JP5297142B2 - Foreign object detection method and apparatus - Google Patents

Description

The present invention relates to a foreign object detection method and a foreign object detection apparatus that perform differential processing on a dual energy X-ray image of an object to be measured to detect the presence or absence of a mixed foreign object.

  Dual energy of low energy and high energy is an effective standard method for performing X-ray inspection of foreign substances such as foreign substances in food and residual bones in meat with high sensitivity using high-contrast different component images. An energy subtraction (see Patent Document 1) is known in which an X-ray image is logarithmically converted into a low energy and high energy equivalent thickness image pair and then subjected to weighted difference processing to separate different component images.

The intensities of low energy X-rays and high energy X-rays that have passed through an object to be measured such as food and meat are determined by an energy-specific photointegration X-ray optical system using a plurality of indirect conversion X-ray detectors (see Patent Document 2). Alternatively, a low-energy and high-energy X-ray image pair, that is, dual-energy X-rays, can be obtained by an energy-specific photocounting X-ray optical system using a single direct-conversion X-ray detector that discriminates the photon energy every arrival of X-ray photons. Can be converted to an image.

In the case of the energy-specific photointegration X-ray optical system, an image registration process is required to eliminate the relative positional shift of the dual energy X-ray image generated depending on the X-ray optical system. In the case of the energy-specific photocounting X-ray optical system, the relative displacement does not occur.

  A dual energy X-ray image having no relative positional deviation is set if the energy band is appropriately set so that the mixed foreign matter can be detected as a different component with high contrast and the weight parameter of the weighted difference processing is optimal. By applying the energy subtraction, the high-contrast different-component image can be immediately separated, and the presence / absence determination of the contaminated foreign matter can be realized by simple threshold processing for the different-component image.

JP 04-11473 A JP 2002-365368 A

  The energy subtraction is a method that exerts an effect when it is desired to perform X-ray inspection of foreign matters such as extraneous foreign matter in food and residual bone in meat with high sensitivity using high-contrast different component images. It has been a problem to overcome the disadvantage that the work of optimally setting the weighting parameters of the weighted difference processing depending on the imaging condition of the energy X-ray image and the object to be measured such as food and meat is delicate and troublesome.

An object of the present invention is to provide a foreign object detection method and a foreign object detection device that can overcome the drawbacks by automatically setting the weight parameter.

The inventor of the present invention pays attention to the fact that the X-ray absorption spectrum of a foreign substance such as an extraneous foreign substance in food or a residual bone in meat is different from the X-ray absorption spectrum of an object to be measured such as food or meat. As a result of component analysis, we have found through experiments that the contaminated foreign matter can be separated as a different component image of the object to be measured. By applying this fact, the weight parameter for energy subtraction, which is troublesome to set, is automatically set optimally. The present invention has been reached.

  A foreign matter detection method according to claim 1 of the present invention is a foreign matter detection method for detecting the presence or absence of a mixed foreign matter by performing differential processing on a dual energy X-ray image of an object to be measured. Generating an equivalent thickness image pair of an equivalent thickness image and a high energy equivalent thickness image, and applying independent component analysis so as to separate the mixed foreign substance component and the measured object component into the equivalent thickness image pair Obtaining a separation matrix, obtaining a weight parameter for the difference processing based on two separation vectors that are elements of the separation matrix, and performing the difference processing using the weight parameter from the equivalent thickness image pair. The method includes a step of separating a component image of the mixed foreign matter and a step of detecting a foreign matter by performing threshold processing on the component image of the mixed foreign matter.

  With this configuration, it is possible to automatically set the optimum weight parameter used for the difference processing in accordance with the dual energy X-ray image of the object to be measured, and highly sensitive foreign matter detection can be performed.

