JP5644580B2 - Anomaly judgment device and anomaly judgment method - Google Patents

Description

  The present invention relates to an abnormality determination device and an abnormality determination method.

  Conventionally, a method for measuring a reflection spectrum or an absorption spectrum by irradiating measurement light to a container from the outside of the container for the purpose of detecting an abnormality of contents such as a liquid filled in the container is known. (For example, refer to Patent Documents 1 and 2).

Utility Model Registration No. 308743 JP 2010-197157 A

  However, according to the methods described in Patent Documents 1 and 2, since the absorption / reflection of the measurement light by the container is much larger than the absorption / reflection of the measurement light by the contents, the obtained reflection spectrum or absorption spectrum is The impact is very large. Therefore, it is difficult to detect minute changes related to the reflection spectrum and absorption spectrum of the contents, and it is difficult to detect minute changes in the contents.

  The present invention has been made in view of the above, and provides an abnormality determination device capable of detecting an abnormality of an inspection object filled in a container with higher accuracy and an abnormality determination method using the abnormality determination device. For the purpose.

  In order to achieve the above object, an abnormality determination device according to the present invention determines whether there is an abnormality in an inspection object based on a spectral image obtained by imaging a container in which the inspection object is stored. An apparatus for irradiating the container with light of a plurality of wavelengths, and receiving reflected light from the container by the light emitted from the irradiation means, thereby imaging the container and obtaining a spectral image. An imaging unit that obtains, an extraction unit that extracts an object imaging pixel including information on the inspection object from all pixels included in the spectrum image obtained by imaging by the imaging unit, and an extraction unit that extracts The average of the spectral intensities for each wavelength in all the object imaging pixels is calculated, and the average reflection spectrum is calculated by the averaging means for calculating the average reflection spectrum. Based on the serial average reflectance spectrum, characterized in that it comprises a determination means for determining presence or absence of abnormality of the inspection object.

  Further, the abnormality determination method according to the present invention is an abnormality determination method for determining the presence or absence of an abnormality of the inspection object based on a spectrum image obtained by imaging a container in which the inspection object is accommodated. An irradiation step of irradiating the container with light of a plurality of wavelengths, and an imaging step of capturing a container to obtain a spectral image by receiving reflected light from the container by the light irradiated in the irradiation step; An extraction step for extracting object imaging pixels including information on the inspection object from all pixels included in the spectrum image obtained by imaging in the imaging step, and all the objects extracted in the extraction step An averaging step of calculating an average reflection spectrum for each wavelength in the object imaging pixel and calculating an average reflection spectrum, and the averaging step Tsu on the basis of the average reflectance spectrum calculated by flop, and having a a judgment step of judging whether the abnormality of the inspection object.

  According to the above-described abnormality determination device and abnormality determination method, first, an object imaging pixel including information related to the inspection object is extracted from all pixels included in the spectrum image obtained by imaging. In this way, by limiting the pixels used for the determination, it is possible to reduce erroneous determination caused by noise caused by the container. In addition, the average of the spectrum intensity with respect to each wavelength in the object imaging pixel is obtained, an average reflection spectrum is calculated, and the presence / absence of an abnormality is determined using this. Thus, by calculating and using the average reflection spectrum, it is possible to reduce misjudgments derived from noise or the like included in each pixel, and to detect abnormalities of the inspection object filled in the container with higher accuracy. It becomes possible to detect.

  Here, the said irradiation means can be set as the aspect which irradiates the light of the several wavelength contained in the range of wavelength 1200-2400nm. By adopting an aspect in which light in this wavelength range is irradiated, it is possible to perform abnormality determination with higher accuracy when the inspection object contains moisture.

  Further, the extraction means includes a reflectance at a first wavelength included in a wavelength range of 1200 to 2400 nm and a first wavelength included in a wavelength range of 1200 to 2400 nm in each pixel included in the spectrum image. It is possible to calculate the ratio of the reflectance at the second wavelength different from the above and determine whether the pixel is the object imaging pixel based on the result.

  In addition, the extraction unit may determine whether each pixel included in the spectrum image has the target based on whether the reflectance at the third wavelength included in the wavelength range of 1200 to 2400 nm is within a predetermined range. It can be set as the aspect which judges whether it is an object imaging pixel.

