JP2009116419A - Outline detection method and outline detection device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an outline detection method and device, allowing detection of an outline of an object. <P>SOLUTION: Each constituent pixel constituting an acquisition image from a camera is classified in each luminance value, and a histogram indicating a distribution of the number of the pixels in each luminance value is formed (S1). A change of the number of the pixels according to the luminance value is approximated by a Bezier curve obtained from the histogram (S2), the minimum detected on a lowest luminance side among the minimums each indicating a trough in the Bezier curve is extracted (S3), and a threshold is set based on the luminance value in the minimum (S4). Each constituent pixel of the acquisition image is binarized into the constituent pixel on a lower luminance side than the threshold or the constituent pixel on a higher luminance side than the threshold by use of the threshold to form a binarized data image (S5), and the outline of a chip image is detected from the binarized data image (S6). <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a contour detection method and a contour detection apparatus for detecting a contour of an object.

  Conventionally, when bonding a chip such as a semiconductor to a lead frame, a bonding apparatus has been used (for example, Patent Document 1). In the bonding apparatus, the picked-up chip is transferred to a bonding point for bonding. It is configured.

  The bonding apparatus is configured to set a wafer ring, and a wafer sheet is stretched on the wafer ring. A wafer is stuck on the wafer sheet, and the wafer is diced. Thus, dicing streets are formed vertically and horizontally on the wafer, and chips are separated by the streets.

  In the bonding apparatus in which the wafer ring is set, an image of the wafer is acquired, and the contour and position of the chip are recognized by performing image processing on the wafer image.

As this recognition method, pattern matching is known, and the outline and position of the target chip are recognized by comparing the chip image in the acquired wafer image with a template registered in advance.
JP 2006-128231 A

  However, in such a conventional technique, teaching work for inputting a chip image and registering it as a template has to be performed in advance, which is difficult.

  The present invention has been made in view of such a conventional problem, and an object of the present invention is to provide a contour detection method and a contour detection device capable of detecting the contour of an object.

  In order to solve the above-mentioned problem, in the contour detection method according to claim 1 of the present invention, in the contour detection method for detecting the contour of the object image from the acquired image obtained by imaging the object, the constituent pixels of the acquired image are changed. Histogram forming stage for forming a histogram showing the distribution of the number of pixels for each luminance value by classifying for each luminance value, and approximating the change in the number of pixels according to the luminance value with a Bezier curve obtained from the histogram, A threshold setting step for setting a threshold based on the luminance value at the minimum value detected on the lowest luminance side among the minimum values indicating valleys in the Bezier curve, and binary for each component pixel of the acquired image A contour detection method comprising: a binarized data forming step for generating binarized data and a contour detecting step for detecting a contour of the object image from the binarized data.

  That is, when detecting the contour of the target image from the acquired image obtained by capturing the target object, the constituent pixels of the acquired image are classified for each luminance value, and a histogram indicating the distribution of the number of pixels for each luminance value is formed. To do. Next, the change in the number of pixels according to the luminance value is approximated by a Bezier curve obtained from the histogram, and among the minimal values indicating valleys in the Bezier curve, the minimal value detected on the lowest luminance side A threshold is set based on the luminance value at.

  At this time, a region with a large number of pixels on the lower luminance side, that is, the lowest luminance side than the trough indicated by the minimum value set in this threshold indicates the background of the outer peripheral portion of the object, An area where the number of pixels is higher on the higher luminance side than the valley indicated by the minimum value set as the threshold indicates the part of the object.

  Then, by using this threshold value, each component pixel of the acquired image is binarized into a component pixel on the luminance side lower than the threshold value and a component pixel on the luminance side higher than the threshold value to form binary data. The contour of the object image is detected from the binarized data.

  At this time, the constituent pixels on the luminance side lower than the threshold value indicate the background of the object, and the constituent pixels on the luminance side higher than the threshold value indicate the part of the object. For this reason, the said outline is detected by extracting the location from which the binarization value of an adjacent component pixel differs.

  In the contour detection method according to claim 2, the object is a chip provided on a wafer sheet, and in the contour detection stage, a boundary between the chip and a street surrounding the chip is used as the contour. To detect.