The foreign matter detection method according to claim 2 of the present invention is the foreign matter detection method according to claim 1 of the present invention, wherein the equivalent matrix image is obtained before the step of obtaining a separation matrix by applying independent component analysis to the equivalent thickness image pair. A step of reducing the image noise of the thick image pair is added.
In other words, a foreign matter detection method for detecting the presence or absence of mixed foreign matter by performing differential processing on a dual energy X-ray image of an object to be measured, the equivalent thickness of a low energy equivalent thickness image and a high energy equivalent thickness image as the dual energy X-ray image Generating an image pair, reducing image noise of the equivalent thickness image pair, and separating the mixed foreign substance component and the measured object component from the equivalent thickness image pair with reduced image noise Applying independent component analysis to obtain a separation matrix, obtaining a weight parameter for the difference processing based on two separation vectors that are elements of the separation matrix, and the difference using the weight parameter. Separating the component image of the mixed foreign matter from the pair of equivalent thickness images by processing, and detecting the foreign matter by performing threshold processing on the component image of the mixed foreign matter It has a step that, a configuration including a.

  With this configuration, it is possible to automatically set the optimum weight parameter used for the difference processing in accordance with the dual energy X-ray image of the object to be measured with reduced noise, and highly sensitive foreign matter detection can be performed.

The foreign object detection apparatus according to claim 3 of the present invention is an X-ray irradiating means 2 for irradiating the object to be measured 3 with X-rays, and the X-ray intensity transmitted through the object to be measured is an X-ray image of low energy and high energy. X-ray detection means 4 for converting into a pair, image input means 5 for storing the X-ray image pair as a dual energy X-ray image, and differential processing of the dual energy X-ray image is mixed into the object to be measured. In the foreign matter detection apparatus comprising the image processing means 6 for detecting the presence or absence of foreign matter, the foreign matter detection method of claim 1 is used for the image processing means 6.
That is, the image processing means 6 using the foreign object detection method according to claim 1 generates an equivalent thickness image generating means 61 for generating an equivalent thickness image pair of a low energy equivalent thickness image and a high energy equivalent thickness image from the dual energy X-ray image. A separation matrix calculating means 63 for obtaining a separation matrix by applying independent component analysis so as to separate the mixed foreign substance component and the measured object component from the equivalent thickness image pair, and elements of the separation matrix A parameter calculation unit 64 for obtaining a weight parameter for the difference processing based on two separation vectors, and a different component for separating the component image of the mixed foreign matter from the equivalent thickness image pair by the difference processing using the weight parameter. An image separation unit 67 and a foreign matter detection unit 68 that detects a foreign matter by performing threshold processing on the component image of the mixed foreign matter are provided.

  With this configuration, it is possible to automatically set the optimum weight parameter used for the difference processing according to the object to be measured, and a highly sensitive foreign matter detection apparatus can be realized.

  A foreign object detection apparatus according to a fourth aspect of the present invention is the foreign object detection apparatus according to the third aspect of the present invention, comprising image noise reduction means 62 for reducing image noise of the equivalent thickness image pair, and obtaining the separation matrix. The means is to obtain a separation matrix for the equivalent thickness image pair with reduced noise.

  With this configuration, it is possible to automatically set the optimum weight parameter used for the difference processing in accordance with the dual energy X-ray image of the object to be measured with reduced noise, and a highly sensitive foreign matter detection apparatus can be realized.

  The foreign matter detection apparatus according to claim 5 of the present invention is the foreign matter detection apparatus according to claim 3 or 4 of the present invention, wherein the weight parameter obtained by the parameter calculation means 64 is stored in the image processing means 6. A parameter storage unit 69 and a parameter reference unit 60 for referring to the stored weight parameter are added.

  With this configuration, the optimum weight parameter used for the difference processing obtained according to the specific object to be measured is stored, and the weight parameter is calculated from the separation matrix calculation that has been performed each time for the same kind of object to be measured. A high-sensitivity foreign matter detection device that can inspect a measurement object of the same type as the specific measurement object for which the optimum weight parameter has been obtained by referring to the weight parameter instead of the calculation process leading to realizable.

  In the foreign matter detection method according to claim 1 of the present invention, it is possible to automatically set the optimum weighting parameter used for the difference processing in accordance with the dual energy X-ray image of the object to be measured. It can be performed.

