JP4053032B2 - Foreign object detection method, foreign object detection program, and foreign object detection device - Google Patents

Description

The present invention relates to a foreign object detection method, a foreign object detection program, and a foreign object detection apparatus that detect the presence or absence of a mixed foreign object by processing an X-ray image of an inspection object.

  In the production line of food and various industrial products, foreign matter detection using an X-ray image has been conventionally performed in order to eliminate foreign matters. For example, Patent Literature 1 and Patent Literature 2 describe conventional techniques in detail. When such a technique is applied to an inspection object such as a can having a high X-ray absorption rate at the peripheral edge of the inspection object, the peripheral edge of the can is often masked as a foreign object detection target. For example, Patent Document 3 describes in detail a peripheral edge masking technique.

FIG. 2 is a flowchart for explaining a conventional typical foreign object detection method for canned objects. The X-ray image input in step S100 is subjected to masking processing in step S101 with the peripheral portion of the object to be excluded as a foreign object detection target, and an image in which the foreign object is emphasized is generated in step S102. Processing is performed to detect the presence or absence of mixed foreign matter.

Various image filters can be applied to the generation of an enhanced image of foreign matter, but in order to apply an image filter that emphasizes the peripheral portion more than the foreign matter, the peripheral portion must be effectively masked. It is desirable that the area that is not subject to foreign object detection by the masking process is the minimum necessary in order to reduce the sensitivity although the second foreign object detection is performed for this area.

  Japanese Patent Application Laid-Open No. 2003-228561 describes a method for reducing the area that is not subject to foreign object detection. It does not change and does not match the actual shape of the peripheral edge. In particular, there is a problem in the masking process of the joint portion in the state where the inspection objects are lined up at an interval as in the X-ray image example of FIG.

As a masking processing method for eliminating such a problem, a blob (single lump) in contact with the background of an object to be inspected as shown in FIG. It is possible to extract the blob in contact with the other blob and use the blob that touches the background as a mask pattern. However, the foreign matter adjacent to the peripheral portion is included in the blob that touches the background and is masked. There is a problem.

W098 / 11456 gazette Japanese Patent Laid-Open No. 2001-307069 JP 2001-281173 A

  The conventional foreign matter detection method for canned objects has a problem in the method of masking the peripheral portion, and the detection accuracy of the mixed foreign matter is insufficient. The subject of this invention is providing the foreign material detection method, foreign material detection program, and foreign material detection apparatus which can detect reliably the foreign material mixed in the to-be-inspected to-be-tested object.

In order to solve the above-mentioned problem, the foreign matter detection method according to claim 1 is a foreign matter detection method for detecting the presence or absence of a mixed foreign matter in the inspection object by processing an X-ray image transmitted through the canned inspection object. In the method, a foreign object candidate image generation step S3 for generating a foreign object candidate image including a peripheral part foreign object image showing a peripheral part and a foreign substance adjacent to the peripheral part and an isolated foreign object image showing a foreign object isolated from the peripheral part from the X-ray image; A first foreign matter extraction step S4 for extracting the isolated foreign matter from the foreign matter candidate image; and by detecting a protruding convex portion on the inner periphery of the peripheral foreign matter image from the foreign matter candidate image, adjacent to the peripheral portion A second foreign matter extraction step S5 for extracting the foreign matter, and a foreign matter synthesis output step S6 for synthesizing and outputting the extracted foreign matter and the foreign matter adjacent to the peripheral edge. Is performed whether detection of the mixed foreign substance.