  In addition, when the reflectance at the fourth wavelength included in the wavelength range of 1200 to 2400 nm is 1 or more in each pixel included in the spectrum image, the extraction unit extracts the pixel from the object imaging pixel. It can be set as the aspect excluded. By having this aspect, it is possible to exclude pixels in which regular reflection occurs from the pixels used for the abnormality determination, and to perform abnormality determination with higher accuracy.

  ADVANTAGE OF THE INVENTION According to this invention, the abnormality determination apparatus which can detect the abnormality of the test object with which it filled in the container with a higher precision, and the abnormality determination method by this abnormality determination apparatus are provided.

It is a figure which shows the structure of the abnormality determination apparatus which concerns on this embodiment. It is a figure explaining a hyperspectral image.

  DESCRIPTION OF EMBODIMENTS Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the accompanying drawings. In the description of the drawings, the same elements are denoted by the same reference numerals, and redundant description is omitted.

  An abnormality determination apparatus 100 according to the present embodiment will be described with reference to FIG. The abnormality determination apparatus 100 according to the present embodiment is an apparatus that inspects for the presence or absence of abnormality such as alteration of an inspection object inside the container 1 that moves in a direction indicated by an arrow A by a conveyance unit (not shown). Examples of the inspection object of the abnormality determination device 100 according to the present embodiment include foods and medicines filled in the container 1. The abnormalities of these inspection objects include alterations such as rot of the inspection object, contamination with bacteria, and the like. The abnormality determination device 100 measures the spectrum of diffuse reflected light obtained by irradiating the container 1 with measurement light from the outside of the container 1 filled with the inspection object, and the inspection object based on the spectrum. Detect anomalies. The abnormality determination device 100 includes a light source 10, a mirror 15, a camera 20 (imaging unit), and a control unit 30.

  The light source 10 irradiates measurement light including light of a plurality of wavelengths toward a predetermined irradiation area. This irradiation region is a position through which the container 1 moved by the transport means passes, and is a linear region extending in a direction perpendicular to the moving direction A of the container 1. The wavelength range of the measurement light emitted by the light source 10 is appropriately selected according to the inspection object in the container 1 or the abnormality of the inspection object that is the detection object. When near infrared light is used as measurement light, specifically, light including a plurality of components having a wavelength in the range of 1200 nm to 2400 nm is preferably used, but visible light is used as measurement light instead of near infrared light. It is also possible to use it. For example, a halogen lamp is preferably used as the light source 10 that emits such light. In the present embodiment, the two light sources 10 that irradiate the same irradiation region are provided at different angles with respect to the irradiation region. The light source 10 may be provided with a cylindrical lens as a means for emitting light to the linear irradiation region.

  Near-infrared light L1 output from the light source 10 is diffusely reflected by the container 1 located in the irradiation region. A part of the light then enters the camera 20 through the mirror 15 as diffusely reflected light L2.

The camera 20 has a function as a hyperspectral sensor that acquires a hyperspectral image. Here, the hyperspectral image in this embodiment is demonstrated using FIG. FIG. 2 is a diagram for explaining the outline of the hyperspectral image. As shown in FIG. 2, the hyperspectral image H is an image configured by _N pixels P _{1 to} P _N. Specifically it shows the two pixel P _n and P _m as them example in FIG. Each of the pixels P _n and P _m includes spectral information S _n and S _m including a plurality of intensity data. And the intensity data is data indicating a spectral intensity at a particular wavelength (or wavelength band), 2, 15 intensity data has been held as the spectral information S _n and S _m, superposition of these It is shown in the state. As described above, the hyperspectral image H is characterized by having a plurality of intensity data for each pixel constituting the image, so that a three-dimensional image having both a two-dimensional element as an image and an element as spectral data. Configuration data. In the present embodiment, the hyperspectral image H refers to an image composed of pixels having intensity data in at least three wavelength bands per pixel. In FIG. 2, the inspection object 3 is also shown. That is, in FIG. 2, P _n is a pixel that images the inspection object 3, and P _m is a pixel that images the background (for example, conveying means).