  In the foreign matter detection method according to claim 2 of the present invention, it is possible to automatically set the optimum weight parameter used for the difference processing according to the dual energy X-ray image of the object to be measured with reduced noise, which is extremely high. Sensitive foreign object detection can be performed with high sensitivity.

  The foreign object detection apparatus according to claim 3 or 4 of the present invention can automatically set the optimum weight parameter used for the difference processing in accordance with the object to be measured, and therefore performs highly sensitive and stable foreign object detection. be able to.

  The foreign object detection apparatus according to claim 5 of the present invention stores the optimum weight parameter used for the difference processing obtained in accordance with a specific object to be measured, and refers to the weight parameter to thereby obtain the same kind of weight parameter. Since the object to be measured can be inspected at high speed, high-sensitivity and high-speed foreign object detection can be performed.

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

1. 1st Embodiment (FIGS. 1-2)
As shown in FIG. 1, the foreign object detection apparatus according to the present embodiment includes a conveying unit 1 that conveys a measurement object 3, an X-ray irradiation unit (X-ray source) 2 that irradiates the measurement object 3 with X-rays, X-ray detection means 4 for converting the X-ray intensity transmitted through the object to be measured into a low energy and high energy X-ray image pair; an image input means 5 for storing the X-ray image pair as a dual energy X-ray image; Image processing means 6 for detecting the presence or absence of foreign matter mixed in the object to be measured by performing differential processing on the dual energy X-ray image, and image display means 7 as necessary.

The transport unit 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 detection unit 4 that are opposed to 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 reach the X-ray detection means 4.

  The X-ray detection means 4 is realized by, for example, a direct conversion type X-ray line sensor that discriminates the photon energy every arrival of an X-ray photon, and forms an energy-specific photocounting X-ray optical system, thereby By sampling in the conveying direction at a sampling pitch substantially equal to the sampling pitch on one line, the X-ray intensity transmitted through the object to be measured 3 and the belt conveyor is taken into the memory as image data for each energy band, and low energy and high Convert to energy X-ray image pairs.

Although not shown in the figure, by using two indirect conversion type X-ray line sensors for the low energy and the high energy instead of the direct conversion type X-ray line sensor as the X-ray detection means 4, the photo-integration X for each energy A linear optical system may be formed.

The image input unit 5 stores the X-ray image pair of the DUT 3 as a dual energy X-ray image without a relative positional shift, and outputs it to the image processing unit 6. When the photo-integrated X-ray optical system for each energy is formed, image registration processing is performed here in order to eliminate the relative displacement of the X-ray image pair captured on the memory.

The image processing means 6 is a CPU or the like on which an image processing program or the like is installed, and the foreign matter detection method according to the present invention shown in FIG. 2 is implemented, and the dual energy X output from the image input means 5 without relative positional deviation. In response to the input of the line image (step S101), an equivalent thickness image pair of a low energy equivalent thickness image and a high energy equivalent thickness image is generated by performing logarithmic conversion in the equivalent thickness image generating means 61 (step S102), and necessary. Accordingly, the image noise reduction means 62 applies edge preserving smoothing processing or the like to the equivalent thickness image pair to reduce the image noise (step S103), and the separation matrix calculation means 63 performs independent component analysis on the equivalent thickness image pair. Is applied to obtain a separation matrix (step S104), and the separation vector discriminating means 641 calculates the difference from low energy to high energy. The separation vector of the different component is discriminated as the separation vector of the different component (step S105), and the separation vector normalizing means 642 normalizes the separation vector of the different component into a form in which the high energy scaled from the low energy is subtracted (step S105). In step S106, the parameter setting means 643 automatically sets the normalized separation vector as a weighting parameter for the difference processing (step S107). That is, the parameter calculation means 64 comprising means 641 to 643 (steps S105 to S107) obtains and automatically sets the weighting parameter for the difference processing based on two separation vectors that are elements of the separation matrix. Further, the component image of the mixed foreign matter is separated from the equivalent thickness image pair by the difference processing using the weight parameter by the different component image separation unit 67 using the automatically set weight parameter (Step S108). The detection means 68 detects the presence or absence of the mixed foreign matter by performing threshold processing on the component image of the mixed foreign matter and detecting the foreign matter (step S109).

