JP2006155248A - Method for detecting gravity center of object - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a gravity center detecting method capable of correctly detecting a gravity center of an object without being influenced by the shape of the object or image capturing conditions. <P>SOLUTION: A gravity center is calculated by totaling moments at each pixel position, setting the number of pixels to 1 for a pixel (245-264) of the object, to a value between 0 and 1 (0.854...), for a pixel (265) near an edge, and to 0 for a pixel in background, and totalling the number of the pixels of the object. According to this configuration, since the number of pixels near the edge of the object is set to a value smaller than the number of pixel corresponding to one pixel, the moment near the edge can be calculated by a unit (sub-pixel unit) finer than resolution of one pixel, which makes it possible to correctly detect the gravity center of the object. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a method for detecting the center of gravity of an object, and more specifically, detects the center of gravity of an object by subjecting an image of the object such as an electronic component to image processing in sub-pixel units (units finer than pixel resolution). Regarding the method.

  Conventionally, in an electronic component mounting machine, an electronic component picked up by a component supply device is imaged by a CCD camera, and the image is processed to obtain the center of gravity (component center) of the electronic component and correct the suction posture. Electronic components are mounted on circuit boards.

  Here, a method of capturing an image of an object such as an electronic component and detecting the position of the center of gravity of the image pattern (graphics) is known, for example, in Patent Document 1, and the center of gravity coordinates (p, q) are the values of the image. The density value of each pixel is used as a weight (in the case of a binary image, the weight is set to 1), and can be expressed by the following equation using a moment.

p = M (1,0) / M (0,0), q = M (0,1) / M (0,0)
M (0, 0): Area of figure M (1, 0): Moment with respect to Y-axis, M (0, 1): Moment with respect to X-axis In the electronic component mounting machine, when calculating the center of gravity of the electronic component, the electronic component , A predetermined window area is set, the window area is scanned along a number of lines, and an image of the electronic component is used as multivalued image data. For example, in FIG. Multi-valued data (0 to 255) is shown at pixel positions (X coordinate) from “240” to “270” with respect to (261st line; Y coordinate). As shown in FIG. 4B, this multi-valued image data is “0” for data smaller than a predetermined threshold (= 109), and “255” for data larger than that. As shown in FIG. 4C, the image data from which noise has been removed is divided into a background and an object, and the first and last coordinates (edges) of the object are extracted (FIG. 4 ( D)), the number of pixels (corresponding to the weight of the pixel) is set to 1 for the pixel of the object, and the number of pixels is set to 0 for the background pixel. The sum of moments (number of pixels at the position × coordinate value) and the sum of the number of pixels (number of pixels) in “240” to “270” is obtained (right side in FIG. 5). Then, this is obtained for all lines “255” to “285”, and the center of gravity is obtained from the sum of the moments obtained for each line and the sum of the number of pixels of the object (right side in FIG. 7). .
7-113975

  However, the conventional detection of the center of gravity using binary image data is because the edge portion on the image data of the captured object is processed in units of one pixel, that is, the weight corresponding to the pixel is a value corresponding to one pixel. Since the center of gravity is calculated with (the number of pixels being 1), there is a problem in that sufficient accuracy cannot be obtained depending on the object and the imaging state at present when accuracy in micron units is required.

  Referring to FIG. 6, the detection of the center of gravity based on binary image data is performed when the density value is higher than the threshold value (near the edge) even when the density value is close to the threshold value (near the edge). Even in the center of the object (FIG. 6A), the number of pixels is equivalent (FIG. 6B), and the change in the density value near the edge is not utilized, and an error occurs in the position of the detected center of gravity. It may occur.

  In addition, even in the case of image data taken sequentially under the same conditions, the density value at each position changes each time. As a result, the threshold value changes, or the threshold value changes because the specified window area is changed, etc. There is a problem that an error (variation) due to a change in threshold value becomes large at the binary centroid.

  The present invention solves such problems, and its object is to provide a center-of-gravity detection method that can accurately detect the center of gravity of an object without being affected by the shape of the object and imaging conditions. To do.