  Returning to FIG. 1, the camera 20 according to the present embodiment is provided with a slit and a spectroscope (not shown), and receives a linear region extending in a direction perpendicular to the traveling direction of the container 1 (y-axis direction). As a region, light from the light receiving region is incident through the mirror 15. The camera 20 includes a light receiving surface in which a plurality of light receiving elements are two-dimensionally arranged, and each light receiving element includes a light receiving unit that receives light. As a result, the light receiving unit receives light of each wavelength of the diffusely reflected light L2 reflected at each position along the direction (y-axis direction) perpendicular to the traveling direction of the container 1. Each light receiving element outputs a signal corresponding to the intensity of received light as information on a two-dimensional planar point composed of a position and a wavelength. A signal output from the light receiving element of the light receiving unit inside the camera 20 is sent from the camera 20 to the control unit 30 as image data related to the hyperspectral image.

  The control unit 30 obtains the spectrum of the diffusely reflected light L2 from the input signal, and determines whether there is an abnormality based on the obtained spectrum. When detecting the abnormality of the inspection object in the container 1, the inspection is performed according to the following principle.

  When the contents to be inspected are abnormal, a change in spectral intensity occurs in a specific absorption band in the wavelength range of measurement light (near infrared light in the present embodiment) output from the light source 10. Therefore, with reference to the spectrum information included in each pixel in the hyperspectral image H by the diffusely reflected light L2 diffusely reflected by the container 1 and composed of a plurality of intensity data, the peak intensity of the specific reflected light originating from the abnormality is determined. By detecting the change, the presence / absence of an abnormality is determined. The result of determination by the control unit 30 is output to, for example, a monitor connected to the control unit 30, a printer, or the like. The operator is notified.

  The control unit 30 includes a CPU (Central Processing Unit), a RAM (Random Access Memory) and a ROM (Read Only Memory) that are main storage devices, a communication module that performs communication with other devices such as a detection unit, and a hard disk It is comprised as a computer provided with hardware, such as auxiliary storage devices. And the function as the control part 30 is exhibited when these components operate | move.

  The control unit 30 includes an extraction unit 31, an averaging unit 32, and a determination unit 33. Image data sent from the camera 20 is first sent to the extraction unit 31.

  The extraction unit 31 has a function of extracting an object imaging pixel including information related to the inspection object from all pixels included in the image data. As shown in FIG. 2, the image data includes pixels in which the inspection target is included in the captured image and pixels in which the inspection target is not included. Therefore, the extraction unit 31 extracts an object imaging pixel that is a pixel that is relatively less affected by the absorption and reflection of the measurement light by the container and obtains information related to the inspection object. As a specific method, for example, when the ratio between the reflectance at the first wavelength and the reflectance at the second wavelength different from the first wavelength is within a predetermined range, the pixel is a container. There is a method of determining that the pixel is a pixel in which the information on the object to be inspected can be obtained, that is, the pixel in which the contents in the container 1 are imaged. Specifically, for example, when the content is a substance containing moisture, the reflectance at a wavelength of 1384 nm where an absorption peak derived from water appears and the reflectance at a wavelength of 1300 nm where the influence of the absorption peak derived from water becomes small In this method, when the ratio is within a predetermined range, it is determined that the pixel is a pixel obtained by imaging the content containing moisture.

  The reflectance referred to in the present embodiment refers to the ratio of the intensity of reflected light when the reflected light intensity is 1 when the reflected light is measured using a standard diffuser, and is temporarily reflected by the standard diffuser. When the intensity of the reflected light is stronger than the above, the reflectivity exceeds 1. However, when the reflectance exceeds 1 in this way, it is preferable to determine that regular reflection is occurring in the inspection object and exclude it from the object imaging pixel. This is because when specular reflection occurs, it is highly possible that the intensity data of the reflected light does not include information on the inspection object.

  As another method, when the reflectance at a specific wavelength is within a predetermined range, the pixel is relatively less affected by the absorption and reflection of the measurement light by the container, and information on the inspection object is obtained. There is a method of determining that the pixel is a pixel obtained by imaging the content in the container 1. Specifically, for example, in the case where the content is a substance containing moisture, when the reflectance at a wavelength of 1384 nm where an absorption peak derived from water appears is smaller than a predetermined value, the pixel has moisture. This is a method for determining that the pixel contains captured content.

