JP2014122827A - Type discrimination device for object - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide the type discrimination device of an object capable of accurately discriminating the type of the object based on the size of the object without depending on the loading state or dry/wet state of the object.SOLUTION: A type discrimination device of an object comprises: height image creation means 40 for acquiring the height image information of an object by three-dimensional image measurement; smoothing processing means 41 for smoothing the height image information acquired by the height image creation means to acquire smoothed height image information; difference processing means 42 for acquiring difference image information between the height image information and the smoothed height image information acquired by the smoothing processing means; a first primary differentiation processing means 43 for performing primary differentiation processing to the difference image information acquired by the difference processing means to calculate difference image primary differentiation information; first area processing means 44 for calculating the area of a region in which the value of the difference image primary differentiation information acquired by the first primary differentiation processing means is equal to or more than a first reference value; and first discrimination means 45 for comparing the area obtained by the first area processing means with a first discrimination value to discriminate the type of the object.

Description

  The present invention relates to an object type discriminating apparatus for an object containing particles of different sizes, for example, an object such as concrete aggregate.

  As stipulated in JIS A 5005 as crushed stone and sand for concrete, there are various sizes of crushed stone, crushed sand and / or recycled aggregate and the like in the concrete aggregate.

  In general, in batch-type concrete manufacturing plants, various concrete aggregates transported by trucks, etc. are stored in aggregate storage units designated by type and size, respectively. Aggregates are selected from these aggregate stores to produce concrete. Therefore, in this type of concrete production, it is essential that the concrete aggregate transported and stored in the aggregate storage section is the correct type and size of aggregate, and the technology for automatically distinguishing between them It is desirable to have

  In Patent Document 1, as a concrete aggregate discrimination system, vibration generated at the time of receiving aggregate is measured, and the measured vibration is subjected to spectrum analysis to determine the amplitude of each frequency. Based on the calculated amplitude, Discriminating the type is disclosed.

  In Patent Document 2, as a concrete aggregate sorting device, an image of a large aggregate is obtained by illuminating and photographing the conveyed aggregate and expanding the edge portion of the obtained binary image. It is disclosed that a large block aggregate is discriminated by extracting and comparing the size and / or area of the large block aggregate with a reference value from the image.

JP 2004-216781 JP2008-2127778

  However, since the concrete aggregate discrimination system disclosed in Patent Document 1 measures vibration during movement, it is necessary to move the aggregate with a transport belt or the like, and the bone that has been transported by a truck or the like. It was impossible to discriminate the type while the material was mounted on the truck. Furthermore, although it was possible to distinguish between sand and gravel, the type based on the size of gravel could not be discriminated at all.

  Further, the concrete aggregate sorting device disclosed in Patent Document 2 can discriminate the massive aggregate by comparing the size of the aggregate with a reference value, but the difference in the size of gravel. It was not possible to discriminate the type of sand or the type of aggregate such as sand.

  Furthermore, as disclosed in Patent Document 2, a method for discriminating an image obtained by capturing an illumination reflected light of an aggregate as it is is based on a difference in color of the aggregate, a difference in brightness, There is a possibility that an erroneous determination is made due to a difference in loading form (stacking method) and / or a difference in wetting method. For example, if the gravel color is different, the brightness is uneven, the gravel is partially piled up, or the sand is partially wet, an erroneous determination will be made .

  In addition, the technology for discriminating the size and shape by imaging an article has already been used in sorting devices for fruits, vegetables, etc., but these discriminating technologies are applied only when a single image of an object can be obtained. It is a possible technology, such as concrete aggregates mounted on transport means such as trucks and conveyor belts, many objects are irregularly mounted, such as being piled up in part in a mountain shape, It was totally impossible to apply when part was wet.

  Therefore, the object of the present invention can be determined even when the object is irregularly mounted, and the type based on the size can be accurately determined without depending on the mounting state. The object is to provide an object type discrimination device.

  Another object of the present invention is to provide an object type discriminating apparatus capable of accurately discriminating the type based on the size without depending on the wet and dry state of the object.

  According to the present invention, height image creation means for obtaining height image information of an object by three-dimensional image measurement, and smoothed height image information obtained by smoothing height image information obtained from the height image creation means. Smoothing processing means to obtain, difference image means for obtaining difference image information of height image information and smoothed height image information obtained from the smoothing processing means, and differential differentiation of difference image information obtained from the difference processing means The first primary differential processing means for calculating the differential image primary differential information, and the area of the region where the value of the differential image primary differential information obtained from the first primary differential processing means is equal to or greater than the first reference value. A first area processing means for determining the object, and a first determination means for determining the type of the object by comparing the area obtained from the first area processing means with a first determination threshold. A type discriminating apparatus is provided.

  After measuring the three-dimensional image of the object and obtaining its height image information, the height image information is smoothed. Next, difference image information between the height image information and the smoothed height image information is obtained, and this is first-order differentiated. The area of the region in which the value of the obtained differential image primary differential information is equal to or greater than the first reference value is determined, and the type of the object is determined by comparing with the first determination threshold value. The difference image information between the height image information and the smoothed height image information becomes unevenness information of the object, and is information corresponding to the size of the object. By first differentiating this information, a portion having a large change in size can be extracted, and by obtaining the area, information regarding the size of the object can be obtained for the entire image area, and this is compared with a determination threshold value. . With such a discrimination process, whether or not the object is only a very small size can be determined without depending on the mounting state even if the object is mounted irregularly. It is possible to reliably determine without depending on the wet and dry state of the object.

