JP2008071780A - Method of detecting film pasting status - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To correctly detect the pasting status of a translucent film in a chip. <P>SOLUTION: Image information of a chip 1 is acquired in step S1, an image to be recognized is extracted on the basis of binarized data of the image information in step S2, and a coarsely detected film region of the translucent film 2 is extracted in step S3. A plurality of divided contour lines 8 divided by a pattern 5 in a right side of the coarsely detected film region are acquired in step S4, a plurality of smooth contour lines 9 are acquired by feature extraction in step S5, a folded portion of coordinates is removed in step S6, and the smooth contour lines 9 are coupled to acquire an entire contour line 10 in step S7. A contour line 11 complying to standards is acquired in step S8, and an approximated right side 12 is acquired on the basis of the contour line 11 complying to the standards in step S9. The steps S4-S9 are repeatedly executed for each side of the coarsely detected film region to acquire approximated bottom, left and top sides 13, 14 and 15, and the position and profile of the translucent film 2 are decided on the basis of each of approximated sides 12-15 in step S10. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a film sticking state detection method for detecting a sticking state of a translucent and predetermined film stuck on the main surface of an object having a pattern on at least the main surface side. The object is specifically a chip-shaped electronic component or a circuit board.

  Conventionally, in order to mount a chip-like electronic component (hereinafter referred to as “chip”) such as a semiconductor chip on a circuit board made of a printed circuit board, a lead frame or the like, there is a method using an adhesive member made of a die bond paste, an adhesive sheet or the like. in use. In recent years, a package in which a plurality of chips are stacked on a circuit board (so-called stack type package) has been adopted in response to demands for downsizing electronic devices and increasing the capacity of storage elements. ing. Also in this stack type package, an adhesive sheet, that is, an adhesive film (hereinafter referred to as “film” as appropriate) is used together with the die bond paste for ease of handling. Specifically, a film is stuck on a chip, and another chip is mounted on the chip (see, for example, Patent Document 1).

  By the way, in order to detect the pasting state of the film on the circuit board or chip which is the object to which the film is stuck, the object to which the film is to be stuck is photographed and the obtained image is binarized. It was. And based on the obtained binarized image, it was detected whether or not the film was mounted or whether or not the film was mounted at an appropriate position.

  However, according to the conventional technology described above, there are the following problems. In particular, in a stack type package, the film is thinned in response to a request for thinning the package, and accordingly, a film having translucency (translucent film) is used. Specifically, by using a translucent film, the pattern of the target chip or circuit board is captured in the image, and by applying a binarization process, the translucent film sticking position The problem is that sufficient accuracy cannot be obtained. Moreover, even if it replaces with a binarization process and performs a multi-value process, since the pattern of a target object is seen through a semi-transparent film, a translucent film and the pattern of a target object will be mixed. For these reasons, it has been difficult to properly extract an image of a translucent film that is a recognition target.

JP 2001-291821 A (page 4, FIG. 1, FIG. 6)

  The problem to be solved by the present invention is that when a translucent film is used, the image of the translucent film is appropriately extracted by being affected by the pattern of the chip or circuit board that is the object. It is difficult to detect the state of film sticking.

  Reference numerals in parentheses in the following description are described for the purpose of facilitating the comparison between the terms in the description and the components shown in the drawings. Further, these symbols do not mean “interpreting the terms in the description limited to the components shown in the drawings”.

  In order to solve the above-described problems, a method for detecting a film application state according to the present invention is a translucent and predetermined shape that is applied to a main surface of an object (1) having a pattern (5) on at least the main surface side. A method for detecting a film sticking situation for detecting a sticking situation of a film (2) having a step of obtaining image information of an object (1) to which the film (2) should be stuck, and 2 A step of obtaining binarized data by binarization, a step of extracting a recognition target image consisting of an image of the film (2) from image information by performing segmentation based on the binarized data, and based on the recognition target image Extracting the rough detection film region consisting of the region corresponding to the image of the rough detection of the film (2) by performing feature extraction, and inspecting at the outer edge of the rough detection film region Generating a plurality of contour lines (8) separated by the pattern (5) with respect to the outer edge by generating a window (7) and performing multi-valued edge detection processing in the inspection window (7); Obtaining a plurality of smooth contour lines (9) from which small waveforms have been removed by performing feature extraction based on the split contour lines (8); removing a fold of coordinates in the plurality of smooth contour lines (9); A step of obtaining the overall contour line (10) with respect to the outer edge by combining a plurality of smooth contour lines (9) and a rough detection film region and the overall contour line (10) are compared to determine a predetermined one of the overall contour lines (10). The process of obtaining the standard contour line (11) satisfying the standard, and the position of the film (2) by performing an approximation based on the standard contour line (11) Together and a step of determining the Jo, the object (1) is characterized in that it is a chip-like electronic component (1) or the circuit board.