  Furthermore, the foreign matter detection method according to claim 2 is the foreign matter detection method according to claim 1, wherein the foreign matter candidate image generation step is a first step on the higher density side than the main distribution of the density histogram of the X-ray image. A first binarized image showing a peripheral portion of the object to be inspected on the X-ray image and a foreign object candidate is generated based on the threshold value of the X-ray image. The image of the object to be inspected on the X-ray image is expanded by expanding the image binarized by the second threshold value on the lower density side than the distribution until it overlaps the outer periphery of the peripheral portion of the first binarized image. Including a second binarized image generation step S41 for generating a second binarized image indicating the blob image overlapping the second binarized image from the first binarized image. A marginal foreign object image, overlying the second binarized image A first labeling process S42 for distinguishing and extracting a non-wrapped blob image as an isolated foreign object image is performed, and the second foreign object extraction stage includes the peripheral edge foreign object image extracted by the first labeling process and the second foreign object image. The third binarized image is synthesized from the binarized image of step S51, and the inner periphery of the peripheral edge foreign object image is obtained by self-smoothing difference using a median filter with respect to the third binarized image. Step S52 of performing a median filter difference, which is a process for detecting protruding protrusions in the image, and canceling the detection due to the shape noise on the inner periphery of the peripheral edge foreign object image with respect to the detection result of the median filter difference A step S53 of performing a second labeling process.

The foreign matter detection program according to claim 3 is a foreign matter detection program for processing the X-ray image transmitted through the canned inspection object to detect the presence or absence of the mixed foreign matter in the inspection object. A first binarized image indicating a peripheral portion of the object to be inspected on the X-ray image and a foreign object candidate is generated by a first threshold value on a higher density side than the main distribution of the density histogram of the image, and the X-ray is generated. The image binarized by the second threshold value on the lower density side than the main distribution of the density histogram of the image is expanded until it overlaps the outer periphery of the peripheral portion of the first binarized image. A second binarized image showing the background of the object to be inspected is generated, and a blob image overlapping the second binarized image is adjacent to the peripheral portion and the peripheral portion from the first binarized image As a peripheral foreign matter image showing the foreign matter And, let performed first labeling processing for extracting as an isolated foreign substance image indicating the isolated foreign matter other than the second binarized image and the overlapping blobs image from the image periphery,
A third binarized image is synthesized from the peripheral edge foreign object image extracted by the first labeling processing and the second binarized image, and a median filter is used for the third binarized image. By performing the self-smoothing difference, the median filter difference, which is a process for detecting protruding protrusions on the inner periphery of the peripheral edge foreign object image, is performed.
The detection result of the median filter difference is subjected to a second labeling process for canceling a detection due to the shape noise of the inner periphery of the peripheral edge foreign object image, and extracted by the first labeling process A computer is caused to perform a process of combining and outputting the isolated foreign object image and the result of the second labeling process.

Further, the foreign object detection apparatus according to claim 4 includes an X-ray irradiation means 2 for irradiating the canned inspection object 3 with X-rays, and an X-ray detector 4 for detecting X-rays transmitted through the inspection object. A foreign matter detection apparatus having image processing means 6 for detecting the presence or absence of foreign matter mixed in an inspection object by subjecting an X-ray image generated from X-ray transmission data output from the X-ray detector to image processing The image processing means generates a foreign object candidate image including a peripheral part foreign object image indicating a peripheral part and a foreign substance adjacent to the peripheral part and an isolated foreign object image indicating a foreign object isolated from the peripheral part from the X-ray image. By detecting an image generating means 61, a first foreign matter extracting means 62 for extracting the isolated foreign matter from the foreign matter candidate image, and a protruding convex portion on the inner periphery of the peripheral foreign matter image from the foreign matter candidate image. Adjacent to the peripheral edge Second foreign matter extracting means 63 for extracting an object, foreign matter synthesis output means 64 for synthesizing and outputting the foreign matter extracted by the first foreign matter extracting means and the foreign matter extracted by the second foreign matter extracting means, It is composed of

Further, the foreign matter detection apparatus according to claim 5 is the foreign matter detection apparatus according to claim 4, wherein the foreign matter candidate image generation means is a first higher-density side than the main distribution of the density histogram of the X-ray image. A first binarized image generating unit 611 that generates, as the foreign object candidate image, a first binarized image indicating a peripheral part of the inspection object on the X-ray image and a foreign object candidate according to the threshold of Further, the image binarized by the second threshold value on the lower density side than the main distribution of the density histogram of the X-ray image is expanded until it overlaps with the outer periphery of the peripheral portion of the first binarized image. And a second binarized image generating unit 612 that generates a second binarized image indicating the background of the inspection object on the X-ray image. By labeling the candidate image, the second 2 A first labeling processing unit 621 for distinguishing between a converted image, a blob image that overlaps and a blob image that does not overlap, and extracting a blob image that does not overlap as the isolated foreign matter, and the second foreign matter extraction unit Is a binarization that combines the second binarized image and the blob image that overlaps with the second binarized image distinguished by the first labeling processing unit and the second binarized image as a third binarized image. The image synthesis means 631, the median filter difference means 632 for performing a median filter difference by self-smoothing difference using a median filter for the third binarized image, and the detection result of the median filter difference, A labeling process is performed to cancel the detection due to the shape noise of the inner periphery of the peripheral edge foreign object image, and is set in advance. Blob excluding come of following micro blob and a second labeling processing unit 633 for extracting a foreign matter adjacent to the periphery.