  The second primary differential processing means for calculating the height image primary differential information obtained by primary differentiation of the height image information, and the value of the height image primary differential information obtained from the second primary differential processing means are: The type of the object is discriminated by comparing the area obtained from the second area processing means for obtaining the area of the region that is equal to or greater than the second reference value and the plurality of second discrimination thresholds. It is preferable to further include a second discriminating means. Subjecting the height image information to first differentiation, obtaining the area of the region where the obtained height image primary differentiation information value is greater than or equal to the second reference value, and comparing the area with a plurality of second discrimination thresholds Determine the type of object. By first differentiating the height information, it is possible to extract a portion having a large change in size, and by obtaining the area, information on the size of the object is obtained for the entire image area, and this is compared with a plurality of discrimination thresholds. To determine. By such discrimination processing, the type related to the size of the object is not dependent on the mounting state, even if the object is mounted irregularly, and also depends on the dry and wet state of the object. It is possible to discriminate reliably.

  Means for discriminating whether or not the concrete aggregate is sand by comparing the area obtained by the first discriminating means with the first discriminating threshold with the first discriminating threshold. It is also preferable.

  The object is concrete aggregate, and the second discrimination means compares the area obtained from the second area processing means with a plurality of second discrimination thresholds, and the concrete aggregate is classified according to size. It is also preferable to be means for discriminating whether the gravel is of a kind.

  It is also preferable that the height image creating means is means for obtaining height image information of an object mounted on the transport means or the transport belt by three-dimensional image measurement.

  It is also preferable that the height image creating means is means for obtaining height image information of the object by three-dimensional image measurement using a stereo method.

  According to the present invention, whether or not the object is only of a very fine size (for example, sand) (for example, gravel larger than sand) in a state where the object is irregularly mounted. Even if it exists, it can discriminate | determine reliably, without depending on a mounting state, and not depending on the dry / wet state of a target object.

It is a figure which shows roughly the structure of the acceptance equipment of the concrete aggregate in the concrete manufacturing factory as one Embodiment of this invention. It is a figure explaining the distance and measurable range of the camera and truck bed bottom in the embodiment of FIG. It is a block diagram which shows roughly the computer structure of the concrete aggregate type discrimination | determination apparatus in embodiment of FIG. It is a block diagram which shows roughly the structure of the kind discrimination device constructed | assembled in embodiment of FIG. It is a flowchart explaining operation | movement of the computer of the kind discrimination device of embodiment of FIG. It is a flowchart explaining the detailed operation | movement of the production process of the height image in FIG. It is the photograph showing (A) left camera image before parallelization correction processing, and (B) right camera image. It is a photograph showing (A) left camera image and (B) right camera image after parallelization correction processing. It is the photograph showing (A) the image after the left camera parallel correction process about the crushed stone 8040, and (B) the height image. It is a photograph showing the smoothing processing image of (A) height image and (B) height image in case concrete aggregate is crushed sand. It is a graph showing the pixel value (pixel value) characteristic of the height image for one line (X axis) of the height image when the concrete aggregate is crushed sand, and the smoothed image. It is a photograph showing the smoothing processing image of (A) height image and (B) height image in case concrete aggregate is crushed stone 2005. It is a graph showing the pixel value characteristic of the height image of 1 line (X-axis) of the height image in case a concrete aggregate is the crushed stone 2005, and its smoothing process image. It is a photograph showing the smoothing process image of (A) height image and (B) height image in case a concrete aggregate is the crushed stone 4020. It is a graph showing the pixel value characteristic of the height image of one line (X axis) of the height image when the concrete aggregate is crushed stone 4020 and the smoothed image. It is a photograph showing the smoothing processing image of (A) height image and (B) height image in case concrete aggregate is crushed stone 8040. It is a graph showing the pixel value characteristic of the height image of 1 line (X axis) of the height image in case a concrete aggregate is the crushed stone 8040, and its smoothing process image. It is a photograph showing the difference image of the height image in case a concrete aggregate is crushed sand, and its smoothing process image. It is a graph showing the pixel value characteristic of 1 line (X-axis) of the difference image of the height image in case a concrete aggregate is crushed sand, and its smoothing process image. It is a photograph showing the difference image of the height image in case a concrete aggregate is the crushed stone 2005, and its smoothing process image. It is a graph showing the pixel value characteristic of 1 line (X-axis) of the difference image of the height image in case a concrete aggregate is the crushed stone 2005, and its smoothing process image. It is a photograph showing the difference image of the height image in case a concrete aggregate is the crushed stone 4020, and the smoothing process image. It is a graph showing the pixel value characteristic of 1 line (X-axis) of the difference image of the height image in case a concrete aggregate is the crushed stone 4020, and its smoothing process image. It is a photograph showing the difference image of the height image in case a concrete aggregate is the crushed stone 8040, and its smoothing process image. It is a graph showing the pixel value characteristic of 1 line (X-axis) of the difference image of the height image in case a concrete aggregate is the crushed stone 8040, and its smoothing process image. It is a photograph of the image showing the absolute value of the primary differential along the horizontal direction of the difference image in case a concrete aggregate is crushed sand. It is a graph showing the pixel value characteristic of 1 line (X-axis) of the image showing the absolute value of the primary differential along the horizontal direction of a difference image in case a concrete aggregate is crushed sand. It is a photograph of the image showing the absolute value of the primary differential along the horizontal direction of a difference image in case a concrete aggregate is crushed stone 2005. It is a graph showing the pixel value characteristic of 1 line (X-axis) of the image showing the absolute value of the primary differential along the horizontal direction of a difference image in case the concrete aggregate is crushed stone 2005. It is the photograph of the image showing the absolute value of the 1st derivative along the horizontal direction of a difference image in case a concrete aggregate is the crushed stone 4020. It is a graph showing the pixel value characteristic of 1 line (X-axis) of the image showing the absolute value of the primary differential along the horizontal direction of a difference image in case a concrete aggregate is the crushed stone 4020. It is the photograph of the image showing the absolute value of the 1st derivative along the horizontal direction of a difference image in case a concrete aggregate is the crushed stone 8040. It is a graph showing the pixel value characteristic of 1 line (X-axis) of the image showing the absolute value of the primary differential along the horizontal direction of a difference image in case the concrete aggregate is the crushed stone 8040. It is the photograph of the image showing the area | region where the absolute value of the 1st differential process result of a difference image is 0.15 or more, (A) is concrete aggregate crushed sand, (B) is concrete aggregate crushed stone 2005, (C). Is the case where the concrete aggregate is crushed stone 4020, and (D) is the case where the concrete aggregate is crushed stone 8040. It is the photograph of the image showing the 1st differential processing result of a height image, (A) is concrete aggregate crushed stone 2005, (B) is concrete aggregate crushed stone 4020, (C) is concrete aggregate crushed stone 8040. Is the case. It is the photograph of the image showing the area | region where the absolute value of the 1st derivative process result of a height image is 2.0 or more, (A) is the concrete aggregate 2005, (B) is the concrete aggregate 4020, (A) C) is the case where the concrete aggregate is crushed stone 8040. It is a graph showing the type discrimination | determination situation of the concrete aggregate in one Example of this invention.