  Furthermore, as another method, when the difference between the reflectance at a specific first wavelength and the reflectance at a second wavelength different from the first wavelength is within a predetermined range, the pixel depends on the container. There is a method for determining that the pixel has a relatively small influence on the measurement light absorption / reflection and can obtain information on the inspection object, that is, a pixel obtained by imaging the contents in the container 1. By combining the above methods, it may be determined whether the pixel is a pixel obtained by imaging the inspection object. The reflectance information of the pixels determined to be the object imaging pixels by the extraction unit 31 is sent to the averaging unit 32.

  The averaging unit 32 obtains an average of spectral intensities for each wavelength from reflectances included in a plurality of object imaging pixels extracted as pixels including information on the inspection object in the extraction unit 31, It has a function of calculating a reflection spectrum. By calculating the average reflection spectrum, one average reflection spectrum is created from one image data. This result is sent to the determination unit 33.

  The determination unit 33 has a function of determining the presence / absence of an abnormality in the inspection target based on the average reflection spectrum calculated by the averaging unit 32. Various methods can be used to determine the presence or absence of this abnormality. For example, there is an abnormality when the ratio between the reflectance at the first wavelength and the reflectance at the second wavelength is within a predetermined range. For example, principal component analysis (PCA), support vector machine (SVM), and the like.

  Then, the determination unit 33 determines whether there is an abnormality in the inspection object imaged with the image data in correspondence with one image data, and outputs the result. The abnormality determination is performed as described above.

  According to the above-described abnormality determination device and the abnormality determination method using the abnormality determination device 100, first, an object imaging pixel including information related to the inspection object is extracted from all pixels included in the spectrum image obtained by imaging. To do. In this way, by limiting the pixels used for the determination, it is possible to reduce erroneous determination caused by noise caused by the container. In addition, the average of the spectrum intensity with respect to each wavelength in the object imaging pixel is obtained, an average reflection spectrum is calculated, and the presence / absence of an abnormality is determined using this. Thus, by calculating and using the average reflection spectrum, it is possible to reduce misjudgments derived from noise or the like included in each pixel, and to detect abnormalities of the inspection object filled in the container with higher accuracy. It becomes possible to detect.

  As mentioned above, although embodiment of this invention was described, this invention is not limited to said embodiment, A various change can be made.

  For example, in the above-described embodiment, the configuration in which the control unit 30 is incorporated in the abnormality determination device 100 has been described. However, the abnormality determination device according to the present invention is incorporated in, for example, a system that performs abnormality analysis of other industrial products. It can also be used. Further, the control unit 30 does not need to be connected to the camera 20 that captures image data as in the above embodiment, and can be used alone.

  EXAMPLES Hereinafter, although this invention is demonstrated more concretely based on an Example and a comparative example, this invention is not limited to a following example at all.

  Image data was captured using the abnormality determination device 1 shown in FIG. Specifically, a container bag was prepared as the container 1, and a reference sample in which a nutritional food was filled as an object to be inspected and a comparative sample in which bacteria were mixed inside the nutritional food were prepared. As the light source 10, a white light source is used, and a linear irradiation region in a direction (y direction) orthogonal to the X direction from a direction that is 45 ± 5 ° downward with respect to the container traveling direction A (x direction). Then, irradiation was performed from two places, and the reflected light from the container 1 was imaged by the camera 20. Imaging of the container 1 was performed by repeating imaging while moving the container 1.

This data was analyzed by the control unit 30. First, in the extraction part 31, the object imaging pixel which imaged the test target object was extracted. As this extraction method, three methods were adopted.
(1) A pixel having a reflectance of 0.35 or less at a wavelength of 1384 nm and a reflectance of less than 1 at a wavelength of 1500 nm was determined as an object imaging pixel.
(2) A pixel in which the ratio of the reflectance at a wavelength of 1300 nm to the reflectance at a wavelength of 1384 nm is less than 1.2 and the condition that the reflectance at a wavelength of 1500 nm is less than 1 is determined as an object imaging pixel.
(3) All pixels were used as object imaging pixels.