The present invention
The image data in the specified window area is multi-valued for each pixel, the pixel position is determined by setting the pixel above the predetermined threshold as the pixel of the object, and the number of pixels is set for the pixel of the object. In the centroid detection method for detecting the centroid of the object from the sum of moments at the pixel position and the sum of the number of pixels of the object,
For each pixel other than the edge portion of the object, the number of pixels is set to the first numerical value, and for the pixels of the edge portion, the number of pixels is smaller than the first numerical value and is larger than 0 The sum of moments and the sum of the number of pixels are obtained by setting numerical values, and the center of gravity of the object is detected from each of the obtained sums.

  In the present invention, the threshold value is set as the first threshold value, a second threshold value that is larger than the first threshold value by a predetermined value is set, and the interval between the first and second threshold values is set. A pixel having a value of is set as a pixel in the edge portion. The number of pixels is set to 1 for pixels greater than or equal to the second threshold, the number of pixels is set to 0 for pixels less than the first threshold, and the first and second thresholds are set. For pixels between values, the number of pixels is set to a value between 1 and 0, and the center of gravity is detected.

  According to the present invention, since the number of pixels at the edge portion of the object is set to a value smaller than the number of pixels corresponding to one pixel, the moment near the edge portion or the edge is sub-unit (subpixel) smaller than the resolution of one pixel. The center of gravity of the object can be accurately obtained without being affected by the shape of the object or the imaging conditions of the object.

  In the present invention, the number of pixels is set to 1 for the pixel of the object, and the object is calculated from the sum of moments at each pixel position of the object and the sum of the number of pixels of the object (corresponding to the area of the object). In this case, the number of pixels of the edge portion of the object adjacent to the background portion is set to a value between 1 and 0, and the center of gravity is detected by increasing the resolution of the portion. The present invention will be described in detail with reference to the accompanying drawings.

  FIG. 1 schematically shows configurations of an electronic component mounting machine 20 according to an embodiment of the present invention and an image processing apparatus 12 that processes a captured image of the electronic component.

  In the figure, the suction nozzle 1 is controlled by a machine control device 13 and moves in the X and Y axis directions to suck an electronic component 2 supplied from a component supply device (not shown). The machine control device 13 moves the suction nozzle 1 that sucks the electronic component 2 to the imaging position, where the electronic component 2 illuminated by the illumination device 3 is replaced with a standard CCD camera 4 or The image is picked up by one of the high-resolution CCD cameras 5.

  As will be described later, the captured image of the electronic component is subjected to image processing in the image processing device 12, component recognition is performed, the component center (center of gravity) and the suction angle are calculated, and the suction posture of the component is detected. Thereafter, the electronic component 2 is moved to the position of the substrate 14 by the machine control device 13, the suction posture is corrected, and the electronic component 2 is mounted at a predetermined position on the circuit board 14.

  A substrate mark is formed on the substrate 14, and this substrate mark is photographed by a substrate recognition camera 15 attached to a head provided with the suction nozzle 1, and the mark image is processed by the image processing device 12 to obtain a substrate position. Is detected. If the substrate 14 is misaligned, the misalignment is corrected, so that the electronic component 2 is mounted on the substrate 14 with high accuracy.

  The image processing apparatus 12 stores an image of an object such as an A / D converter 6 that converts an image from the cameras 4, 5, and 15 into a digital value, and an electronic component 2 or a board mark that has been converted into a digital image. The memory 7 includes a memory 7, a working memory (RAM) 8, a gravity center detection program 9, a CPU 10, and an interface 11. The captured image of the electronic component 2 and the mark of the board 14 is stored in the image memory 7. The CPU 10 reads and executes the center-of-gravity detection program 9 and designates a window having a predetermined size including an image of an object such as the electronic component 2 or the board mark stored in the image memory 7 and is designated. The image in the window area is processed to detect the center of gravity of the object.