  FIG. 1 schematically shows the entire construction of a concrete aggregate receiving facility in a concrete manufacturing factory as one embodiment of the present invention, and FIG. 2 shows the distance between the camera and the bottom of a truck bed in this embodiment and measurement is possible. FIG. 3 schematically shows a computer configuration of the concrete aggregate type discriminating apparatus in the present embodiment, and FIG. 4 shows the concrete aggregate type discriminating apparatus constructed in the present embodiment. The configuration is schematically shown, and FIG. 5 illustrates the operation of the computer of the type discrimination device of the present embodiment.

  1 and 2, 10 is a truck loaded with a concrete aggregate 11 on a carrier, 12 is a hopper that receives the concrete aggregate 11 from the truck 10, 13 is a belt conveyor that conveys the concrete aggregate from the hopper 12, and 14 is Two cameras for imaging the concrete aggregate 11 loaded on the truck 10 in the loaded state, 15 is a lighting device for the concrete aggregate 11, 16 is electrically connected to the two cameras 14, The concrete aggregate type discriminating device 17 to which image information is inputted from these cameras 14 is electrically connected to the type discriminating device 16 and accepts the concrete aggregate which receives the aggregate discriminating result from the type discriminating device 16. The equipment control panels are shown respectively.

Each of the two cameras 14 is the same type of digital image pickup camera, and is a camera that uses a semiconductor image pickup device such as a CCD or a CMOS as the image pickup device. These two cameras 14 are installed spaced apart from one another by a predetermined distance, and upwardly spaced from the bed bottom 10a of the track 10 by a distance H _C. In the present embodiment, three-dimensional image measurement is performed by the stereo method based on image information from these two cameras 14. Incidentally, in FIG. 2, H _R represents the measurement distance range. As an example only, the distance H _C is H _C = 5500 mm, the distance range H _R is H _R = 1000 mm, and the distance between the two cameras 14 is 300 mm.

  In the present embodiment, the concrete aggregate 11 loaded on the loading platform of the truck 10 is imaged by the two cameras 14 and the type is discriminated. However, in one modification, as shown in FIG. The concrete aggregate transported by 13 may be picked up by two cameras 14 'to perform type discrimination.

  As shown in FIG. 3, the type discriminating apparatus 16 in this embodiment includes a central processing unit (CPU) 30b, a read only memory (ROM) 30c, and a random access memory (RAM) connected to each other via a bus 30a. 30d, a hard disk drive (HDD) 30e, a communication interface 30f, an audio processor 30g, an image processor 30h, a disk drive 30i such as a Blu-ray disc / digital versatile disc (BR / DVD), and input / output The computer 30 includes an interface 30j and a program for operating the computer 30.

