JP5174540B2 - Wood defect detection device - Google Patents

Description

  The present invention relates to a wood defect detection device, and more particularly to a wood defect detection device that detects defects on a thin and sheet-like single plate surface that is bonded to a plywood or a plywood surface that is bonded to a single plate.

  Japanese Laid-Open Patent Publication No. 2007-147442 discloses a technique for detecting defects on the surface of wood based on color distribution. In this case, since the defect can be detected accurately in consideration of the color, the system can be expensive compared with the case of using a monochrome image. In addition, here, since the defect is determined using a complicated function for the defect determination and the defect pattern collected in advance, the inspection process (determination process) is not fast and the defect is determined in advance. There is a problem that it is necessary to grasp the pattern.

On the other hand, there is one disclosed in Japanese Patent Application Laid-Open No. Hei 8-145914 which can detect a defect in a monochrome image. This is because a one-dimensional TV camera is used, and although it is inexpensive, it has the problem that the brightness difference between the defective part and the non-defective part cannot be obtained sufficiently and cannot be detected stably. Since the appearance in the captured image differs depending on the state, defect detection may not be reliably performed. This can be attributed to factors such as the surface state and moisture content that depend on the inspection target and factors that depend on the luminance level of the defect portion.
JP 2007-147442 A JP-A-8-145914

  The present invention was invented in view of the above-described conventional problems, and it is an object of the present invention to provide a wood defect detection apparatus that can perform stable defect detection while being inexpensive. An object of the present invention is to provide a wood defect detection device capable of detecting defects at high speed by simple processing.

In order to solve the above problems, a wood defect detection apparatus according to the present invention includes an illumination unit that illuminates the surface of the inspection wood that is transferred at a predetermined speed, and an imaging unit that images the surface of the inspection wood illuminated by the illumination unit. consists of a processing means for detecting a defect of inspection wood surface based on the image captured by the imaging means, the illumination means irradiates light in the wavelength range 420nm~530nm the inspection surface of the inspection timber There, the processing means state, and are not to extract the defect portion based on luminance information of each pixel in the captured image, the processing means is a predetermined threshold or less of the luminance and the luminance value A region having an area of a predetermined area equal to or greater than a predetermined value is extracted as a defect area, and when the area value of a defective area having a luminance value equal to or lower than a predetermined luminance value is smaller than a predetermined value for determining a defect, the predetermined luminance value or lower In the area An area value of a defect region having a luminance value equal to or lower than a predetermined luminance value is determined as a defect according to an angle of gradient from the lowest luminance of the luminance distribution centering on the lowest luminance pixel to a predetermined luminance value. Determining a defect according to the angle of the gradient from the lowest luminance of the luminance distribution centering on the lowest luminance pixel in the area below the predetermined luminance value to the predetermined luminance value when the value is smaller than the predetermined value for determining the defect It is characterized by being. By using light of the above-mentioned wavelength, a high contrast can be obtained between the normal part (non-defective part) and the defective part. For this reason, the defective part can be determined well. And since the area | region which is the brightness | luminance below a predetermined threshold value and the area of the area | region which has the luminance value is more than a predetermined value is extracted as a defect area | region, what includes a defect can be extracted reliably. Further, when the area value of the defective area having a luminance value equal to or lower than the predetermined luminance value is smaller than the predetermined value for determining as a defect, the luminance distribution in at least four directions centered on the lowest luminance pixel in the defective area is any direction. In this case, by determining a defect having at least one inflection point as a defect, a node defect that is originally a defect but is not determined as a defect by area can be determined as a defect. Further, when the area value of the defective area having a luminance value equal to or lower than the predetermined luminance value is smaller than the predetermined value for determining as a defect, the predetermined value is determined from the lowest luminance of the luminance distribution centering on the lowest luminance pixel in the area equal to or lower than the predetermined luminance value. By determining a defect according to the angle of the gradient until the luminance value is exceeded, the defect type can be stably classified even in a defect region having the same minimum luminance.