According to the present invention, the detection accuracy of foreign matters mixed in a canned inspection object is remarkably improved as compared with the conventional method in which the peripheral portion of the inspection object is masked with a set number of pixels.

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

(First embodiment)
FIG. 1 is a flowchart for explaining the foreign object detection method of the present invention. In step S2, a density histogram is generated for the X-ray image input in step S1, and a main distribution is obtained. Thereafter, the first binarized image showing the peripheral portion of the object to be inspected and foreign substance candidates on the X-ray image by the first threshold value set on the higher density side than the main distribution of the density histogram obtained in step S2. Is generated in step S3, and an image binarized by a second threshold value set to a lower density side than the main distribution of the density histogram obtained in step S2 is defined as an outer periphery at a peripheral portion of the first binarized image. In step S41, a second binarized image that is expanded until it overlaps and shows the background of the inspection object on the X-ray image is generated, and the second binarized image is generated from the first binarized image. A blob image that overlaps with the image is extracted as a marginal foreign object image that indicates a peripheral part and a foreign substance adjacent to the peripheral part, and a foreign object that is isolated from the peripheral part of the image other than the blob image that overlaps the second binarized image Isolated A first labeling process that is extracted as an object image is performed in step S42, and a third binarized image is extracted from the peripheral edge foreign object image extracted in the first labeling process and the second binarized image in step S51. The median filter difference, which is a process for detecting the protruding portion on the inner periphery of the peripheral edge foreign object image by the self-smoothing difference using the median filter with respect to the third binarized image. In step S52, a second labeling process is performed in step S53 for canceling the detection due to the shape noise on the inner periphery of the peripheral edge foreign object image with respect to the detection result of the median filter difference in step S53. The isolated foreign object image extracted by the first labeling process and the result of the second labeling process are combined and output in step S6.

  For example, when the X-ray image example of FIG. 4 is input in step S1, the first binarized image is FIG. 5, the second binarized image is FIG. 6, and the result of the first labeling process is FIG. FIG. 8 shows the third binarized image, FIG. 9 shows the result of the second labeling process, and FIG. 10 shows the combined output of the extracted foreign matter. This embodiment will be described below.

  In the following explanation, the maximum value and the median filter using a kernel of N × N pixels are expressed as Max (N) and Median (N), respectively, and the median filter difference (the median filter image and the image before the median filter execution) Difference image) is expressed as MedianS (N), binarization by threshold C is expressed as BIN (C), density inversion of the image is expressed as INV (image), and OR of the two images is ORed (image 1) , Image 2), and a series of image processing is indicated by “image → processing 1 → processing 2 →...”.

In step S2 of FIG. 1, the density histogram of the X-ray image is generated in the range of density [0,255], and the high density portion [B,
255], the range in which the density histogram frequency is greater than 1 / (A + 2) of the peak is obtained as the main distribution. Here, the setting steps of the parameters A and B to be set in advance are not shown in FIG. 1, but set values estimated in advance according to the type of the inspection object are used. For example, A = 10 and B = 20 are used for the products shown in FIG.

  In step S3 of FIG. 1, the position on the high density side of the main distribution obtained in step S2 is set as a first threshold value, and a first binarized image is generated by X-ray image → BIN (first threshold value). .