  The audio processing unit 30g is connected to the speaker 30k, and the image processing unit 30h is connected to the display 30l. The disc drive device 30i can be mounted with a Blu-ray disc / digital versatile disc / compact disc (BR / DVD / CD) 30m. The input / output interface 30j has a mouse 30n, a keyboard 30o, two cameras 14, and concrete. The control panel 17 of the aggregate receiving facility is connected.

  The CPU 30b is configured to execute a program stored in the ROM 30c or the HDD 30e according to a basic program such as an operation system (OS) or a boot program stored in the ROM 30c or the HDD 30e to perform the processing of this embodiment. . The CPU 30b is configured to control operations of the RAM 30d, the HDD 30e, the communication interface 30f, the sound processing unit 30g, the image processing unit 30h, the disk drive device 30i, and the input / output interface 30j.

  The RAM 30d is used as a main memory of the type discriminating device, and is configured to store programs and data transferred from the HDD 30e and the disk drive device 30i. The RAM 30d is also used as a work area for temporarily storing various data during program execution.

  The HDD 30e is configured to store programs and data in advance, or to store programs and data captured from an external network via the communication interface 30f.

  The audio processing unit 30g is configured to perform processing for reproducing audio data in accordance with an instruction from the CPU 30b and output the audio signal to the speaker 30k.

  The image processing unit 30h is configured to perform image processing in accordance with an instruction from the CPU 30b and generate image data. The generated image data is output to the display 301 and displayed on the screen.

  The disk drive device 30i is configured to read a program or data from the set BR / DVD / CD 30m and transfer it to the RAM 30d in accordance with an instruction from the CPU 30b. It is also possible to write programs and data to the set BR / DVD / CD 30m.

  The input / output interface 30j controls the exchange of data between the mouse 30n and the keyboard 30o and the CPU 30b or the RAM 30d, inputs the imaging information from the two cameras 14, and further outputs the determination result to the concrete aggregate receiving facility. It is configured to output to the control panel 17.

  The communication interface 30f is connected to a network (not shown), and is configured to be able to exchange programs and data with this network according to instructions from the CPU 30b.

  In the computer 30 having such a configuration, when starting the program, the CPU 30b first secures a program storage area, a data storage area, and a work area in the RAM 30d, and fetches the program and data from the HDD 30e or from the outside to obtain the program storage area. And stored in the data storage area. Next, the processing flow shown in FIG. 5 is executed based on the program stored in the program storage area. As the CPU 30b executes the program, a type discriminating device 16 as schematically shown in FIG. 4 is constructed in this computer.

  As shown in FIG. 4, the type discriminating device 16 constructed on the computer is a height image creation that obtains the height image information of the concrete aggregate by the three-dimensional image measurement of the stereo method based on the imaging information of the two cameras 14. Means 40, smoothing processing means 41 for obtaining smoothed height image information obtained by smoothing height image information obtained from height image creating means 40, and smoothing obtained from height image information and smoothing processing means 41. Difference processing means 42 for obtaining difference image information of the converted height image information, and first primary differentiation processing means 43 for calculating difference image primary differential information obtained by firstly differentiating difference image information obtained from the difference processing means 42. A first area processing unit 44 for obtaining an area of a region in which the value of the differential image primary differential information obtained from the first primary differential processing unit 43 is equal to or greater than a first reference value; and a first area processing Means 44 A first discriminating means 45 for discriminating the type of the object by comparing the obtained area with a first discriminating threshold, and a second for calculating the height image primary differential information obtained by linearly differentiating the height image information. First differential processing means 46 and second area processing means for obtaining the area of the area where the value of the height image primary differential information obtained from the second primary differential processing means 46 is equal to or greater than the second reference value. 47 and second discrimination means 48 for discriminating the type of the object by comparing the area obtained from the second area processing means 47 with a plurality of second discrimination thresholds.

Before describing the processing operation of the type discriminating apparatus 16, the calibration processing (camera calibration) of the two cameras 14 will be briefly described. This calibration process is to obtain the camera parameters of each camera 14 and the relative positional relationship between the two cameras 14 when the type discriminating apparatus is shipped from the factory, when the camera is installed, or when necessary.
(1) First, the two cameras 14 are arranged at a predetermined interval.
(2) Next, the camera calibration plate (black circles of uniform size on a white background are arranged in a matrix of 7 columns and 7 rows (49 pieces) at regular intervals. The image of the plate) drawn by the plate area which is a square frame is taken by the two cameras 14 and stored in a predetermined storage area such as the RAM 30d or the HDD 30e.
(3) The plate area of the camera calibration plate image stored in this storage area is read out.
(4) The position and orientation of the camera calibration plate are estimated from the black circles in the read plate area, and temporarily stored as black circle position information (WMp) in the work storage area.
(5) The camera calibration plate is moved, and the above processes (2) to (4) are repeated 45 times.
(6) Based on the black circle position information (WMp) of the calibration plate obtained from the 45 images thus obtained, the camera parameters of the two cameras 14 and the relative positions of the two cameras 14 The relationship is calculated and stored in the storage area in a usable state.

Further, if necessary, create by measuring the height H _C of the camera 14 to measure the height of the aggregate to loading floor 10a of the "reference image".