  Moreover, the detection method of the film sticking condition which concerns on this invention is the above-mentioned detection method. WHEREIN: The predetermined shape of a film (2) is a quadrilateral, In the process of extracting a rough detection film area | region, each side which a quadrilateral has In the step of generating four primary straight lines indicating the contour line (11) within the standard, parameters based on the four primary straight lines are used as a predetermined standard, and the position and shape of the film (2) are determined. The process is characterized in that the positions of the respective corners (16) of the film (2) are determined by approximating the contour lines (11) within the standard by respective linear lines.

  According to the present invention, binarized data is obtained based on the image information of the object (1) which is composed of the chip-shaped electronic component (1) or the circuit board and has the pattern (5) on the main surface side. The following steps are sequentially performed based on the value data. The recognition target image is extracted by segmentation, the rough detection film region is extracted by feature extraction, the multi-valued edge detection is performed in the inspection window (7) generated at the outer edge of the rough detection film region, and the pattern (5) is used for the outer edge. Acquisition of a plurality of divided contour lines (8) obtained, acquisition of a plurality of smooth contour lines (9) by feature extraction based on the plurality of divided contour lines (8), removal of coordinate wrapping in the plurality of smooth contour lines (9), a plurality of Acquisition of the overall contour line (10) related to the outer edge by combining the smooth contour lines (9) of the two, acquisition of the contour line (11) within the standard by comparing the rough detection film region and the overall contour line (10), and within the standard It is each process of determination of the position and shape of a film (2) by performing the approximation based on a contour line (11). Thereby, when the pattern (5) which a chip | tip or a circuit board has is taken in into the image through the film (2) which has translucency, the influence of those patterns (5) is excluded, and a film (2 ) Position and shape can be determined. Therefore, even if the film (2) on the object (1) is translucent, the presence / absence of film sticking and the film sticking situation can be accurately detected.

  Further, according to the present invention, the predetermined shape of the film (2) is a quadrilateral, and each corner (16) of the film (2) is based on four primary lines indicating each side of the quadrilateral. Determine the position. Thereby, the position and shape of the film (2) having a quadrilateral outer shape can be determined. Therefore, in the case where the film (2) on the object (1) is quadrilateral and translucent, the presence / absence of film sticking and the film sticking situation can be accurately detected.

  In step S1, image information of the chip (1) to which the translucent film (2) having a rectangular planar shape is attached is acquired. In step S2, the image information is binarized to obtain binarized data, and a recognition target image is extracted by segmentation based on the binarized data. In step S3, the rough detection film region of the translucent film (2) is extracted by morphology processing and feature extraction. In step S4, multivalued edge detection is performed in the inspection window (7) generated on the right side which is a part of the outer edge of the rough detection film region, and a plurality of cut contour lines (figure divided by the pattern (5) with respect to the right side ( 8) is acquired. In step S5, a plurality of smooth contour lines (9) are obtained by feature extraction based on the plurality of divided contour lines (8). In step S6, the return of coordinates in the plurality of smooth contour lines (9) is removed. In step S7, a plurality of smooth contour lines (9) are combined to obtain an overall contour line (10) related to the outer edge. In step S8, by comparing the rough detection film region with the entire contour line (10), a contour line (11) satisfying a predetermined standard is obtained. In step S9, the approximate right side (12) is acquired by performing approximation based on the contour line (11) within the standard. After that, the approximated lower side (13), left side (14), and upper side (15) are obtained by repeating steps S4 to S9 for each side of the rough detection film region of the translucent film (2). In step S10, the position and shape of the translucent film (2) are determined based on the approximated sides (12 to 15).

  Example 1 of the method for detecting a film application state according to the present invention will be described with reference to FIGS. In addition, in order to make it easy to understand, any figure is typically abbreviate | omitted or exaggerated suitably. In addition, portions not related to the description of the present invention are enclosed in a rectangular frame and are not shown.

  FIG. 1 is a flowchart (flow chart) showing a method for detecting a film sticking state according to the present invention, omitting loop processing. FIG. 2 is a sample diagram showing an example of image information of a chip to which a translucent film is attached. FIG. 3 is a sample diagram showing a result of binarizing and segmenting the image information of FIG. FIG. 4 is a sample diagram showing a rough detection film region of a translucent film extracted by performing morphological processing on the image information of FIG. 3 to extract a recognition target image regarding the translucent film and then performing feature extraction. Up to here, each process until the rough detection film region of the translucent film is extracted by image processing is described.