  That is, the object is a chip provided on a wafer sheet, and a boundary between the chip and a street surrounding the chip is detected as a contour.

  Thus, by acquiring and processing an image of the wafer, the outline of the chip is detected, and the chip position is detected based on this outline.

  In the contour detection device according to claim 3 of the present invention, in the contour detection device for detecting the contour of the object image from the acquired image obtained by imaging the object, the constituent pixels of the acquired image are classified for each luminance value. Histogram forming means for forming a histogram indicating the distribution of the number of pixels for each luminance value, and a change in the number of pixels according to the luminance value is approximated by a Bezier curve obtained from the histogram, Threshold setting means for setting a threshold based on the luminance value at the minimum value detected on the lowest luminance side among the minimum values indicating, and binarization by binarizing each constituent pixel of the acquired image with the threshold Binarized data forming means for forming data; and contour detecting means for detecting the contour of the object image from the binarized data.

  That is, when detecting the contour of the target image from the acquired image obtained by capturing the target object, the constituent pixels of the acquired image are classified for each luminance value, and a histogram indicating the distribution of the number of pixels for each luminance value is formed. To do. Next, the change in the number of pixels according to the luminance value is approximated by a Bezier curve obtained from the histogram, and among the minimal values indicating valleys in the Bezier curve, the minimal value detected on the lowest luminance side A threshold is set based on the luminance value at.

  At this time, a region with a large number of pixels on the lower luminance side, that is, the lowest luminance side than the trough indicated by the minimum value set in this threshold indicates the background of the outer peripheral portion of the object, An area where the number of pixels is higher on the higher luminance side than the valley indicated by the minimum value set as the threshold indicates the part of the object.

  Then, by using this threshold value, each component pixel of the acquired image is binarized into a component pixel on the luminance side lower than the threshold value and a component pixel on the luminance side higher than the threshold value to form binary data. The contour of the object image is detected from the binarized data.

  At this time, the constituent pixels on the luminance side lower than the threshold value indicate the background of the object, and the constituent pixels on the luminance side higher than the threshold value indicate the part of the object. For this reason, the said outline is detected by extracting the location from which the binarization value of an adjacent component pixel differs.

  Further, in the contour detection apparatus according to claim 4, the object is a chip provided on a wafer sheet, and the contour detection means uses a boundary between the chip and a street surrounding the chip as the contour. To detect.

  That is, the object is a chip provided on a wafer sheet, and a boundary between the chip and a street surrounding the chip is detected as a contour.

  Thus, by acquiring and processing an image of the wafer, the outline of the chip is detected, and the chip position is detected based on this outline.

  As described above, in the contour detection method according to claim 1 and the contour detection apparatus according to claim 3 of the present invention, each component pixel of the acquired image is set to the background of the object by using the threshold value obtained from the Bezier curve. And a component pixel indicating the part of the object can be binarized and the contour of the object image can be detected from the binarized data.

  Accordingly, it is possible to detect the contour of the target object without performing the teaching work of inputting an image of the target object and registering it as a template as in the prior art, and it is possible to automatically detect the contour by using the apparatus. Further, by using this contour, the position of the target object can be extracted.

  In the contour detection method according to claim 2 and the contour detection apparatus according to claim 3, the boundary between the chip provided on the wafer sheet and the street surrounding the chip can be detected as the contour. Based on this, the chip position can be detected.

  Therefore, when the chip is picked up from the wafer sheet in the bonding apparatus, the presence / absence of the chip at the pick-up position and the state of the chip can be detected, and the center position of the chip can be picked up accurately.

  Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a contour detection device 1 according to the present embodiment, and the contour detection device 1 is a device that realizes a contour detection method for detecting the contour of an object.

  In FIG. 1, a configuration in which the contour detection device 1 is applied to a bonding device 11 is shown as an example. The bonding device 11 includes a chip 14 on a wafer sheet 13 stretched on a wafer ring 12. Is a device for picking up and transferring to a lead frame (not shown) for bonding.

  A wafer 21 is stuck and fixed on the wafer sheet 13, and the wafer 21 is diced. Dicing streets 22,... Are formed in the wafer 21 vertically and horizontally, and a plurality of chips 14,.