As the independent component analysis, FastICA (Fast
Common and known signal separation algorithms such as Independent Component Analysis (SOBI) and Second Order Blind Identification (SOBI) can be applied. An outline of the independent component analysis is shown in FIG. FIG. 5 (a) shows a low energy equivalent thickness image, FIG. 5 (b) shows a high energy equivalent thickness image, FIG. 5 (c) shows a component image of a mixed foreign object, and FIG. 5 (d) shows a measured object. The component image of the object is shown, and the independent component analysis is to obtain a separation matrix so as to separate the mixed foreign substance component and the measured object component from the equivalent thickness image pair. It should be noted that the equivalent thickness image pair serving as an input for the independent component analysis needs to include at least the main part of the DUT 3 and includes even a foreign component foreign material to be separated. It does not have to be. For example, if the object to be measured can be regarded as a substantially single component as in the case of foreign matter inspection of sausage, the two types of signals that can be separated as independent components by the independent component analysis are those of the object to be measured (sausage). This is because it can be regarded as a signal and a signal other than that (hard foreign matter such as residual bone, metal, glass, and stone).

Assuming that the two separation vectors that are elements of the separation matrix calculated in step S104 are (a, b) and (c, d), a vector having a density pair in the corresponding pixel of the equivalent thickness image pair as a component (L , H), the two independent components are given by X = aL + bH and Y = cL + dH. However, it remains undefined which of X and Y corresponds to the density of the different component image, and X and The density magnification of Y is also indefinite including the inverted state, and cannot be linked to energy subtraction by the optimum weight parameter.

Therefore, in step S105, when L is a low energy component, H is a high energy component, and a positive weight parameter is k, an energy subtraction of Z = L−kH corresponding to the difference processing is a separation vector (1, − Focusing on the fact that it is equal to the component that can be separated by k), one of (a, b) and (c, d) that is closer to (1, -k) is discriminated as a different component separation vector. Specifically, since the low energy equivalent thickness image has a higher contrast and clearer than the high energy equivalent thickness image, k is set to 1 if the logarithmic transformation is performed so that the density of the equivalent thickness image pair is approximately the same. Based on the knowledge that the values are close to each other, for example, the calculated values of b / a and d / c may be evaluated, and the separation vector that has calculated the value close to −1 may be determined as the separation vector of the different component.

In step S106, the determined separation vector is (a, b), (a, b) is multiplied by a magnification 1 / a, and transformed into the form (1, b / a). Normalize the vector.

In step S107, the normalized separation vector is set to (1, b / a), and k = −b / a is automatically set for the difference processing with Z = L−kH.

In step S108, the difference process of Z = L−kH is executed.

The image display means 7 displays inspection result information corresponding to the foreign object detection result of the foreign object detection means 68 on a flat panel display or the like.

2. Second Embodiment (FIGS. 3 to 4)
In the present embodiment, in the image processing means 6 of the first embodiment, a parameter storage means 69 for storing automatically set weight parameters, and a parameter reference means 60 for referring to the stored weight parameters, The other configurations are the same as those in the first embodiment. Therefore, the parameter storage unit 69 and the parameter reference unit 60 will be described. The same parts as those in the first embodiment are denoted by the same reference numerals as those in the first embodiment, and the description of the first embodiment is used. The description will not be repeated.

The parameter storage means 69 receives the difference processing weight parameter automatically set by the parameter calculation means 64 and outputs it to the different component image separation means 67. When the operation mode signal (not shown) indicates parameter storage, the parameter storage means 69 Is stored in a reference memory (not shown) (step S202).

When the operation mode signal does not indicate parameter reference, the parameter reference means 60 outputs the equivalent noise image output from the equivalent thickness image generation means 61, and further through the image noise reduction means 62 when the image noise reduction means 62 exists. The thick image pair is received and output to the separation matrix calculation means 63. Then, when the operation mode signal indicates parameter reference, the flow from step S104 to step S202 described above is switched, and the stored weight parameter of the difference processing is read from the reference memory and the different component image separation is performed. Output directly to the means 67 (step S201).