  If the center of gravity position is detected normally, the center of gravity position is displayed on the monitor 16 and the result is returned to the machine control device 13 via the interface 11. If it is not detected normally, it returns an error. When the machine control device 13 normally receives the result of the center of gravity position, the machine control device 13 calculates a correction amount for correcting the suction posture or the substrate position according to the information, and controls the movement amount of the suction nozzle 1 according to the correction amount. As described above, the electronic component 2 is mounted on the substrate 14 at a predetermined position.

  Next, the flow of image processing for detecting the center of gravity will be described with reference to FIG.

  First, the image processing apparatus 12 controls the designated camera 4, 5, or 15 to capture an image of an object such as a mark formed on the electronic component 2 or the substrate 14 (step S 1). Digitized by the D converter 6, stored as multivalued image data in the image memory 7, and displayed on the monitor 16.

  One line (261 (Y coordinate value) lines) of the multivalued image data stored in the image memory 7 is shown in FIG. 4A, and “240” to “270” shown above are The pixel position (X coordinate value) along the one line is shown, and the density data at the pixel position is multivalued as image data “0” to “255”.

  Subsequently, the center-of-gravity detection program 9 creates a histogram in the designated window area for the data in the image memory 7. And the threshold value which distinguishes a target object and a background part by a discriminant analysis method is calculated (step S2).

  This calculation method is shown in FIG. 3, which shows a histogram showing how many pixels having the density (frequency) for each density value of the image in the window area. . Min is the minimum density value of the image, Max is the maximum density value, and if the threshold value that separates the object and the background portion is Thr, the density value up to the maximum density value Max above the threshold Thr is the target. The density value of the object M2 is the density value of the background portion M1 that is less than the threshold Thr and reaches the minimum density value Min.

Here, the threshold value Thr is changed from the minimum density value Min to the maximum density value Max, and the lower limit threshold value ThrMin corresponding to the average density value of the background portion M1 at each threshold value Thr and the average density of the object M2 An upper threshold value ThrMax corresponding to the value is obtained, and the variance D between the background part M1 and the object M2 is set, where PMin is the number of pixels of the background part M1 and PMax is the number of pixels of the object M2.
D = PMin × PMax × {(ThrMin−ThrMax) squared}
And a threshold value Thr that maximizes the variance is obtained.

  FIG. 3 shows the threshold value Thr, the upper limit threshold value ThrMax, and the lower limit threshold value ThrMin when the variance becomes maximum in this way.

  In this embodiment, the threshold value Thr and the upper limit threshold value ThrMax are used to detect the center of gravity. However, when the density value of the edge portion of the target object is close to the threshold value, the characteristics of the target object, the imaging conditions, etc. Depending on the case, it is possible to make adjustments such as using an intermediate value between the lower limit threshold value ThrMin and the threshold value Thr or a value of 3/4.

  Next, in step S3, threshold value acquisition determination is performed, and the maximum density value in the specified window area is lower than the reference value of the threshold value that was passed as information when the execution of the centroid detection process was instructed. In this case, it is determined that there is only a background in the designated window area, and the center-of-gravity detection program 9 notifies the machine control device 13 of an error via the interface 11 without acquiring the threshold value by the discriminant analysis method ( Step S9) The process is terminated.

  Also, filtering is performed on the image data (step S4), and when there is noise in the designated window area, the multi-value image data is binarized with the threshold value acquired in step S2, and AND− Data subjected to the OR filter is created and stored in the work memory 8. The binarized data is shown in FIG. 4B, where the threshold value is “109” and the image data in FIG. 4A is less than “109” and “0” ( (Background), the data more than that is binarized as “255” (object). Further, as shown in FIG. 4C, the image data at the pixel positions “244”, “266”, and “267” is changed to the data “255” of the object, and the data at the pixel position “270” is obtained by the filtering process. Is converted into background data “0”.

  Subsequently, in step S5, the center of gravity is detected. In this case, when detecting the X coordinate of the center of gravity, the density value and the discriminant analysis method for each pixel position are detected for each line in the Y direction of the designated window region, and when detecting the Y coordinate, for each line in the X direction. Each pixel position is classified by comparing the obtained two kinds of threshold values, that is, the threshold value Thr and the upper limit threshold value ThrMax under the following conditions.