  Moreover, the following figures are diagrams for explaining each process until the position and shape of the film are determined by image processing based on the rough detection film region of the translucent film. FIG. 5 is divided by a pattern obtained by generating an inspection window on the right side of the portion corresponding to the translucent film in the rough detection film region of FIG. 4 and performing multi-valued edge detection processing in the inspection window. It is a sample figure which shows several division | segmentation contour lines. FIG. 6 is a sample diagram showing a plurality of smooth contour lines from which small waveforms have been removed, obtained by performing feature extraction based on the plurality of cut contour lines in FIG. FIG. 7 shows the overall contour lines for the outer edge (in this case, the right side) of the translucent film, obtained by removing the folds of the coordinates in the plurality of smooth contour lines of FIG. 6 and then combining the plurality of smooth contour lines. It is a sample figure. FIG. 8 is a sample diagram showing in-standard contours that satisfy a predetermined standard, obtained by comparing the rough detection film region and the overall contours in the overall contour lines in FIG. 7. FIG. 9 is a sample diagram showing approximate contour lines obtained by performing a first-order straight line approximation based on the standard contour lines in FIG. FIG. 10 is a sample diagram showing the results obtained by acquiring the approximate contour lines of FIG. 9 for the four sides of the rough detection film region of the translucent film and then detecting the four sides and the four corners of the translucent film. . 5 to 10, the periphery of the translucent film is shown.

  First, in step S1 shown in FIG. 1, with the translucent film attached to the main surface of the chip, the chip including the translucent film is photographed using a camera. Thereby, as shown in FIG. 2, the image information of the chip 1 can be acquired in a state where the translucent film 2 is stuck on the main surface of the chip 1. In this embodiment, since the translucent film 2 is cut so that each corner is 90 °, the planar shape of the translucent film 2 is a quadrilateral with each corner being 90 °, that is, a rectangle. (Including squares).

  Here, the chip 1 will be described. On the main surface side of the chip 1, an electrode 3 for transmitting and receiving an electrical signal between the chip 1 and the outside thereof, and a wiring 4 for transmitting and receiving an electrical signal inside the chip 1 are provided. Yes. The electrode 3 and the wiring 4 together form a pattern 5. A protective film (not shown) is provided on the pattern 5 as necessary. Thereby, the appearance (brightness etc.) of the pattern 5 in the image information differs depending on the presence or absence of the protective film on the pattern 5.

  Next, in step S2 shown in FIG. 1, binarization is performed using an appropriate threshold value based on the image information of FIG. Thereby, binarization data is acquired about the chip | tip 1 with which the translucent film 2 was affixed on the main surface. Subsequently, in step S2, segmentation is performed based on the binarized data. Thereby, the recognition target image which consists of the image of the translucent film 2 is extracted from the image information shown by FIG. As a result, as shown in FIG. 3, a region including the recognition target image of the translucent film 2 can be acquired.

  Next, in step S3 shown in FIG. 1, a rough detection film comprising a region corresponding to an image obtained by roughly detecting the translucent film 2 by sequentially performing morphology processing and feature extraction based on the image of FIG. Extract regions. As a result, as shown in FIG. 4, an image including the rough detection film region of the translucent film 2 can be acquired. Here, the accuracy of the acquired image is restricted by the accuracy of the pixels when binarization is performed. Further, an error occurs when the rough detection film region is approximated to a rectangle. In addition, an overlapping portion 6 consisting of a portion overlapping the outer edge of the translucent film 2 in the pattern 5 is taken into the rough detection film region. Thereby, the rough detection film region does not accurately reflect the size and shape of the translucent film 2. Therefore, when the rough detection film region of the translucent film 2 shown in FIG. 4 is used as it is, it is difficult to accurately detect the pasting state of the translucent film 2.

  Next, in step S4 shown in FIG. 1, an inspection window 7 is generated on the right side of the rough detection film region of the translucent film 2 shown in FIG. In the inspection window 7, multi-valued edge detection processing is performed using gray value information. Thereby, as shown in FIG. 5, a plurality of divided contour lines 8 divided by the overlapping portion 6 in the pattern 5 can be obtained with respect to the outer edge of the rough detection film region (the right side of the translucent film 2 in FIG. 5). it can.

  Next, in step S5 shown in FIG. 1, feature extraction is performed on the plurality of cut contour lines 8 shown in FIG. 5 using, for example, roundness, length, and height. As a result, the contour edge formed of a small waveform in the plurality of divided contour lines 8 is removed. As shown in FIG. 6, a plurality of smooth contour lines 9 from which small contour edges are removed can be acquired.