  In step S41 of FIG. 1, the density B + d is set as the second threshold value, and a second binarized image is generated by INV (X-ray image → BIN (second threshold value)) → Max (D). Here, the kernel size D of Max (D) is set to a size sufficient for the second binarized image and the first binarized image to overlap by about one pixel on the outer periphery in the peripheral portion, and the kernel shape is circular. And Here, the parameter d and kernel size D setting steps to be set in advance are not shown in FIG. 1, but set values estimated in advance according to the type of the inspection object are used. For example, d = 0 and D = 7 are used for the product of FIG.

  In step S42 of FIG. 1, first labeling is performed by distinguishing and extracting, from the first binarized image, a peripheral foreign matter image indicating a peripheral portion and a foreign matter adjacent to the peripheral portion and an isolated foreign matter image indicating a foreign matter isolated from the peripheral portion. Process. In this labeling process, after removing a minute blob due to image noise, a blob that overlaps the second binarized image is used as a peripheral part foreign object image indicating a peripheral part and a foreign substance adjacent to the peripheral part, and a blob that is not is an isolated foreign object. By using the image, the peripheral edge foreign object image (gray blob in FIG. 7) and the isolated foreign object image (black blob in FIG. 7) are extracted. Here, although the setting step of the threshold value F1 with respect to the area of the blob for erasing the minute blob is not shown in FIG. 1, a setting value estimated in advance according to the type of the inspection object is used. For example, F1 = 5 is used for the variety shown in FIG.

  In step S51 in FIG. 1, a third binarized image is synthesized by OR (second binarized image, peripheral foreign object image).

  In step S52 of FIG. 1, the median filter difference is performed by the third binarized image → MedianS (E). Here, the setting step of the kernel size E is not shown in FIG. 1, but a setting value estimated in advance according to the type of the inspection object is used. For example, E = 17 is used for the product of FIG.

  In step S53 in FIG. 1, a second labeling process, which is a process for canceling the detection due to the shape noise on the inner periphery of the peripheral foreign matter image, is performed on the detection result of the median filter difference. Here, the setting step of the threshold value F2 with respect to the area of the blob for erasing the minute blob caused by the shape noise is not shown in FIG. 1, but the setting value estimated in advance according to the type of the object to be inspected is used. Use. For example, F2 = 3 is used for the product of FIG.

  In step S6 of FIG. 1, the result of the first labeling process (FIG. 7) and the result of the second labeling process (black blob in FIG. 9) are combined and output (FIG. 10).

  As described above, the foreign object detection method according to the present invention uses a mask pattern obtained by extracting the peripheral edge itself as a blob as well as a contaminated foreign object as compared with the conventional method in which the peripheral edge of the inspection object is masked with a set number of pixels. Since the resulting distortion of the mask pattern shape is also detected, the detection accuracy of foreign matters mixed in the canned inspection object is remarkably improved.

(Second Embodiment)

FIG. 3 is a block diagram for explaining a foreign matter detection apparatus using X-rays, which is an embodiment of the foreign matter detection apparatus of the present invention. That is, a conveying means 1 for conveying a canned object 3 (inspected object), an X-ray source 2 (X-ray irradiating means) for irradiating the canned object 3 with X-rays, and a canned object to be measured An X-ray detector 4 for detecting an X-ray transmission image 3, an image input means 5 for logarithmically converting an X-ray transmission image (X-ray transmission data) output from the X-ray detector and capturing it as an X-ray image; It is a foreign object detection device provided with image processing means 6 for processing the X-ray image and determining the presence or absence of mixed foreign substances. Image display means 7 may be added as necessary.

The transport means 1 is realized by, for example, a belt conveyor that transmits X-rays well, and transports the object to be measured 3 between the X-ray source 2 and the X-ray detector 4 that are arranged to face each other. The X-rays irradiated from the X-ray source 2 are absorbed by the object to be measured 3 and slightly absorbed by the belt conveyor, pass through them, and then reach the X-ray detector 4.

The X-ray detector 4 is realized by an X-ray line sensor, for example, and converts the X-ray transmission image of the inspection object 3 into a digital image. This digital image is sampled in the transport direction at a sampling pitch substantially equal to the sampling pitch on one line by the X-ray line sensor, and is taken into the memory as an X-ray transmission image.