As described above, the present invention automatically sets the optimum weight parameter for energy subtraction in accordance with the dual energy X-ray image of the object to be measured so that highly sensitive and stable foreign object detection can be easily performed. The present invention is useful for all foreign object detection devices having detection means for outputting line data.

It is a figure which shows schematic structure of the foreign material detection apparatus concerning 1st Embodiment of this invention. It is a flowchart which shows the general | schematic flow of the foreign material detection method concerning 1st Embodiment of this invention. It is a figure which shows schematic structure of the foreign material detection apparatus concerning 2nd Embodiment of this invention. It is a flowchart which shows the general | schematic flow of the foreign material detection method concerning 2nd Embodiment of this invention. It is a figure which shows the outline of a general and well-known independent component analysis.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Conveyance means 2 ... X-ray source 3 ... DUT 4 ... X-ray detection means 5 ... Image input means 6 ... Image processing means 7 ... Image display means
60 ... Parameter reference means 61 ... Equivalent thickness image generation means 62 ... Image noise reduction means 63 ... Separation matrix calculation means 64 ... Parameter calculation means 641 ... Separation vector discrimination means 642 ... Separation vector normalization means 643 ... Parameter setting means 67 ... Different Component image separation means 68 ... Foreign matter detection means 69 ... Parameter storage means

Claims (5)

A foreign matter detection method for performing differential processing on a dual energy X-ray image of a measurement object to detect the presence or absence of mixed foreign matter, wherein the dual energy X-ray image is an equivalent thickness image of a low energy equivalent thickness image and a high energy equivalent thickness image. Generating a pair; applying an independent component analysis to the equivalent thickness image pair so as to separate the mixed foreign matter component and the measured object component into a separation matrix; and Obtaining a weight parameter for the difference processing based on two separation vectors that are elements, separating the component image of the contaminated foreign matter from the equivalent thickness image pair by the difference processing using the weight parameter, And a step of detecting a foreign substance by performing threshold processing on a component image of the mixed foreign substance. Prior to the step of applying an independent component analysis to obtain a separation matrix so as to separate the contaminated foreign material component and the measured object component from the equivalent thickness image pair, the image noise of the equivalent thickness image pair is determined. The foreign matter detection method according to claim 1, further comprising a reduction step. X-ray irradiation means (2) for irradiating the object to be measured (3) with X-rays, and X-ray detection means for converting the X-ray intensity transmitted through the object to be measured into a pair of low energy and high energy X-ray images ( 4), image input means (5) for storing the X-ray image pair as a dual energy X-ray image, and differential processing of the dual energy X-ray image to detect the presence or absence of foreign matter mixed in the object to be measured The image processing means (6), wherein the image processing means (6) generates an equivalent thickness image pair of a low energy equivalent thickness image and a high energy equivalent thickness image from the dual energy X-ray image. An equivalent thickness image generating means (61) to be generated, and a component for obtaining a separation matrix by applying independent component analysis to the equivalent thickness image pair so as to separate the mixed foreign substance component and the measured object component. The matrix calculation means (63), the parameter calculation means (64) for obtaining a weight parameter for the difference processing based on two separation vectors as elements of the separation matrix, and the equivalent processing by the difference processing using the weight parameter. A different component image separating unit (67) for separating the component image of the mixed foreign matter from the thick image pair, and a foreign matter detecting unit (68) for detecting the foreign matter by performing threshold processing on the component image of the mixed foreign matter. A foreign object detection device. The image noise reduction means (62) for reducing image noise of the equivalent thickness image pair, wherein the separation matrix obtaining means obtains a separation matrix for the noise-reduced equivalent thickness image pair. 4. The foreign object detection device according to 3. A parameter storage means (69) for storing the weight parameter obtained by the parameter calculation means and a parameter reference means (60) for referring to the stored weight parameter are added to the image processing means (6). The foreign object detection device according to claim 3 or 4.