  Next, in step S6 shown in FIG. 1, the folding of the coordinates is removed from the plurality of smooth contour lines 9 shown in FIG. As a result, it is possible to obtain a plurality of smooth contour lines from which the folding of coordinates is removed.

  Next, in step S7 shown in FIG. 1, a plurality of smooth contour lines from which the folding of coordinates has been removed are combined. Thereby, as shown in FIG. 7, with respect to the outer edge (right side in FIG. 7) of the translucent film 2, it is possible to obtain the entire contour line 10 indicating the outer edge. Here, the reason for combining a plurality of smooth contour lines is to determine the position / shape of the right side of the translucent film 2 by increasing edge information when determining the position / shape of the right side of the translucent film 2 in a later step. This is because the accuracy can be increased.

  Next, in step S8 shown in FIG. 1, the coordinates of the entire contour line 10 are applied to a linear equation obtained by approximating the right side of the translucent film 2 in the rough detection film region. As a result, the rough detection film region and the entire contour line 10 are compared. Then, among the data constituting the entire contour line 10, by adopting a portion having coordinates existing within a predetermined parameter set value, that is, within a predetermined standard, as shown in FIG. Can be acquired. Here, examples of the predetermined parameter include the slope and intercept of the linear line obtained by approximating the right side of the translucent film 2, the distance from the linear curve, and the like. If the process of obtaining the contour line 11 within the standard is described in detail, first, the entire contour line 10 in the part having coordinates existing within a predetermined standard is left. Next, in the part which exists between those remaining parts in the whole contour line 10, in other words, in the part which deviated from a predetermined standard, those remaining parts are connected with a straight line.

  Next, in step S9 shown in FIG. 1, first, a linear straight line approximation is performed based on the within-standard contour line 11 shown in FIG. As a result, as shown in FIG. 9, the right side 12 that is partial and approximated can be acquired.

  Next, although illustration in FIG. 1 is omitted, the process of steps S4 to S9 shown in FIG. 1 is subsequently performed for the other outer edge of the translucent film 2, that is, the lower side, the left side, and the upper side ( Loop processing). As a result, as shown in FIG. 10, the approximated lower side 13, the approximated left side 14, and the approximated upper side 15, which are each partial, can be acquired. And based on each approximated sides 12-15, the intersection of each side in the translucent film 2, ie, the four corners 16 which the translucent film 2 has, can be detected. Thereby, the outer edge (in other words, external shape) 17 of the affixed translucent film 2 is detectable. Therefore, in step S10 shown in FIG. 1, the position and shape of the translucent film 2 can be determined. Further, the center of the translucent film 2 can be detected based on the four corners 16 of the translucent film 2. When determining the position and shape of the translucent film 2, it is necessary to determine the relative positional relationship of the translucent film 2 in the chip 1. In this case, two known points on the chip 1, for example, two corners in a diagonal positional relationship are used.

  As described above, according to the method for detecting the state of film sticking according to the present embodiment, the influence of the pattern 5 provided on the main surface side of the chip 1 is eliminated, and the translucent sticking to the main surface is performed. The position and shape of the film 2 can be determined. Therefore, even when the film on the chip 1 that is the object is the translucent film 2, it is possible to accurately detect the presence / absence of the film sticking and the film sticking situation.

  Hereinafter, Example 2 of the detection method of the film sticking condition which concerns on this invention is demonstrated with reference to FIG.1, FIG.2 and FIG.4. In the present embodiment, the planar shape of the translucent film 2 shown in FIG. 2 is a shape other than a rectangle. Also in this case, the position and shape of the translucent film 2 can be determined in the same manner as the flowchart shown in FIG. Specifically, when the inspection window 7 is generated in step S4 of FIG. 1 (see FIG. 4), the inspection window 7 finely divided according to the planar shape of the translucent film 2 is generated. Further, according to the inspection window 7 that is finely divided, the contour lines that are finely divided are also acquired for various contour lines. In the first embodiment, four approximate sides are acquired. In the second embodiment, a larger number of approximate sides are acquired.

  As described above, according to this embodiment, the position and shape of the translucent film 2 can be determined even when the planar shape of the translucent film 2 is a shape other than a rectangle. Examples of the planar shape of the translucent film 2 include a shape obtained by chamfering at least one corner of a rectangle, a triangle, a pentagon or more polygon. Also, for a semi-transparent film 2 having a planar shape composed of a shape in which a part of a circle is connected to one side of a rectangle, a shape in which a part of a circle is connected to two opposite sides of the rectangle (oval type), etc. Also, this embodiment can be applied.

