value for the corner terminal is found in a block having a certain number of pixels or more, the corner terminal is determined. Judge that it exists. If a certain number or more of points equal to or larger than the threshold value are not found even after scanning the entire area A ', it is determined that there is no corner terminal. The number of pixels that are determined to be a certain number or more blocks that are regarded as corner terminals is set for each resolution. Thereby, the detection accuracy of the corner terminal can be improved.
[0062]
【The invention's effect】
As described above, according to the present invention, since the component recognition is performed by removing a corner terminal that is unnecessary for the component recognition not related to the centering, the component recognition accuracy can be improved, and the component mounting accuracy can be improved. it can.
[0063]
Further, in the present invention, if the corner terminal is close to the adjacent terminal depending on the type of component, the corner terminal may not be recognized, so the image is scanned along a diagonal line obtained from the corner coordinates, A corner terminal is detected by scanning, and the detected corner terminal is painted out to remove the corner terminal. By performing the terminal filling process in this manner, terminals that are not desired to be detected can be deleted from the image, so that not only unnecessary terminals themselves are not detected, but also detection of surrounding terminals is not affected. Only necessary terminals can be accurately detected, and accurate component recognition can be performed.
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a schematic configuration of a component mounting apparatus.
FIG. 2 is a configuration diagram showing a configuration of a component imaging device and an image processing device.
FIG. 3 is a flowchart showing a flow of component recognition.
FIG. 4 is an explanatory diagram showing types of components.
FIG. 5 is an explanatory diagram showing a flow for obtaining a lead row straight line.
FIG. 6 is an explanatory diagram showing a flow of obtaining a reference straight line for obtaining a lead line straight line.
FIG. 7 is an explanatory diagram for explaining an effect of a corner terminal on terminal detection depending on a component type.
FIG. 8 is an explanatory diagram showing sampling points for obtaining a binarization threshold.
FIG. 9 is a flowchart illustrating a flow of detecting whether or not a corner terminal exists in a component by diagonal scan.
FIG. 10 is an explanatory diagram illustrating a diagonal scan method.
FIG. 11 is an explanatory diagram showing a flow for obtaining a contour of a detected corner terminal.
FIG. 12 is a flowchart showing a flow of a corner terminal deletion process and a different component inspection for components of types 1 and 2;
FIG. 13 is an explanatory diagram showing a state in which a lead pitch is determined.
FIG. 14 is an explanatory diagram showing a flow of corner terminal deletion.
FIG. 15 is a flowchart showing a flow of a corner terminal deletion process and a different component inspection for a type 3 component.
FIG. 16 is an explanatory diagram showing a corner terminal deletion process for a type 3 component.
FIG. 17 is an explanatory diagram showing a process for increasing corner terminal detection.
[Explanation of symbols]
5 Imaging device
7 Image processing device
7b Image memory
7d parts data memory

Claims (4)

In a component recognition method of processing a captured image of a component and recognizing a posture of the component based on a terminal of the component detected by the image processing,
Processing the image of the captured part to obtain corner coordinates,
Determine whether there is a corner terminal based on the acquired corner coordinates,
If it is determined that a corner terminal exists, delete the corner terminal,
A component recognition method characterized by recognizing a posture of a component based on a terminal after removing a corner terminal. 2. The component recognition method according to claim 1, wherein a corner terminal is detected by scanning along a diagonal line, and whether or not recognition of the specified component is recognized is performed based on the detection result. In a component recognition device that processes a captured image of a component and recognizes the orientation of the component based on the terminal of the component detected by the image processing,
An image processing device that processes an image of the captured component;
Means for deleting the corner terminal when it is determined that there is a corner terminal based on the corner coordinates obtained from the image processing,
Means for calculating the component center and its inclination based on the terminal after removing the corner terminal,
A component recognition device comprising: 4. A method according to claim 3, further comprising the step of: providing means for scanning an image along a diagonal line determined from the corner coordinates, detecting a corner terminal by the scanning, and checking whether or not the specified component is recognized based on the detection result. Component recognition device as described.

Images