Assuming that the X-ray absorption rate of the substance is α and the thickness of the substance is L, the intensity S ′ after the X-ray of the intensity S is transmitted through the substance is theoretically S ′ = S · exp (−α · L ). If transformation is performed by taking the logarithm of both sides, it can be written as α · L = log (S) −log (S ′). The X-ray transmission image corresponds to a two-dimensional distribution of S ′, and can be converted into an X-ray absorption image showing a two-dimensional distribution of the amount of absorption α · L due to a substance by modifying the logarithm as described above. Both the X-ray transmission image and the X-ray absorption image include physical property information of the object to be inspected, that is, the amount of absorption α · L due to the substance. This is advantageous for emphasizing and detecting foreign matter having a high X-ray absorption rate. In this case, for example, in an X-ray absorption image of an object to be inspected, it is possible to evaluate the likelihood of a foreign object by treating a portion having a high density locally or a sharp edge portion as a foreign object candidate point.

The image input means 5 logarithmically converts the X-ray transmission image into an X-ray absorption image to form an X-ray image and outputs the X-ray image to the image processing means 6. In this logarithmic conversion, correction depending on the physical properties of the inspection object and the X-ray wavelength may be added.

The image processing means 6 is composed of a CPU or the like having a parameter setting function, receives the input of the X-ray image output from the image input means 5, and based on the preset parameters, the book described with reference to FIG. The presence / absence detection of the mixed foreign matter is performed by executing the image processing including the foreign matter detection method of the invention. As a function of the image processing, the foreign matter candidate image generation unit 61, the first foreign matter extraction unit 62, and the second The foreign matter extraction means 63 and the foreign matter composition output means 64 are realized.

  The foreign object candidate image generation unit 61 includes a first binarized image generation unit 611 and a second binarized image generation unit 612. The X-ray image output from the image input unit 5 has a peripheral portion and a peripheral portion. A foreign object candidate image including a peripheral foreign object image indicating an adjacent foreign object and an isolated foreign image indicating a foreign object isolated from the peripheral part is generated. The first binarized image generating means 61 1 1 indicates the peripheral part of the object to be inspected on the X-ray image and the foreign object candidates by the first threshold value on the higher density side than the main distribution of the density histogram of the X-ray image. 1 binarized image is generated as a foreign object candidate image, and the second binarized image generating means 612 binarizes with a second threshold value on the lower density side than the main distribution of the density histogram of the X-ray image. The image is expanded until it overlaps with the outer periphery of the peripheral portion of the first binarized image to generate a second binarized image indicating the background of the inspection object on the X-ray image.

The first foreign matter extraction unit 62 performs a labeling process on the foreign matter candidate image generated by the foreign matter candidate image generation unit 61, thereby distinguishing the second binarized image from the overlapping blob image and the non-overlapping blob image. Then, the blob image that does not overlap is extracted as an isolated foreign object.
In order to extract an isolated foreign object, it is distinguished by overlapping with a binarized image indicating the background of the expanded inspection object, but the outer periphery of the inspection object is detected by a filter that detects an edge such as differentiation or Laplacian. And may be distinguished by overlapping with the outer periphery.

The second foreign matter extracting means 63 is composed of a binarized image synthesizing means 631, a median filter difference means 632, and a second labeling processing means 633, and detects protruding convex portions on the inner periphery of the peripheral part foreign matter image. To extract foreign matter adjacent to the peripheral edge.
The binarized image synthesizing unit 631 synthesizes the second binarized image distinguished by the first labeling processing unit and the blob image that overlaps the second binarized image and the second binarized image as a third binarized image. Then, the median filter difference means 632 performs a median filter difference on the third binarized image by self-smoothing difference using a median filter, and the second labeling processing means 633 detects the median filter difference. For the result, a labeling process for canceling the detected amount caused by the shape noise on the inner periphery of the peripheral part foreign object image is performed, and a blob excluding a minute blob having a predetermined size or less is adjacent to the peripheral part. Extract as a foreign object.

  The foreign matter synthesis output unit 64 synthesizes and outputs the isolated foreign matter image extracted in the first labeling process 621 and the result of the second labeling process 633.