  At this time, it is preferable that the illuminating means illuminates light substantially parallel to the fiber direction of the inspection wood at an angle of 40 to 75 ° with respect to the inspection surface of the inspection wood. Defects can be detected with a simple configuration of the imaging unit and processing unit utilizing the characteristics of wood, and various defects can be detected without collecting defect patterns in advance.

  The processing means extracts, as a defect candidate area, an area having a luminance equal to or lower than a predetermined threshold and an area having the luminance value equal to or lower than the predetermined value, and the predetermined area centered on the defect candidate area In this case, when the peripheral brightness is high, the threshold level is increased, and when the peripheral brightness is low, the threshold level is decreased to remeasure the area value of the region, and the defect determination is performed based on the area value. Since the threshold value is adjusted according to the peripheral luminance, the defect can be reliably detected even when the luminance difference between the defective portion and the non-defective portion is not sufficient.

  The processing means sets a threshold for binarization based on the average luminance in a predetermined area where the lowest luminance pixel of the defective area exists and the average luminance of all captured images, and based on the threshold. If the defect area is extracted, the brightness of the peripheral area (non-defective part area) can be reflected to accurately extract the defect area.

  The processing means obtains a defect area using a plurality of threshold values, obtains a defect area at an overlapping position among defect areas determined to have a defect at each threshold value, and further overlaps the defect area. If the defect area is to be deleted while leaving the largest defect area among the defect areas, the defect area can be stably extracted even if the brightness of the defect portion is not stable due to the surface condition of the inspection wood or the imaging state. Can do.

  Furthermore, if the processing means is to determine the surface color unevenness defect based on the average luminance difference between the predetermined areas on the surface of the inspection wood, the color unevenness defect that has been conventionally determined from the color distribution using a color image is also included. Can be detected.

  Since this invention uses what irradiates the light of the wavelength range of 420 nm-530 nm to the test | inspection surface of test | inspection wood as an illumination means, it can obtain high contrast with a normal part (good part) and a defective part. For this reason, it is possible to satisfactorily determine the defective portion based on the luminance.

  Hereinafter, the present invention will be described based on an embodiment shown in the accompanying drawings. The illustrated example is for detecting various defects appearing on the surface of a thin sheet-like single plate attached to the surface of a plywood. As shown in FIG. 1 and FIG. 2, at least one camera 2 is arranged above the inspection wood 1 conveyed by the conveyor 4, and a light source 3 that illuminates a portion that is an imaging range by the camera 2 is arranged. . A line sensor type camera is used as the camera 2 as the imaging means, and oblique light illumination is given to the inspection wood 1 by using a light emitting diode as the light source 3 as the illumination means.

  Here, a light source 3 that outputs light having a wavelength of 420 nm to 530 nm is used. FIG. 3 shows the result of measuring the reflectance of the normal part of the inspection wood 1 and the reflectance of the defective part (cane stripe defect) by changing the wavelength. In this case, since the contrast between the normal part (non-defective part) and the defective part becomes large in the range of 420 to 530 nm, the light having the above wavelength is used. As is apparent from the figure, 430 nm to 480 nm. Since the contrast is maximized at a wavelength of 1, it is possible to perform better defect detection.

  Further, the illumination by the light source 3 is applied substantially parallel to the fiber direction S of the inspection wood 1 and from 40 to 75 °, in the illustrated example, from 60 ° above. Since there are generally conduits for supplying moisture from the wood in the fiber direction S, irregularities appear in the direction perpendicular to the fiber direction S as shown in FIG. In addition, when there are grain, a defect, etc., the uneven shape may be different from other unevenness. This is because the shape is large and the color of the uneven part is different from other uneven parts. By irradiating light parallel to the fiber direction S from diagonally above, other uneven parts such as grain and defects can be obtained. It can be emphasized rather than the unevenness. Furthermore, the light penetrating into the wood is not easily affected by the unevenness, and the light easily returns to the surface (is easily diffused). On the contrary, when it irradiates from the direction orthogonal to the fiber direction S, it becomes difficult to identify wood grain and defects. There may be irregularities that run in directions other than the fiber direction S, but the amount is minute and not a problem level.