The “reference image” is created as follows.
(A) A surface that is a reference in the depth direction (height direction) (in this embodiment, the cargo bed floor surface 10a of the truck 10) is photographed by the two cameras 14 and stored in a storage area.
(B) As will be described later with reference to FIG. 6, the image information is read out (the process of step S2a in FIG. 6 is performed).
(C) As will be described later, a parallelization correction process is performed on the captured image information (the process of step S2b in FIG. 6 is performed).
(D) As described later, a parallax image creation process is performed on the image information that has undergone the parallelization correction process (the processes of steps S2c and S2d in FIG. 6 are performed).
(E) As will be described later, a depth direction image creation process is performed (the processes in steps S2e and S2f in FIG. 6 are performed), and the created depth direction image is used as reference image information in a storage area in a usable state. save.

  Next, the type determination processing operation in the present embodiment will be described with reference to the processing flow of FIG.

  When a start signal, which is an external input signal composed of a signal from a sensor (not shown) for detecting that the truck 10 has entered the concrete aggregate receiving unit, a signal by an operator's button operation, or the like, is applied. The determination device 16 starts the process shown in FIG.

  First, image information obtained by imaging the concrete aggregate 11 on the loading platform with the two left and right cameras 14 provided above the loading platform of the truck 10 is captured, and predetermined information such as RAM 30d or HDD 30e is obtained for each of the left and right image information. Are stored in the work storage area (step S1).

  Next, a height image of the concrete aggregate 11 is created by the stereo method based on the image information from the left and right cameras 14, and the created height image information is stored in a predetermined work storage area (step S2).

  FIG. 6 illustrates a detailed processing flow of the height image creation processing.

  In the height image creation processing, first, image information (matrix pixel information) captured by the left and right cameras 14 stored in a predetermined work storage area is read (step S2a), and the parallelization correction processing is performed on the image information. Is stored in a predetermined work storage area (step S2b). This parallelization correction process uses the information on the relative positional relationship between the two cameras 14 obtained by the calibration process (camera calibration) of the two cameras 14 and the camera parameters, and reads the left camera read from the storage area. Using the image as a reference, the right camera image is corrected so as to be parallel to the image, and is stored as WMh in the work storage area (hereinafter, this image information is referred to as right camera image information). Note that the left camera image may be corrected so as to be parallel to the right camera image as a reference, and stored in the work storage area as WMh.

  FIG. 7 shows (A) the left camera image and (B) the right camera image before the parallelization correction processing, and FIG. 8 shows (A) the left camera image and (B) the right after the parallelization correction processing. A camera image is shown.

  In these drawings, when attention is paid to the aggregate surrounded by a broken line, before the parallelization correction processing in FIG. 7, the images 70 and 71 of the aggregate corresponding to (A) the left camera image and (B) the right camera image. Are not located on the same straight line, but after the parallelization correction process of FIG. 8, the corresponding aggregate images 80 and 81 are located on the same straight line in the left camera image and (B) right camera image. You can see that

  Next, the post-parallelization corrected image information stored as WMh in the work storage area is read (step S2c), a parallax image creation process is performed on the image information, and the image information is stored in a predetermined work storage area (step S2d). ). In this parallax image creation processing, in order to calculate a three-dimensional position, first, it is specified which point on the two left and right images is a certain point in space. That is, using the left camera image as a reference, template matching (normalized cross correlation function) is used for a certain point in the image, and a corresponding point is searched from the right camera image (WMh) after the parallelization correction processing. . Next, the parallax of the corresponding point when viewed in the left camera image and the right camera image is calculated from the shift in the horizontal direction. Thereafter, the obtained parallax image information is stored as WMs in a predetermined work storage area. In this embodiment, parallelization is performed to improve the efficiency of template matching. However, parallax can be calculated by searching corresponding points between the left camera image and the right camera image without performing parallelization. good.

  Next, the parallax image information stored as WMs in the work storage area is read (step S2e), a depth direction image is created from the image information, and stored in a predetermined work storage area (step S2f). In this depth direction image creation processing, the distance in the depth direction is obtained by calculation based on the relative positional relationship between the parallax images (WMs) and the two cameras 14 and camera parameters, and an image in the depth direction is created.

  Thereafter, the depth direction image information stored in the work storage area is read (step S2g), a height image creation process is performed from the image information, and is stored as WMt in a predetermined work storage area (step S2h). This height image creation processing obtains height image information by reading the reference image information created in advance and stored in the storage area, and subtracting the depth direction image information from the reference image information for each pixel. To save.

  FIG. 9 shows (A) an image after the left camera parallel correction processing and (B) a height image for crushed stone 8040 (JIS A 5005). In FIG. 9A, a frame line 90 indicates that the inside is an image processing area, and FIG. 10B indicates a height image of only the processing area. However, in FIG. 5B, the height image is represented by black and white contrast, and the white portion represents a higher position and the black portion represents a lower position. .

  Hereinafter, the description will be returned to the processing flow of FIG. After the height image creation processing in step S2, the height image information (WMt) stored in a predetermined work storage area is read out, smoothed, and smoothed image information (WMsm) is used for the work. Save in the storage area (step S3). This smoothing process is a smoothing filter process for making a smooth image in order to remove random noise on the image or reduce fine changes in pixel values and obtain an easy-to-view image. In the embodiment, the following averaging filter is used.