The image display means 7 performs OR operation on the extracted foreign substance composite output from the foreign substance composite output means 64 of the image processing means 6 and the X-ray image and displays them on a CRT or the like.

  As described above, the present invention extracts a blob indicating a peripheral portion and a foreign matter adjacent to the peripheral portion, a blob indicating a foreign matter image isolated from the peripheral portion and a blob indicating a foreign matter isolated from the peripheral portion from the X-ray image of the inspection object, In order to extract the foreign matter adjacent to the peripheral portion by detecting the protruding portion protruding from the inner periphery from the peripheral portion and the blob indicating the foreign matter adjacent to the peripheral portion, the X-ray transmission amount and the foreign matter at the edge of the inspection object This is useful for an X-ray foreign matter inspection apparatus that inspects for the presence or absence of foreign matter adjacent to the edge of an object to be inspected at an equivalent X-ray transmission amount.

It is a flowchart for demonstrating the foreign material detection method of this invention. It is a flowchart for demonstrating the conventional foreign material detection method. It is a figure for demonstrating embodiment of the foreign material detection apparatus of this invention. It is an example of an X-ray image. It is a 1st binarized image example. It is a 2nd binarized image example. It is a 1st labeling process example. It is a 3rd binarized image example. It is a 2nd labeling process example. It is a composite output example of the extracted foreign material.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Conveyance means, 2 ... X-ray source (X-ray generation means), 3 ... Measuring object (inspection object),
DESCRIPTION OF SYMBOLS 4 ... X-ray detector, 5 ... Image input means, 6 ... Image processing means, 7 ... Image display means, 61 ... Foreign substance candidate image generation means, 62 ... First foreign substance extraction means,
63 ... second foreign matter extraction means, 64 ... foreign matter synthesis output means,
611 ... first binarized image generating means, 612 ... second binarized image generating means, 621 ... first labeling processing means, 631 ... third binarized image synthesizing means,
632 ... Median filter difference means, 633 ... Second labeling processing means

Claims (5)

