JP4075799B2 - Method for detecting powder particles - Google Patents

Description

  The present invention relates to a method for detecting a granular material that automatically detects the filling state of the granular material on a holding body.

  The need to automatically determine and detect the filling state of powder particles on a holder is required in a wide range of industries and the like. For example, the need for automatically detecting the state of filling of industrial materials and products on a conveyor such as a belt conveyor is very high in terms of product management.

  Japanese Unexamined Patent Application Publication No. 2002-221481 (Patent Document 1) discloses a technique that uses image processing for a granular material. It is a device that measures the particle size by image processing on coal moving on a conveyor such as a belt conveyor, and has been devised to add a baffle plate to eliminate distance fluctuations between the imaging device and the subject. Yes.

  Moreover, there exists a indication of Unexamined-Japanese-Patent No. 2003-83868 (patent document 2) as what added the device to this image processing method for a particle size distribution measurement. This is because a “blurred image” obtained by performing “blurring processing” on the original image is obtained from the captured original image, and the “blurred image” is binarized to obtain a measurement target having a predetermined particle diameter or more. While measuring the distribution of the object, and binarizing the “difference image” formed by the difference between the captured original image and the “blurred image”, the particle size of the measurement object less than the predetermined particle size This is a particle size distribution measurement method for measuring the distribution and measuring the entire particle size distribution based on the measurement results of these two types of particle size measurement distributions.

Furthermore, there is a disclosure of Japanese Patent Laid-Open No. 5-126749 (Patent Document 3) as a similar technique although the object is different. This is because, in order to evaluate the similarity of the melon mesh tension in a box unit in a boxed melon, the power spectrum image is obtained from each melon image, and the commonality of the luminance distribution of each power spectrum is examined. Yes.
Japanese Patent Laid-Open No. 2002-221481 JP 2003-83868 A Japanese Patent Laid-Open No. 5-126749

  In the methods disclosed in Patent Document 1 and Patent Document 2, binarization processing is performed with a predetermined threshold after normalization processing as preprocessing in order to obtain a particle image. The normalization processing is for removing low-frequency gradual luminance undulations, and there are various methods such as shading correction processing, filtering by Fourier transform, and filtering by wavelet transform.

  FIG. 6 is a diagram illustrating an example of a preprocessing process, and illustrates an example in which filtering by wavelet transform is applied to an image obtained by capturing coal particles conveyed by a belt conveyor. The figure shows (1) the original image, (2) the normalized image, and (3) the binarized image from the top, respectively. (3) Extract the area where the belt conveyor surface is exposed as in the binarized image. can do. When an isolated region having an area larger than the predetermined threshold exists, it is determined that the coal particles are not filled. That is, when the exposure on the surface of the belt conveyor is relatively large, it is possible to detect the filling of the coal particles by assuming that “the granular particles larger than the predetermined threshold cannot exist”.

  However, in practice, it is often necessary to detect the exposure of the belt conveyor surface that is smaller than the assumed size of the granular material. In FIG. 6 (3), the presence of particles larger than the area remaining after binarization is sufficiently considered, and conversely, the belt conveyor equivalent to the area remaining after binarization in (3) If the exposure is not automatically detected, a large error occurs in the particle size distribution measurement. That is, there is a problem that it is not sufficient to evaluate the exposure of the belt conveyor solely by its area rather than the image.

  In addition, Patent Document 3 is characterized in that a specific melon image is compared with other melon images, and so-called pattern matching with other images is performed for the purpose of evaluating the similarity of melon mesh tension. It is said. That is, in Patent Document 3, it is indispensable to collect a plurality of images, and there is a problem that it is not possible to meet the need for examining the filling state only from a specific image intended by the present invention.

  The present invention has been made in view of the above circumstances, and detection of a granular material that determines the filling state of the granular material by utilizing the difference in surface properties between the surface property of the holding body as a background and the granular material. It is to provide a method.

  The present invention captures the state of filling of the granular material on the holding body with a camera, performs a two-dimensional Fourier transform on the obtained image, and within a power spectrum distribution in a frequency space, a predetermined low frequency region and It is a method for detecting a granular material, characterized in that the filling state of the granular material is determined from an average value of a power spectrum in a high frequency range.

  Moreover, this invention is the detection method of the granular material of Claim 1, Comprising: The ratio of the average value of the power spectrum of the said low frequency region and a high frequency region is compared with a predetermined threshold value, and the filling state of a granular material Is a method for detecting a granular material.

  Furthermore, this invention is the detection method of the granular material of any one of Claim 1 or Claim 2, WHEREIN: The said granular material is a coal particle, The detection method of the granular material characterized by the above-mentioned. is there.

  According to the present invention, the obtained image is subjected to two-dimensional Fourier transform, and the ratio of the average values of the power spectra in a predetermined low frequency region and high frequency region in the power spectrum distribution in the frequency space is determined as a predetermined threshold value. Therefore, it is possible to automatically detect the filling state of the granular material on the holding body.