  FIG. 5 shows the measurement of average brightness when light is applied to three kinds of wood (hip (M1), oak (M2), and beech (M3)) by changing the angle of illumination by the light source 3. However, as is apparent from the figure, since a high reflection luminance can be obtained when the angle is 40 ° to 75 °, especially around 60 °, a large amount of reflected light for defect detection can be obtained. .

  The output image of the camera 2 is subjected to the following processing in the processing means 4 shown in FIGS. That is, the processing means 4 comprising an image processing computer is shown in FIG. 6 for each pixel of the two-dimensional captured image obtained by capturing the output of the line sensor type camera 2 in synchronization with the conveyance of the inspection wood 1. As shown in the figure, the threshold value is changed stepwise and binarization processing is performed for each threshold value to group pixels, and an area is extracted from the captured image. Then, the area of the areas grouped by this process is compared with a standard area set in advance for each detection threshold, and if the area is equal to or larger than the standard area, a process of determining the area as a defective area is performed.

  By the way, depending on the state of the picked-up image, the edge portion may become loose (the number of pixels rising from the edge portion becomes large), and the area may not be detected accurately from the image. In order to delete the “full” edge, it is preferable to reset the threshold value and perform binarization processing. FIG. 7 shows a flow in this case. When the flow up to the step “collection of detected pixels for each threshold” in the flow shown in FIG. Continue to run.

  As in the case shown in FIG. 6, first, the area is compared with a preset area for each detection threshold value. If the area is equal to or larger than the area, the area is determined as a defect candidate area, and then the defect candidate area is measured once. Then, the binarization process is performed again by changing the binarization threshold value, and the defect candidate area that is equal to or smaller than the preset area is checked again. A fixed value is given as the value A when changing the threshold value. For example, when the initial detection threshold is 30, and the luminance difference between the defective portion minimum luminance and the non-defective portion average luminance is 110, the defect detection is performed using 55 which is ½ of the luminance difference as the value of A. A value 85 added to the threshold value is set as a threshold value. The threshold value is adjusted according to the peripheral luminance. Then, the determination is performed by re-measuring the area of the area. By performing this process, as shown in FIG. 8, it is possible to detect a defect area that is larger than the detected defect size S1 and substantially coincides with the defect size S0 that can be visually confirmed. Separation from the peripheral luminance can be reliably performed, and the defective portion can be accurately extracted.

  By the way, as shown in FIG. 9, when the inspection wood 1 is a laminate of a plurality of single plates (pieces) on a plywood, a piece P1 where the defect 9 exists and a piece P2 where the defect 9 does not exist When the average brightness is different and the threshold is set with the brightness level of a defect present in any piece when the threshold is set, the defect is detected as a defect in the case where the defect is detected by the piece P1 and in the case of the piece P2. A difference occurs in the area to be detected.

  FIG. 10 shows a flow for dealing with this point. By using the average brightness of the pieces present in the captured image as a reference, taking into account the difference from the average brightness of the pieces with defects, the threshold value can be varied and the average brightness of all pieces can be varied as a reference. The threshold value is set. That is, the threshold value is changed by the value of “total average luminance−defect existence piece average luminance”. Note that the boundary of each piece is determined by detecting the end of the captured image and assigning the piece boundary so as to match the end.

  In addition, the defect area in the captured image has the same threshold value depending on the state of the imaging system (focus blur of the camera 2, aberration of the lens of the camera 2, etc.) and the state of the inspection wood 1 (difference in reflected light amount due to moisture content, etc.). Since it is not detected but is detected from a plurality of threshold regions, a plurality of defect regions A1, A2, A3 overlap as shown in FIG.

  In such a case, the defect determination is performed in the area where the detection area is the maximum, thereby preventing the overlap of the defect areas. FIG. 12 shows another defect having a centroid position within a predetermined range centered on the centroid coordinate of the area detected by determining the defect at the threshold value N in order to delete (delete) the overlapping defect area. If there is a region, the recording of the defective region (defect region having a small area) is deleted.