In the foreign object detection method for processing the X-ray image transmitted through the canned inspection object and detecting the presence or absence of the contaminated foreign material in the inspection object,
A foreign object candidate image generating step (S3) for generating a foreign object candidate image including a peripheral part foreign object image indicating a peripheral part and a foreign substance adjacent to the peripheral part and an isolated foreign object image indicating a foreign object isolated from the peripheral part from the X-ray image;
A first foreign matter extraction step (S4) for extracting the isolated foreign matter from the foreign matter candidate image;
A second foreign matter extraction step (S5) for extracting a foreign matter adjacent to the peripheral portion by detecting a protruding convex portion on the inner periphery of the peripheral foreign matter image from the foreign matter candidate image;
A foreign matter detection method comprising: a foreign matter synthesis output step (S6) for synthesizing and outputting the extracted foreign matter and the foreign matter adjacent to the periphery. The foreign substance candidate image generation step includes a first threshold indicating a peripheral part of the object to be inspected on the X-ray image and a foreign substance candidate by a first threshold value on a higher density side than the main distribution of the density histogram of the X-ray image. Generate a binary image,
In the first foreign substance extraction step, an image binarized by a second threshold value on a lower density side than the main distribution of the density histogram of the X-ray image is defined as an outer periphery at a peripheral portion of the first binarized image. A second binarized image generating step (S41) for generating a second binarized image that is expanded until it overlaps and indicates the background of the inspection object on the X-ray image; A blob image that overlaps the second binarized image is used as the peripheral foreign object image from the binarized image, and other than the blob image that overlaps the second binarized image is distinguished and extracted as an isolated foreign image. Perform the first labeling process (S42),
In the second foreign matter extraction step, a step of combining a third binarized image from the peripheral foreign matter image extracted in the first labeling process and the second binarized image (S51), A step of performing a median filter difference, which is a process for detecting protruding protrusions on the inner periphery of the peripheral foreign matter image, on the third binarized image by self-smoothing difference using a median filter (S52). And a second labeling process (S53) for canceling the detected amount caused by the shape noise of the inner periphery of the peripheral edge foreign object image with respect to the detection result of the median filter difference (S53). The foreign object detection method according to claim 1, wherein: In a foreign object detection program for causing a computer to process an X-ray image transmitted through a canned inspection object and detect the presence or absence of a mixed foreign object in the inspection object,
Generating a first binarized image indicating a peripheral edge of the object to be inspected on the X-ray image and a foreign object candidate by a first threshold value higher than the main distribution of the density histogram of the X-ray image;
The image binarized by the second threshold value on the lower density side than the main distribution of the density histogram of the X-ray image is expanded until it overlaps with the outer periphery of the peripheral portion of the first binarized image. Generating a second binarized image showing the background of the inspection object on the line image;
A blob image that overlaps the second binarized image is extracted from the first binarized image as a marginal foreign object image indicating a peripheral part and a foreign substance adjacent to the peripheral part, and the second binarization is performed. A first labeling process is performed to extract an image other than the blob image that overlaps with the image as an isolated foreign object image indicating a foreign object isolated from the peripheral edge of the image;
A third binarized image is synthesized from the peripheral foreign matter image extracted by the first labeling process and the second binarized image;
With respect to the third binarized image, a median filter difference that is a process for detecting protruding protrusions on the inner periphery of the peripheral edge foreign object image is performed by a self-smoothing difference using a median filter,
For the detection result of the median filter difference, perform a second labeling process for canceling the detection due to the shape noise of the inner periphery of the peripheral edge foreign object image,
A foreign object detection program for combining and outputting the isolated foreign object image extracted by the first labeling process and the result of the second labeling process. From the X-ray detector, an X-ray irradiation means (2) for irradiating the canned inspection object (3) with X-rays, an X-ray detector (4) for detecting X-rays transmitted through the inspection object, In a foreign matter detection apparatus having image processing means (6) for detecting the presence or absence of foreign matter mixed in an inspection object by subjecting an X-ray image generated from output X-ray transmission data to image processing,
The image processing unit generates a foreign object candidate image that generates a foreign object candidate image including a peripheral foreign object image indicating a peripheral part and a foreign substance adjacent to the peripheral part and an isolated foreign object image indicating a foreign object isolated from the peripheral part from the X-ray image. Means (61);
First foreign matter extraction means (62) for extracting the isolated foreign matter from the foreign matter candidate image;
Second foreign matter extracting means (63) for extracting a foreign matter adjacent to the peripheral portion by detecting a protruding convex portion on the inner periphery of the peripheral foreign matter image from the foreign matter candidate image;
A foreign object comprising a foreign object extracted by the first foreign object extraction means and a foreign object composition output means (64) for combining and outputting the foreign object extracted by the second foreign substance extraction means. Detection device. The foreign substance candidate image generating means is a first unit that indicates a candidate for a peripheral part of the inspection object and foreign substances on the X-ray image by a first threshold value on a higher density side than the main distribution of the density histogram of the X-ray image. First binarized image generating means (611) for generating a binarized image as the foreign object candidate image, and further by a second threshold value on the lower density side than the main distribution of the density histogram of the X-ray image. The binarized image is expanded until it overlaps the outer periphery of the peripheral portion of the first binarized image to generate a second binarized image indicating the background of the inspection object on the X-ray image. Second binarized image generation means (612),
The first foreign matter extraction means performs a labeling process on the foreign matter candidate image to distinguish between the second binarized image and a blob image that does not overlap, and a blob image that does not overlap. First labeling processing means (621) for extracting as an isolated foreign matter,
The second foreign matter extracting means performs third binarization on the blob image and the second binarized image that overlap with the second binarized image distinguished by the first labeling processing means. A binarized image synthesizing unit (631) for synthesizing as an image, and a median filter difference unit (632) for performing a median filter difference on the third binarized image by self-smoothing difference using a median filter. Then, a labeling process is performed on the detection result of the median filter difference to cancel the detection due to the shape noise on the inner periphery of the peripheral edge foreign object image, and a minute blob having a predetermined size or less is removed. 5. The foreign matter detection apparatus according to claim 4, further comprising second labeling processing means (633) for extracting the blob as foreign matter adjacent to the peripheral edge.

Images