  The present invention seeks to detect the filling state of a granular material by utilizing the difference between the surface property of the holding body as a background and the surface property of the granular material. The best mode for carrying out the present invention will be described below with reference to the drawings. FIG. 2 is a diagram illustrating an example of a device configuration for carrying out the present invention, and is an example in which a belt conveyor is used as a holding body and coal particles are used as a granular material. In the figure, 1 is a coal particle, 2 is a belt conveyor, 3 is an area sensor camera, 4 is a camera controller, 5 is a storage device, 6 is a computing device, and 7 is a display device. First, the area sensor camera 3 controlled by the camera controller 4 images the state of the scattered coal particles 1 conveyed by the belt conveyor 2.

  The area sensor camera 3 is installed at a position where the belt conveyor 2 and the coal particles 1 can be sufficiently confirmed using a two-dimensional CCD camera or the like. Ideally, the image collection conditions are selected from values that match the speed of the belt conveyor 2 and the field of view range of the area sensor camera 3, and a place that well represents the filling state of the coal particles 1 is ideally selected.

  The obtained image is immediately stored in the storage device 5 for recording, and a calculation process for obtaining the filling state is performed in the arithmetic device 6 and the result is displayed on the display device 7 as a detection result of the filling state. Since the arithmetic processing in the arithmetic unit 6 is the core of the present invention, it will be described in detail below.

  FIG. 1 is a diagram for explaining the flow of a filling state detection process by image processing in the arithmetic device 6. The processing is started at the beginning (S100), and first, an image captured by the area sensor camera 3 is collected (S101).

  Next, two-dimensional Fourier transform (S102) is performed on this image to perform spectrum analysis. The spectrum analysis by the two-dimensional Fourier transform can be easily performed by first performing a one-dimensional Fourier transform in a certain direction and further performing a one-dimensional Fourier transform in a direction orthogonal to the result. As the one-dimensional Fourier transform, for example, the well-known FFT (Fast Fourier Transform) by Cooley and Turkey may be used.

FIG. 3 shows an original image (left side) with different coal particle filling states and a spectrum distribution diagram (right side) after each two-dimensional Fourier transform. The two-dimensional spectrum distribution diagram is obtained by performing spectrum analysis up to the Nyquist frequency for both the vertical (f _v ) and horizontal (f _h ) axes, and expanding the low frequency region of the result. Is different from the square in the vertical and horizontal directions.

  In this two-dimensional spectrum distribution diagram, the power spectrum of the corresponding frequency is expressed by the density of the figure, that is, from light to dark. The original image shows that the coal particles change from dense to sparse, that is, the filling rate changes from large to small as it goes from the top of Fig. 3 (A) to the bottom of (E). The proportion is increasing. Looking at the two-dimensional spectral distribution map, it can be seen that the lower the density, the clearer the difference between the outer part and the central part.

  FIG. 4 is a diagram in which the change of the power spectrum due to the filling state of the coal particles is schematically written in one dimension. In a state where the coal particles are densely packed (the state where the belt conveyor surface does not come out: the top of FIG. 4), a flat power spectrum that extends to the high frequency range caused only by the surface properties of the coal particles. Appears. Then, from the top to the bottom of the figure, the packing of coal particles from dense to sparse, that is, as the properties of the belt conveyor surface are exposed, another component appears with a spectral distribution with strong power in the low frequency range. . This other component is attributed to the properties of the belt conveyor surface, and when the conveyor surface is completely exposed as shown in the bottom of FIG. 4, the component widely spread from the low frequency region to the high frequency region disappears. In the intermediate stage of FIG. 4, it is expressed by superposition of the two spectral components described above.

  The two-dimensional spectrum distribution diagram of FIG. 3 can be easily explained by extending the above one-dimensional case. That is, when the conveyor surface is almost completely exposed as shown in Fig. 3 (E), the power of the central part (low frequency range) of the two-dimensional spectral distribution due to the properties of the belt conveyor surface becomes large (the lightest in the figure). The outer portion has almost no power (the darkest in the figure). On the other hand, when almost covered with coal particles as shown in FIG. 3 (A), it spreads widely from the central part (low frequency range) due to only the surface properties of the coal particles to the outer part (high frequency band). A flat power spectrum (in the figure, there is little difference in shading between the central part and the outer part) appears. Note that the two-dimensional spectral distribution in FIG. 3 shows the two components of the coal particles and the belt conveyor, the DC component, the lowest frequency band, the brightness offset, and the gentle brightness undulation caused by the illumination system. It is comprised with what produces | generates.

  When the spectrum analysis by the above two-dimensional Fourier transform (S102) is completed, the process proceeds to the step of calculating the spectrum unevenness degree (S103). Here, an index that characterizes the change in the power spectrum of the image according to the exposure of the surface property of the belt conveyor in the frequency space, the degree of spectral unevenness, is introduced.