Here, h (i, j) is the pixel value in the i-th row and j-th column of the averaged image, g (i, j) is the pixel value in the i-th row and j-th column in the height image, and N × N is the filter Each size is shown.

For example, when N = 3, the pixel value h (i, j) of the averaged image is expressed as follows, and becomes an averaging filter having a size of 3 × 3 pixels.

  As the smoothing filter, in addition to the averaging filter, a median filter, a weighted averaging filter, a Gaussian filter, a k nearest neighbor averaging filter, a bilateral filter, or the like can be used.

  Since the unevenness information (high-frequency component) of the aggregate is removed by this smoothing process, the smoothed image indicates how the aggregate is piled up.

  FIG. 10 shows a smoothed image of (A) a height image and (B) a height image when the concrete aggregate is crushed sand. Moreover, FIG. 11 represents the pixel value characteristic of the height image of one line (X axis) 100 of the height image when the concrete aggregate is crushed sand, and the smoothed image.

  FIG. 12 shows (A) a height image and (B) a smoothed image of the height image when the concrete aggregate is crushed stone 2005, respectively. FIG. 13 shows the height image of one row (X axis) 120 of the height image when the concrete aggregate is crushed stone 2005, and the pixel value characteristic of the smoothed image.

  FIG. 14 shows (A) a height image and (B) a smoothed image of the height image when the concrete aggregate is crushed stone 4020, respectively. FIG. 15 shows the height image of one row (X axis) 140 of the height image when the concrete aggregate is crushed stone 4020 and the pixel value characteristic of the smoothed image.

  FIG. 16 shows (A) a height image and (B) a smoothed image of the height image when the concrete aggregate is crushed stone 8040, respectively. FIG. 17 represents the height image of one row (X axis) 160 of the height image when the concrete aggregate is crushed stone 8040 and the pixel value characteristic of the smoothed image.

  After the smoothing process in step S3 of FIG. 5, the smoothed image information (WMsm) stored in a predetermined work storage area is read out, and the height image information (WMt) and the smoothed image information (WMsm) are read out. (WMd = WMt−WMsm), and the difference image information (WMd) is stored in the work storage area (step S4).

  FIG. 18 shows a difference image between a height image when the concrete aggregate is crushed sand and a smoothed image thereof, and FIG. 19 shows a height image when the concrete aggregate is crushed sand and a smoothed process thereof. The pixel value characteristic of one line (X axis) 180 of the difference image with the image is represented.

  FIG. 20 shows a difference image between a height image when the concrete aggregate is crushed stone 2005 and a smoothed image thereof, and FIG. 21 shows a height image when the concrete aggregate is crushed stone 2005 and its smoothed image. The pixel value characteristic of one line (X axis) 200 of the difference image with the digitized image is represented.

  22 shows a difference image between a height image when the concrete aggregate is crushed stone 4020 and a smoothed image thereof, and FIG. 23 shows a height image when the concrete aggregate is crushed stone 4020 and its smoothed image. The pixel value characteristic of one line (X axis) 220 of the difference image with the digitized image is represented.

  FIG. 24 shows a difference image between a height image when the concrete aggregate is crushed stone 8040 and a smoothed image thereof, and FIG. 25 shows a height image when the concrete aggregate is crushed stone 8040 and its smoothed image. The pixel value characteristic of one line (X axis) 240 of the difference image with the digitized image is represented.

  As can be seen from these figures, the difference between the height image and the smoothing correction processing image, which is the unevenness information of the aggregate, is rare when the concrete aggregate is crushed sand, but when the concrete aggregate is crushed stone The difference is larger than in the case of crushed sand. In particular, the larger the aggregate diameter, the greater the difference.

  After the difference processing in step S4 in FIG. 5, the difference image information (WMd) stored in a predetermined work storage area is read, and the difference image information (WMd) is set to 1 along the horizontal direction (X-axis direction). Next differentiation and primary differentiation along the vertical direction (Y-axis direction), and the absolute values (WMbsx) and (WMbsy) of the values obtained by the first differentiation are stored in the work storage area (step S5). . These primary differentials are for extracting a portion where the difference is drastically changed. When the concrete aggregate is crushed sand, the value is lower than that of crushed stone. Note that, before performing the primary differentiation, smoothing processing may be performed on the difference image information (WMd), and the primary differentiation may be performed on the smoothed difference image information.

  FIG. 26 shows an image of the absolute value of the first derivative along the horizontal direction of the difference image when the concrete aggregate is crushed sand, and FIG. 27 shows the horizontal direction of the difference image when the concrete aggregate is crushed sand. 2 represents the pixel value characteristic of one line (X axis) 260 of the image representing the absolute value of the first-order differential along the line.

  FIG. 28 shows an image of the absolute value of the first derivative along the horizontal direction of the difference image when the concrete aggregate is crushed stone 2005, and FIG. 29 shows the difference image when the concrete aggregate is crushed stone 2005. The pixel value characteristic of one line (X axis) 280 of the image representing the absolute value of the first-order differential along the horizontal direction is shown.