  The predetermined range is preferably set by a radius corresponding to a defect size having a maximum standard area value (usually a standard area value applied to a region detected using the maximum luminance value as a threshold value). In addition, this process is preferably performed from an area detected by setting a high-luminance threshold value (an area detected by setting a high threshold value).

  There are various types of defects, and it is necessary to detect a node defect in the defect as a defect even when the prescribed area value is not satisfied. Therefore, when detecting a defect, it is necessary to detect whether or not it is a node defect. However, when the luminance distribution is examined for a portion having a node defect, the luminance distribution is centered on the lowest luminance (region centroid position) in the defect region. It was found that the three patterns shown in FIG. 13 can be classified. From this characteristic, defects can be classified when there are one or more inflection points, and when there are one and two or more inflection points even in the same nodal defect. The normal defect is the same as the luminance distribution shown in FIG. 13A, and in the case of a node defect, the luminance distribution as shown in FIGS. 13B and 13C also exists. It is possible to determine a nodal defect even with a simple classification.

  Based on this point, as shown in FIG. 14, the center of gravity position of the defect area is estimated, and the luminance distribution of the area twice the size of the circumscribed rectangle of the area centered on the obtained center of gravity position is displayed in four directions ( Measurement is performed in eight directions from the position of the center of gravity, and further, whether or not there is an inflection point in the luminance distribution is measured. And if it is two of the above four directions and there are two or more inflection points, it is recorded as a life defect, and if there are one or more inflection points in two directions, it corresponds to the threshold value. Defect detection independent of the area of recording as a defect area is performed.

  Further, as shown in FIG. 15B, the luminance change up to the peripheral luminance, such as a node defect, has a luminance distribution that suddenly reaches the peripheral luminance from the lowest luminance level, as shown in FIG. As described above, there are two types of defects having a luminance distribution that reaches the peripheral luminance pattern while the luminance changes gently. Incidentally, in the case shown in FIG. 15A, the luminance change from the lowest luminance level to the peripheral non-defective part is about 10 pixels, and in the example shown in FIG. 15B, it is about 20 pixels.

  These defects do not satisfy the standard area value of the defect depending on the setting of the luminance value of the threshold value, and even when the defect is not determined, it is determined as a defect from the characteristics of this luminance distribution, thereby performing stable defect detection. be able to.

  FIG. 16 shows a flow for dealing with this point. First, defect determination is performed by area, and when the area value is not satisfied, determination is made using a luminance distribution. The luminance distribution is centered on the position of the center of gravity of the defect candidate, and measures twice the size of the circumscribed rectangle of the defect on both sides (both directions) from the center. Further, this luminance distribution is measured every 45 degrees in four directions passing through the center of gravity position, and when the luminance change is equal to or larger than a predetermined inclination in two or more directions, it is determined as a defect.

  In addition, in the case where a plurality of single plates (pieces) are bonded to the plywood surface, it is preferable to treat a piece having a large difference from other pieces as a defect. In this regard, until now we have measured and evaluated the color difference (color distribution difference) of these pieces with color images, but here we are emitting light in the wavelength range where the luminance contrast is maximum. Is used for the same detection as a color image.

  Specifically, as shown in FIG. 17, area division is performed for each piece to obtain an average luminance for each piece, and when a luminance difference between a piece and an adjacent piece is equal to or greater than a predetermined value, Judged to have a defect. For example, the average brightness of the piece P1 is 68, the average brightness of the piece P2 is 55, the average brightness of the piece P3 is 70, the average brightness of the piece P4 is 65, and the pieces P1 and P3 are in the piece P2. , P4 are adjacent to each other, the average luminance of the piece 2 has a luminance difference greater than or equal to a predetermined value with respect to the peripheral pieces (pieces P1, P3, P4).