  The method for detecting the state of film application according to the present invention may be used for the purpose of inspecting the state of application of the translucent film to be applied to the main surface of the circuit board when the chip is mounted on the circuit board. it can. In this case, the position and shape of the translucent film can be determined by appropriately extracting the image of the translucent film without being affected by the pattern provided on the main surface side of the circuit board. In this case, two alignment marks provided in advance on the circuit board are used.

  In addition, even if the translucent film is not applied for some reason, the method for detecting the film application state according to the present invention indicates that the translucent film to be applied does not exist on the main surface of the object. Can be detected. Therefore, the defect that the translucent film is not stuck can be detected.

  In addition, the present invention is not limited to the above-described embodiments, and may be arbitrarily combined, changed, or selected as necessary without departing from the spirit of the present invention. It can be done.

It is a flowchart (flowchart) which abbreviate | omits a loop process and shows the detection method of the film sticking condition which concerns on this invention. It is a sample figure which shows an example of the image information of the chip | tip with which the translucent film was stuck. It is a sample figure which shows the result of binarizing and segmenting the image information of FIG. It is a sample figure which shows the rough detection film area | region of the translucent film extracted by performing the morphology process of the image information of FIG. 3, extracting the recognition target image regarding a translucent film, and performing feature extraction after that. A plurality of divided contour lines divided by a pattern obtained by generating an inspection window on the right side of the portion corresponding to the translucent film in the rough detection film region of FIG. 4 and performing multi-valued edge detection processing in the inspection window FIG. FIG. 6 is a sample diagram showing a plurality of smooth contour lines obtained by performing feature extraction based on a plurality of cut contour lines in FIG. 5 and from which small waveforms are removed. It is a sample figure which shows the whole contour regarding the outer edge (in this case, right side) of a translucent film acquired by removing the folding | turning of the coordinate in several smooth contour lines of FIG. 6, and combining several smooth contour lines after that. . FIG. 8 is a sample diagram showing in-standard contours satisfying a predetermined standard obtained by comparing the rough detection film region and the overall contour lines in the overall contour lines of FIG. 7. FIG. 9 is a sample diagram showing approximate contour lines obtained by performing a first-order straight line approximation based on the standard contour lines in FIG. 8. It is a sample figure which acquires the approximate contour line of Drawing 9 about four sides in the rough detection film field of a semi-transparent film, detects four sides and four corners of a semi-transparent film after that, and shows the result.

Explanation of symbols

1 chip (object)
2 Translucent film (film)
3 Electrode 4 Wiring 5 Pattern 6 Overlap 7 Inspection Window 8 Divided Contour 9 Smooth Contour 10 Overall Contour 11 In-Standard Contour 12 Approximated Right Side 13 Approximated Lower Side 14 Approximated Left Side 15 Approximated Upper Side 16 Corner 17 Outer Edge

Claims (2)

A method for detecting a film sticking situation for detecting a sticking situation of a translucent film having a predetermined shape to be stuck on the principal surface in an object having a pattern on at least the principal surface side,
Obtaining image information of the object to which the film is to be attached;
Binarizing the image information to obtain binarized data;
Extracting a recognition target image consisting of an image of the film from the image information by performing segmentation based on the binarized data;
Extracting a rough detection film region composed of a region corresponding to an image obtained by rough detection of the film by performing feature extraction based on the recognition target image;
Generating a detection window at an outer edge of the rough detection film region, and obtaining a plurality of divided contour lines divided by the pattern with respect to the outer edge by performing multi-valued edge detection processing in the detection window;
Obtaining a plurality of smooth contours from which small waveforms have been removed by performing feature extraction based on the plurality of cut contours;
Removing the folding of coordinates in the plurality of smooth contour lines;
Obtaining overall contour lines for the outer edge by combining the plurality of smooth contour lines;
Obtaining an in-standard contour that satisfies a predetermined standard among the entire contour lines by comparing the rough detection film region and the entire contour lines;
Determining the position and shape of the film by approximating based on contour lines within the standard,
The object is a chip-like electronic component or a circuit board. In the detection method of the film sticking condition of Claim 1,
The predetermined shape of the film is a quadrilateral,
In the step of extracting the rough detection film region, four primary lines indicating each side of the quadrilateral are generated,
In the step of obtaining the contour lines within the standard, parameters based on the four primary straight lines are used as the predetermined standard,
A method for detecting a film sticking state, wherein, in the step of determining the position and shape of the film, the position of each corner of the film is determined by approximating the contour lines within the standard by respective linear lines.

Images