In this classification, the density value of the pixel position of interest is N,
If N <Thr, “0”
"1" if ThrMax ≤ N
If Thr ≦ N <ThrMax, “2”
And a classification value is assigned to the pixel.

  When the threshold value is “109” and the upper limit threshold value is “170”, the multi-valued image data shown in FIG. 4A is classified by the above algorithm as shown in FIG. The threshold value less than “109” is classified as “0”, the upper threshold value “170” or more is classified as “1”, the threshold value “109” or more and less than the upper threshold value “170” is classified as “2”. . At this time, referring to the corresponding position of the filtered data, even if the classification is “1” or “2”, if the filtered data (C) is “0”, the classification is “0”. To do. The case of the pixel position “270” is this example.

  Subsequently, when data for one line is created, the inside of the first position (first edge) and the last position (second edge) of classification 1 is replaced with classification 1. This is because when there is a portion having a low density value inside the object, the inside is handled as the object. This state is shown in FIG. 4F, and the pixel positions “253”, “256”, “259” to “262” are replaced with “1” as data inside the object.

  Next, looking at the data for one line, the number of pixels for classification 1 pixels is set to 1, the number of pixels for classification 0 is set to 0, and the number of pixels for classification 2 is set to 0. Is set to a pixel number (P) in units of sub-pixels that is finer than one pixel, by assuming that the pixel is an edge portion of the object or a pixel near the edge.

P = (N− (Thr−1)) / (ThrMax− (Thr−1))
The value (number of pixels) calculated at this time is 0.0 <P <1.0. As shown in FIG. 4G, the pixel position “265” has the number of pixels as edge data. It is calculated as "0.854 ...".

  Here, the number of pixels 1 means that the number of pixels at the pixel position is 1 (1 pixel), the number of pixels 0 means no pixels, and the number of pixels is “0.0” and “ When it is between 1.0, it corresponds to the number of sub-pixels obtained by subdividing one pixel according to its value. In other words, the number of pixels indicates the weight of the pixel at the pixel position when calculating the center of gravity. For pixels other than the edge portion of the object, the weight of the pixel is set to 1, The background pixel is assigned a weight of 0, and the edge pixel is assigned a weight between “0.0” and “1.0”.

  In this way, in the example of FIG. 4F, since the pixel position “265” is the classification 2, the moment M is expressed by the equation M = P × C (P = 0.854. 261), and for other pixel positions, P is set to 0 or 1 according to the classification value to obtain moments for each pixel position, and the sum (sum) of moments for one line and the number of pixels Calculate the sum (total) of (pixels). The result is shown in FIG. In the figure, for comparison, what is calculated based on the image data of FIG. 4D according to the prior art is shown on the right side.

  FIG. 6 shows data for all lines in the X direction and Y direction in the designated window area. The pixel positions “240” to “270” (from 255 lines to 285 lines (Y direction)) are shown. Data in the X direction) are shown. In order to avoid complexity, only the data of 259 to 261 lines are shown. FIG. 6A is multi-value image data, FIG. 6B is an example of the prior art, and FIG. 6C is a pixel sub-pixel of the edge portion or the vicinity of the edge as described above according to the present invention. In the 259th and 260th lines, the pixel at the pixel position “266” is converted into a subpixel, and in the 261 line, the pixel at the pixel position “265” is converted into a subpixel, and the number of pixels is a value between 1 and 0. ing.

  Subsequently, for the sum of moments and the sum of pixels obtained for each line, the sum (sum) of all lines in the specified window area is obtained, and MSm is the sum of moments and PSum is the sum of pixels. (G) is obtained. An example of this is shown in FIG. The right side of FIG. 7 shows an example in the case of the conventional example, that is, the case of using FIG.

  The above is the Y coordinate value of the center of gravity, but the X coordinate value can also be obtained by calculating the moment with respect to the Y axis. That is, the moment (the product of the number of pixels and the X coordinate value) at each pixel position for one line is obtained, the sum is obtained and obtained for all lines, and the sum of moments for all lines (MSum) is obtained. The X coordinate value is obtained from the total sum (PSum) of the total number of pixels.