  30 shows an image of the absolute value of the first derivative along the horizontal direction of the difference image when the concrete aggregate is crushed stone 4020, and FIG. 31 shows the difference image when the concrete aggregate is crushed stone 4020. The pixel value characteristic of one line (X axis) 300 of the image representing the absolute value of the first-order differential along the horizontal direction is shown.

  FIG. 32 shows an image of the absolute value of the first derivative along the horizontal direction of the difference image when the concrete aggregate is crushed stone 8040, and FIG. 33 shows the difference image when the concrete aggregate is crushed stone 8040. The pixel value characteristic of one line (X axis) 320 of the image representing the absolute value of the primary differential along the horizontal direction is represented.

  26 to 33 show the results of the primary differentiation along the horizontal direction (X-axis direction) for the difference image information (WMd), the difference image information is used in the primary differentiation process in step S5. (WMd) is first-order differentiated along the vertical direction (Y-axis direction), and the first-order derivative information obtained from both is used here.

  Next, in order to evaluate the entire image, an area where the absolute value of the first derivative of the difference image takes a value of 0.15 or more is calculated, and the area is stored in the work storage area as a first index (WMm1). Save (step S6). That is, the area where the absolute value (WMbsx) obtained by first-order differentiation along the horizontal direction (X-axis direction) of the difference image information takes a value of 0.15 or more, and the vertical direction (Y The total area is obtained by adding the area where the absolute value (WMbsy) of the value obtained by first-order differentiation along the axial direction takes a value of 0.15 or more. In this case, the area calculation process in the X-axis direction and the Y-axis direction actually calculates the number of pixels having an absolute value of the first-order differential of the difference image of 0.15 or more.

  FIG. 34 shows an image in which the area where the absolute value of the first-order differential processing result of the difference image is 0.15 or more is white. FIG. 34 (A) shows the concrete aggregate being crushed sand, and FIG. The aggregate is crushed stone 2005, the same figure (C) is the case where the concrete aggregate is crushed stone 4020, and the same figure (D) is the case where the concrete aggregate is crushed stone 8040. From the figure, it can be seen that for crushed sand, there is almost no region where the absolute value of the first-order differential processing result of the difference image is 0.15 or more. Thus, the first index (WMm1) is effective when separating an aggregate having a small diameter (5 mm or less).

  Thereafter, the height image information (WMt) stored in a predetermined work storage area is read out, and the height image information (WMt) is first-order differentiated along the horizontal direction (X-axis direction) and the vertical direction ( A first-order differentiation is performed along the Y-axis direction), and absolute values (WMbtx) and (WMbty) of values obtained by the first-order differentiation are stored in the work storage area (step S7). These primary differentiations are performed because it is considered that the influence of how the aggregate is piled up is small in a portion where the change in height is severe. Before performing the first differentiation, the smoothing process may be performed on the height image information (WMt), and the first differentiation may be performed on the smoothed height image information.

  FIG. 35 is an image showing the first-order differential processing result of the height image. FIG. 35A shows a concrete aggregate crushed stone 2005, FIG. 35B shows a concrete aggregate crushed stone 4020, and FIG. This is the case where the concrete aggregate is crushed stone 8040.

  Next, in order to evaluate the entire image, an area where the absolute value of the first derivative of the height image takes a value of 2.0 or more is calculated, and the area is used as a second index (WMm2) as a work storage area. (Step S8). That is, the area where the absolute value (WMbtx) of the value obtained by first-order differentiation along the horizontal direction (X-axis direction) of the height image information takes a value of 2.0 or more, and the vertical direction of the height image information. The total area is obtained by adding the area where the absolute value (WMbty) of the value obtained by first-order differentiation along (Y-axis direction) takes a value of 2.0 or more. In this case, the calculation process of the area in the X-axis direction and the Y-axis direction is actually to determine the number of pixels having an absolute value of the first derivative of the height image of 2.0 or more.

  FIG. 36 shows an image in which an area where the absolute value of the first-order differential processing result of the height image is 2.0 or more is white, and FIG. 36 (A) shows the concrete aggregate is crushed stone 2005 and FIG. 36 (B). Is a case where the concrete aggregate is crushed stone 4020, and FIG. From the figure, it can be seen that the larger the crushed stone diameter, the larger the area where the absolute value of the first-order differential processing result of the height image is 2.0 or more. As described above, the second index (WMm2) is effective when a crushed stone which is an aggregate having a large diameter (greater than 5 mm) is divided according to its diameter.

  After the area processing in step S8 of FIG. 5, the type of the concrete aggregate is determined. First, the first index (WMm1) stored in the work storage area is read out, and it is determined whether or not the first index is less than the first determination threshold (step S9 in FIG. 5). In this example, 61000 is used as the first discrimination threshold (when the image processing range is 650 × 800 pixels). Therefore, in step S9, it is determined whether WMm1 <61000.

  When WMm1 <61000 (in the case of YES), it is determined that the concrete aggregate is crushed sand (step S10 in FIG. 5). When WMm1 <61000 is not satisfied (in the case of NO), it is determined that the concrete aggregate is crushed stone (step S11 in FIG. 5).