  As a result, a plurality of chips 14,... As target objects are formed on the wafer sheet 13, and each of the chips 14,. The boundary between the chip 14 and the streets 22 surrounding the chip 14 forms the contour of the chip 14.

  The bonding apparatus 11 includes a camera 32 that acquires an image of the wafer 21 on the wafer sheet 13 set on the wafer table 31. The camera 32 captures the wafer 21 on the wafer sheet 13. The acquired acquired image 33 (see FIG. 2) is configured to be output to the image processing unit 34. By using the camera 32 and the image processing unit 34 provided in the bonding apparatus 11, the contour detection apparatus 1 of the present embodiment is configured.

  The image processing unit 34 is configured around a microcomputer, and by processing the acquired image 33 from the camera 32, each component pixel constituting the acquired image 33 is shown in FIG. The histogram 41 is configured so as to be classified for each luminance value of “0” to “255” and to show the distribution of the number of pixels for each luminance value.

  Further, as shown in FIG. 3, the change in the number of pixels according to the luminance value is approximated by a Bezier curve 51 obtained from the histogram 41, and local values 61 and 62 indicating valleys in the Bezier curve 51 are approximated. Of these, the threshold value is set based on the luminance value at the minimum value 61 detected on the lowest luminance side. As shown in FIG. A binarized data image 71 as binarized data is formed by binarization, and contours 74,... Formed between the chip image 72 and the street image 73,. Configured to detect.

  The operation of the contour detection apparatus 1 according to the present embodiment having the above configuration will be described with reference to the flowchart shown in FIG.

  That is, when the microcomputer of the image processing unit 34 operates in accordance with a program stored in the ROM and the microcomputer executes the contour detection process, the acquired image 33 is input from the camera 32 that has captured the image of the wafer 21, and the acquired image As shown in FIG. 2, each of the constituent pixels constituting the pixel 33 is classified into luminance values “0” to “255”, and the distribution of the number of pixels for each luminance value “0” to “255” is shown. A histogram 41 is formed (S1).

  Then, as shown in FIG. 3, the change in the number of pixels according to the luminance value is approximated by a Bezier curve 51 obtained from the histogram 41 (S2), and a minimum value 61 indicating a valley in the Bezier curve 51 is obtained. 62, the minimum value 61 detected on the lowest luminance side is extracted (S3), and a threshold value is set based on the luminance value at the minimum value 61 (S4).

  More specifically, the Bezier curve 51 can be shown as a function comprising a polynomial, and the local minimum values 61 and 62 are obtained by calculating this function with a microcomputer. Then, the lowest value obtained from each of the minimum values 61 and 62 is selected, and the luminance value “143” at the minimum value 61 is set as the threshold value.

  The luminance value “143” is a value that substantially matches the luminance value at the valley position that appears on the lowest luminance side in the histogram 41. In the present embodiment, the luminance value “143” at the minimum value 61 is set as the threshold value. However, the present invention is not limited to this, and a value calculated based on the luminance value “143” is set as the threshold value. It may be set to.

  A region on the lower luminance side than the valley portion indicated by the minimum value 61 set to the threshold value “143”, that is, a mountain portion where the number of pixels is the highest on the lowest luminance side is constituted by the wafer sheet 13 on the outer periphery of the chip 14 .., And represents the background of the chip 14. On the other hand, the region on the higher luminance side than the valley indicated by the minimum value 61 set to the threshold “143” indicates any part of the chip 14. At this time, the chip 14 is formed with a wafer portion formed of silicon and a pattern formed on the wafer, and the two peaks appearing on the higher luminance side than the minimum value 61 are It is assumed that the wafer portion and the pattern portion are shown.

  Next, by using this threshold value “143”, each constituent pixel of the acquired image 33 is changed to a constituent pixel on the lower luminance side than the threshold value “143” and the threshold value “143” as shown in FIG. A binarized data image 71 is formed as binarized data by binarizing with the higher-luminance component pixels (S5), and the contour 74 of the chip image 72 is detected from the binarized data image 71. (S6).