  FIG. 5 is a diagram showing the spectrum distribution and the region setting in the frequency space in the horizontal direction and the vertical direction. As described above, the shading in the figure shows the power weakness, the central part is in the low frequency region, and the peripheral part is Each of the high frequency ranges is shown. In this step, first, two regions of a high frequency region B indicated by a right-upward oblique line and a low frequency region A indicated by a right-downward oblique line are defined in the upper part of FIG.

  Here, the region A does not include the DC component and the nearest region. The spectrum of the area shown in white inside the area A is often related to the brightness offset and the gentle brightness undulation caused by the illumination system. This may be due to different factors. In addition, for the region B, the region B is not defined continuously from the region A, but a buffer region that is a white portion is provided between the regions A. Will increase.

Next, power averages E _A and E _B are calculated for region A and region B, respectively. Further, using the calculated E _A (average power in the low frequency band A) and E _B (average power in the high frequency band B), the spectral unevenness R _F is calculated by the following equation (1). .
R _F = E _A / E _B (1)
The spectrum is biased toward the low frequency region as the spectrum unevenness degree R _F is larger, and the spectrum is spread over the entire frequency space as the spectrum unevenness degree R _F is smaller. That is, a large degree of exposure of the spectrum ubiquitous degree R _F larger the belt conveyor surface, a dense packing degree is large filling enough granule spectrum ubiquitous degree R _F is small. Note that the zone setting shown in FIG. 5 is an example applied to coal used for coke production, and is actually appropriate depending on the difference in surface properties between the support and the granular material that are the background. The setting should be changed.

  If the degree of spectrum unevenness is obtained, it is compared with a preset threshold value in the next step of spectrum unevenness degree <threshold value (S104). If the degree of spectrum unevenness is smaller than the threshold value, it is determined that the granular material is filled (S105), and if the reverse case, it is determined that the granular material is not filled (S106). The determination criteria in this case should also be changed as appropriate according to the difference in surface properties between the holding body and the powder and granular material, as described in the above-described zone setting.

  Then, in the result display (S107), the determination result and the spectrum uneven distribution degree are displayed on the display device to notify the operator of the detection result. Thereafter, a series of processing is completed at the end (S108).

  FIG. 7 shows an example of a result obtained by applying the present invention to detection of a packed state of coal particles used for coke production. In the figure, the horizontal axis represents the degree of coal filling with numbers from 0 to 5, and the vertical axis plots the degree of spectral unevenness corresponding to each. As the number on the horizontal axis increases, the degree of coal filling decreases, that is, the degree of exposure of the belt conveyor surface increases. However, it is a qualitative grouping of what is visually determined to have almost the same filling degree in six stages, and it should be noted that the same degree of filling is not guaranteed in a strict sense.

  For the above reasons, although there is variation in the degree of spectral unevenness in each group, it can be clearly confirmed that the value of the degree of spectral unevenness decreases as it goes to the left side of the figure, that is, as the coal filling degree increases. In this example, if the threshold value is set to 1.53, it can be seen that it is possible to automatically determine when coal filling is good or not. As described above, the coal filling on the belt conveyor can be confirmed by the image processing of the captured image, and it can be automatically determined whether the obtained image is suitable for the particle size distribution measurement processing.

  In addition to the coal described so far, the detection target can also be applied to various raw materials such as ore, limestone, coke, plastic powder, toner, and other industrial raw materials and products, and agricultural products and medicines.

It is a figure explaining the flow of the detection process of the filling state of the granular material by image processing. It is a figure which shows an example of the apparatus structure for implementing this invention. It is a figure which shows the spectrum after the two-dimensional Fourier transform of the original image from which the filling state of coal particles differs. It is the figure which wrote typically the change of the power spectrum by the filling state of the coal particle in one dimension. It is a figure which shows the spectrum distribution and area | region setting in a frequency space. It is a figure which shows an example of the pre-processing process in the conventional image processing. It is a figure which shows an example of the result applied to the detection of the filling state of coal particles.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Coal particle 2 Belt conveyor 3 Area sensor camera 4 Camera controller 5 Memory | storage device 6 Arithmetic device 7 Display device

Claims (3)

The filling state of the powder on the holding body is imaged with a camera, the obtained image is subjected to two-dimensional Fourier transform, and the power in a predetermined low frequency region and high frequency region in the power spectrum distribution in the frequency space. A method for detecting a granular material, comprising determining a filling state of the granular material from an average value of a spectrum. In the detection method of the granular material of Claim 1,
A method for detecting a granular material, comprising determining a filling state of the granular material by comparing a ratio of an average value of power spectra in the low frequency region and the high frequency region with a predetermined threshold value. In the detection method of the granular material of any one of Claim 1 or Claim 2,
The said granular material is a coal particle, The detection method of the granular material characterized by the above-mentioned.

Images