For the object imaging pixels extracted by the determination methods (1) and (2) above, an average reflection spectrum obtained by performing an averaging process in the averaging unit 32 (one data for one image data: Example) In the determination unit 33, and the case where the averaging is not performed (the reflection spectrum of each of the object imaging pixels is used: corresponding to the comparative example)
(A) Method of determining that there is an abnormality when the ratio of the reflectance at a wavelength of 1450 nm and the reflectance at a wavelength of 1270 nm is within a predetermined range (B) Principal component analysis (PCA)
The determination was performed by the following two methods. Regarding (3), the average reflection spectrum subjected to the averaging process was determined by the two methods (A) and (B). Table 1 shows the respective detection rates (rates at which the comparative sample was determined to be abnormal) and false detection rates (rates at which the reference sample was determined to be abnormal).

  From the results of Table 1, in both cases (1) and (2), the case where the abnormality determination is performed using the average reflection spectrum is compared with the case where the abnormality determination is performed using the reflection spectrum for each pixel. As a result, the detection rate improved and the false detection rate decreased.

DESCRIPTION OF SYMBOLS 100 ... Abnormality determination apparatus, 1 ... Container, 10 ... Light source, 15 ... Mirror, 20 ... Camera, 30 ... Control part, 31 ... Extraction part, 32 ... Averaging part, 33 ... Determination part.

Claims (6)

On the basis of a spectral image obtained by imaging a container in which an inspection object is stored, an abnormality determination device that determines the presence or absence of abnormality of the inspection object,
Irradiating means for irradiating the container with light of a plurality of wavelengths;
An imaging unit that captures the container and obtains a spectrum image by receiving reflected light from the container by the light irradiated from the irradiation unit;
Extraction means for extracting object imaging pixels including information on the inspection object from all pixels included in the spectrum image obtained by imaging by the imaging means;
An averaging means for calculating an average reflection spectrum by obtaining an average of spectrum intensities for each wavelength in all the object imaging pixels extracted by the extraction means;
Based on the average reflection spectrum calculated by the averaging means, determination means for determining the presence or absence of abnormality of the inspection object;
An abnormality determination device comprising:   The abnormality determination device according to claim 1, wherein the irradiation unit irradiates light having a plurality of wavelengths included in a wavelength range of 1200 to 2400 nm. The extraction means includes, in each pixel included in the spectrum image, a reflectance at a first wavelength included in a wavelength range of 1200 to 2400 nm, and a first wavelength included in a wavelength range of 1200 to 2400 nm. The abnormality determination device according to claim 2, wherein a ratio between reflectances at different second wavelengths is calculated, and based on the result, it is determined whether the pixel is the object imaging pixel. . The extraction means determines whether the pixel is the object based on whether or not the reflectance at the third wavelength included in the wavelength range of 1200 to 2400 nm is within a predetermined range in each pixel included in the spectrum image. It is judged whether it is an imaging pixel. The abnormality determination apparatus of Claim 2 or 3 characterized by the above-mentioned. The extraction means excludes the pixel from the target imaging pixel when the reflectance at the fourth wavelength included in the wavelength range of 1200 to 2400 nm is 1 or more in each pixel included in the spectrum image. The abnormality determination device according to claim 3 or 4, characterized in that: Based on a spectrum image obtained by imaging a container in which an inspection object is stored, an abnormality determination method for determining presence or absence of abnormality of the inspection object,
An irradiation step of irradiating the container with light of a plurality of wavelengths;
An imaging step of capturing a container to obtain a spectral image by receiving reflected light from the container by the light irradiated in the irradiation step;
An extraction step of extracting an object imaging pixel including information on the inspection object from all pixels included in the spectrum image obtained by imaging in the imaging step;
An average step of calculating an average reflection spectrum by obtaining an average of spectral intensities for each wavelength in all the object imaging pixels extracted in the extraction step;
Based on the average reflection spectrum calculated by the averaging step, a determination step for determining presence or absence of an abnormality of the inspection object;
The abnormality determination method characterized by having.

Images