  At this time, the constituent pixels on the lower luminance side than the threshold “143” indicate the streets 22,... That are the background of the chip 14, and the constituent pixels on the higher luminance side than the threshold “143” are the chips. Any one of 14 sites is shown. Therefore, the chip image 72 showing the chip 14 is shown in white, the street image 73 showing the streets 22,... Is shown in black, and the binarized values of adjacent constituent pixels are Different locations between white and black can be extracted, and the boundary between the chip 14 and the streets 22 surrounding the chip 14 can be detected as the contour 74.

  At this time, in this binarized data image 71, black dot portions 75,... Are displayed due to noise or the like, but it is configured to recognize a part surrounded by black lines as a chip image 72. ing.

  Thus, by using the threshold value “143” obtained from the Bezier curve 51, each component pixel of the acquired image 33 is replaced with a component pixel indicating the street 22,. It is possible to binarize and classify into 14 constituent pixels. As a result, the microcomputer 74 can easily detect the contours 74,... Of the chip 14 that are difficult to detect from the grayscale acquired image 33 by using the binarized data image 71. be able to.

  Therefore, the contour 74 of the chip 14 can be detected without performing a teaching operation for inputting a chip image and registering it as a template as in the prior art. By using the contour detection device 1, the contour 74, ... can be automatically detected.

  Further, by using the contours 74,..., The position of the chip 14 can be detected based on the contours 74,.

  Therefore, when the chip 14 is picked up from the wafer sheet 13 in the bonding apparatus 11, it is possible to detect the presence / absence of the chip 14 at the pick-up position, chipping of the chip 14, and the like. The center can be picked up accurately.

  Then, since the position of the chip 14 can be grasped, in the post-processing of the contour detection process, the image within the contour 74,... At the grasped chip position can be acquired and automatically registered as a template.

It is a block diagram which shows one embodiment of this invention. It is a figure which shows the acquired image and histogram of the embodiment. It is a figure which shows the acquired image of the same embodiment, a histogram, and a Bezier curve. It is a figure which shows the acquisition image and binarization data image of this Embodiment. It is a flowchart which shows the operation | movement of the embodiment.

Explanation of symbols

1 Contour Detection Device 13 Wafer Sheet 14 Chip 22 Street 33 Acquired Image 34 Image Processing Unit 41 Histogram 51 Bezier Curve 61 Minimal Value 71 Binary Data Image 72 Chip Image 73 Street 74 Contour

Claims (4)

In a contour detection method for detecting a contour of an object image from an acquired image obtained by imaging the object,
A histogram forming step of classifying the constituent pixels of the acquired image for each luminance value and forming a histogram indicating the distribution of the number of pixels for each luminance value;
The change in the number of pixels according to the brightness value is approximated by a Bezier curve obtained from the histogram, and based on the brightness value at the minimum value detected on the lowest brightness side among the minimum values indicating valleys in the Bezier curve. A threshold setting stage for setting the threshold,
A binarized data forming step of binarizing each component pixel of the acquired image with the threshold value to form binarized data;
A contour detection step of detecting a contour of the object image from the binarized data;
A contour detection method comprising: The object is a chip provided on a wafer sheet,
2. The contour detection method according to claim 1, wherein, in the contour detection step, a boundary between the chip and a street surrounding the chip is detected as the contour. In a contour detection apparatus for detecting a contour of a target object image from an acquired image obtained by capturing the target object,
Histogram forming means for classifying the constituent pixels of the acquired image for each luminance value and forming a histogram indicating the distribution of the number of pixels for each luminance value;
The change in the number of pixels according to the brightness value is approximated by a Bezier curve obtained from the histogram, and based on the brightness value at the minimum value detected on the lowest brightness side among the minimum values indicating valleys in the Bezier curve. Threshold setting means for setting the threshold,
Binarized data forming means for binarizing each component pixel of the acquired image with the threshold value to form binarized data;
Contour detecting means for detecting the contour of the object image from the binarized data;
A contour detection apparatus comprising: The object is a chip provided on a wafer sheet,
The contour detection apparatus according to claim 1, wherein the contour detection unit detects a boundary between the chip and a street surrounding the chip as the contour.

Images