  FIG. 8 shows an example in which the designated window area is changed with the same image data, the threshold value is “99”, and the upper threshold value is “168”.

  In the case where the density value matches the threshold value or the upper threshold value, for example, the following conditions can be used.

0: N ≦ Thr
1: ThrMax ≦ N
2: Thr <N <ThrMax
In this case, the number of pixels in category 2 is as follows.

P = (N-Thr) / (ThrMax-Thr)
In this way, by changing the formula for obtaining the number of pixels depending on conditions, how to handle when the density value matches the threshold (whether it is handled as the background side or the object side), etc. It can be adapted to the image processing specifications from time to time.

  In this way, the quality of the obtained center of gravity detection is determined in step S6. When the center of gravity cannot be detected, such as when the object does not fit within the window or when the designated window area has only one pixel, the center of gravity detection program 9 is sent to the machine controller 13 via the interface 11. An error is notified (step S9) and the process is terminated.

  If the center of gravity is detected successfully, the center of gravity is displayed on the image displayed on the monitor 16 (step S7), and the center of gravity detection program 9 notifies the machine controller 13 of the center of gravity through the interface 11. (Step S8), the process ends.

  In the above-described embodiment, the number of pixels other than the edge portion of the object is set to 1, but it may be a positive numerical value n (real number) such as 1.5 or 2. In this case, the number of pixels P in the edge portion may be n times the number of pixels obtained by the above-described formula.

It is the block diagram which showed the structure of the component mounting machine and the image processing apparatus. It is the flowchart which showed the flow which calculates | requires the gravity center of a target object. It is explanatory drawing which showed the method of calculating | requiring a threshold value by discriminant analysis. It is explanatory drawing which showed the transition of the data for 1 line of a designation | designated window area | region. It is a table | surface figure which showed the example which calculates | requires the moment for 1 line. It is explanatory drawing which showed the transition of the data for all the lines of a designation | designated window area | region. It is the table | surface which showed the example which calculates | requires the moment for all the lines. It is explanatory drawing which showed the transition of the data for all the lines at the time of changing a threshold value. FIG. 9 is a table showing an example of obtaining moments for all lines in the example of FIG. 8.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Adsorption nozzle 2 Electronic component 7 Image memory 9 Center-of-gravity detection program 12 Image processing apparatus 13 Machine control apparatus 14 Board | substrate

Claims (5)

The image data in the specified window area is multi-valued for each pixel, the pixel position is determined by setting the pixel above the predetermined threshold as the pixel of the object, and the number of pixels is set for the pixel of the object. In the centroid detection method for detecting the centroid of the object from the sum of moments at the pixel position and the sum of the number of pixels of the object,
For each pixel other than the edge portion of the object, the number of pixels is set to the first numerical value, and for the pixels of the edge portion, the number of pixels is set to a second value larger than 0 with a value smaller than the first numerical value. A method for detecting the center of gravity of an object, wherein a sum of the moments and a sum of the number of pixels are set to numerical values, and the center of gravity of the object is detected from each of the obtained sums.   The threshold value is set as a first threshold value, a second threshold value that is larger than the first threshold value by a predetermined value is set, and has a value between the first and second threshold values. The center-of-object center-of-gravity detection method according to claim 1, wherein a pixel is the pixel of the edge portion.   3. The method for detecting the center of gravity of an object according to claim 1, wherein the first and second threshold values are obtained by a discriminant analysis method.   The number of pixels is 1 for pixels greater than or equal to the second threshold, the number of pixels is 0 for pixels less than the first threshold, and between the first and second thresholds. 4. The method for detecting the center of gravity of an object according to claim 1, wherein the number of pixels is set to a value between 1 and 0. 5.   5. The method for detecting the center of gravity of an object according to claim 4, wherein the number of pixels for the pixels between the first and second threshold values is obtained based on the first and second threshold values.