It is a schematic block diagram of an example of an embodiment of the invention. It is a perspective view same as the above. It is explanatory drawing of the relationship between a reflectance, a reflection ratio, and a wavelength. It shows the unevenness of the surface of the inspection wood, (a) is a plan view, (b) is a cross-sectional view along the line XX. It is explanatory drawing of the relationship between an irradiation direction and a brightness | luminance. It is a flowchart same as the above. It is a flowchart of another example. It is explanatory drawing of an example of luminance distribution. It is a top view of an example of the inspection wood surface. It is a flowchart of other examples. It is explanatory drawing of an example of the binarized image by a some threshold value. It is a flowchart of another example. (a), (b), and (c) are explanatory diagrams of the luminance distribution in the case of a nodal defect. It is a flowchart of another example. (a) (b) is explanatory drawing of the luminance distribution in the case of a node defect, respectively. It is a flowchart of another example. It is a flowchart of another example.

Explanation of symbols

1 Inspection wood 2 Camera 3 Light source 4 Processing means

Claims (7)

Illuminating means for illuminating the surface of the inspection timber transferred at a predetermined speed, imaging means for imaging the surface of the inspection timber illuminated by the illuminating means, and defects on the surface of the inspection timber based on the image captured by the imaging means And processing means for detecting
It said illuminating means is adapted to irradiate light in a wavelength range of 420nm~530nm the inspection surface of the inspection timber,
It said processing means state, and are not to extract the defect portion based on luminance information of each pixel in the captured image,
The processing means extracts a region having a luminance equal to or lower than a predetermined threshold and the area of the luminance value having a luminance value equal to or larger than a predetermined value as a defective region, and a defective region having a luminance value equal to or lower than the predetermined luminance value When the area value is smaller than a predetermined value for determining a defect, the luminance distribution in at least four directions centered on the lowest luminance pixel in the defect region has at least one inflection point in any direction. A wood defect detection device characterized by being determined as a defect. Illuminating means for illuminating the surface of the inspection timber transferred at a predetermined speed, imaging means for imaging the surface of the inspection timber illuminated by the illuminating means, and defects on the surface of the inspection timber based on the image captured by the imaging means And processing means for detecting
The illuminating means irradiates the inspection surface of the inspection wood with light having a wavelength range of 420 nm to 530 nm,
The processing means extracts a defective portion based on luminance information of each pixel in a captured image,
The processing means extracts a region having a luminance equal to or lower than a predetermined threshold and the area of the luminance value having a luminance value equal to or larger than a predetermined value as a defective region, and a defective region having a luminance value equal to or lower than the predetermined luminance value When the area value is smaller than the predetermined value for determining the defect, the defect is determined according to the angle of the gradient from the lowest luminance of the luminance distribution centering on the lowest luminance pixel in the area below the predetermined luminance value to the predetermined luminance value. it characterized in that to determine the wood defect detection apparatus. 3. The illumination means illuminates light substantially parallel to the fiber direction of the inspection wood at an angle of 40 to 75 [deg.] With respect to the inspection surface of the inspection wood. Wood defect detection device. The processing means extracts a region having a luminance equal to or lower than a predetermined threshold and an area having the luminance value equal to or lower than a predetermined value as a defect candidate region, and in a predetermined region centered on the defect candidate region When the peripheral brightness is high, the threshold level is increased, and when the peripheral brightness is low, the threshold level is decreased, and the area value of the region is remeasured, and the defect determination is performed based on the area value. The wood defect detection apparatus according to any one of claims 1 to 3, wherein the detection is performed. The processing means sets a threshold value for binarization based on the average luminance in a predetermined area where the lowest luminance pixel of the defective area exists and the average luminance of all captured images, and the defect is determined based on the threshold value. The wood defect detection device according to any one of claims 1 to 4, wherein the region is extracted. The processing means obtains a defect area using a plurality of threshold values, obtains a defect area at an overlapping position among defect areas determined to have a defect at each threshold value, and further overlaps the defect area. The wood defect detection device according to any one of claims 1 to 5, wherein the defect region is deleted while leaving the largest defect region among the defect regions. The wood defect detection according to any one of claims 1 to 6, wherein the processing means determines a surface color unevenness defect based on an average luminance difference between predetermined regions on the surface of the inspection wood. Equipment .