  Next, the second index (WMm2) stored in the work storage area is read, and the second index is less than the second determination threshold A, which is one of the plurality of second determination thresholds. (Step S12 in FIG. 5). In this example, 1650 is used as the second determination threshold A. Therefore, in step S12, it is determined whether WMm2 <1650.

  When WMm2 <1650 (YES), it is determined that the concrete aggregate is crushed stone 2005 (step S13 in FIG. 5). When WMm2 <1650 is not satisfied (in the case of NO), it is determined whether or not the second index is less than a second determination threshold B which is another one of the plurality of second determination thresholds ( Step S14 in FIG. In this example, 24000 is used as the second determination threshold B (when the image processing range is 650 × 800 pixels (pixels)). Accordingly, in step S12, it is determined whether WMm2 <24000.

  When WMm2 <24000 (in the case of YES), it is determined that the concrete aggregate is crushed stone 4020 (step S15 in FIG. 5). When WMm2 <24000 is not satisfied (in the case of NO), it is determined that the concrete aggregate is crushed stone 8040 (step S16 in FIG. 5).

  According to the above determination processing, when the first index WMm1 is WMm1 <61000, it is determined that the concrete aggregate is crushed sand, and when the second index WMm2 is WMm2 <1650, the concrete aggregate is crushed stone 2005. When the second index WMm2 is 1650 ≦ WMm2 <24000, it is determined that the concrete aggregate is crushed stone 4020, and when the second index WMm2 is 24000 ≦ WMm2, the concrete aggregate is crushed stone 8040. Determined.

  In step S10, S13, S15, or S16 of FIG. 5 described above, after the type of the concrete aggregate is determined, the result is output to the concrete aggregate receiving equipment control panel 17 (see FIG. 1) (see FIG. 1). 5 step S17).

  As described in detail above, according to the present embodiment, the concrete aggregate 11 to be identified is picked up by the two cameras 14 provided above the loading platform while being loaded on the loading platform of the truck 10. Then, a three-dimensional image is measured by the stereo method. Height image information is obtained from the imaging information, difference image information between the height image information and the smoothed height image information is obtained, and this is first-order differentiated. The area of the region where the value of the obtained differential image primary differential information is greater than or equal to the first reference value is obtained and used as the first index. On the other hand, the height information is first-order differentiated, and the area of the region where the value of the obtained height image first-order derivative information is equal to or larger than the second reference value is obtained and used as the second index. By comparing the first index with the first determination threshold, it is determined whether the concrete aggregate 11 is crushed sand or crushed stone. The difference image information between the height image information and the smoothed height image information is unevenness information of the concrete aggregate 11 and is information corresponding to the size of the concrete aggregate 11. A portion with a large change in size can be extracted by first-order differentiation of this information, and information regarding the size of the concrete aggregate 11 can be obtained for the entire image by obtaining the area, and this can be compared with a discrimination threshold. Whether or not the concrete aggregate 11 is only a crushed sand having a very fine size (whether it is a crushed stone larger than the crushed sand) or not is determined in a state where the concrete aggregate 11 is irregularly mounted. Even if it exists, it can discriminate | determine reliably, without depending on a mounting state, and not depending on the dry and wet state of the concrete aggregate 11. FIG.

  Moreover, according to this embodiment, it is discriminate | determined whether the concrete aggregate 11 is the crushed stone 2005, the crushed stone 4020, and the crushed stone 8040 by comparing a 2nd parameter | index with several 2nd discrimination | determination threshold value. . The height image information is first-order differentiated, and the area of the region where the value of the obtained height image first-order derivative information is equal to or greater than the second reference value is obtained and compared with a plurality of second discrimination thresholds. It is determined whether the aggregate 11 is crushed stone 2005, crushed stone 4020, or crushed stone 8040. By first differentiating the height information, it is possible to extract a portion having a large change in size, and by obtaining the area, information on the size of the object is obtained for the entire image area, and this is compared with a plurality of discrimination thresholds. Depending on the type of discrimination processing, the type (crushed stone 2005, crushed stone 4020, crushed stone 8040) related to the size of the concrete aggregate 11 depends on the loaded state even if the concrete aggregate 11 is loaded irregularly. It is possible to reliably determine without depending on the wet and dry state of the concrete aggregate 11.

  Note that, as the three-dimensional image measurement method, the stereo method using parallax is used in the above-described embodiment, but various other known methods such as a light cutting method for projecting slit light may be applied. Is possible.

  As concrete aggregates, four types of crushed sand, crushed stone 2005 (5-20 mm diameter crushed stone), crushed stone 4020 (20-40 mm diameter crushed stone), and crushed stone 8040 (40-80 mm diameter crushed stone) specified in JIS A 5005. Prepared. A total of 68 types of samples were prepared for each of these aggregates, depending on how they were wet (dry, partially wet, all wet), and in 17 different ways. These samples are summarized as shown in Table 1 below.

  Two digital image capturing cameras were arranged in parallel with each other above such a sample and imaged. In this example, the distance between the camera and the sample was 1600 mm, and the camera interval (distance between optical axes) was 100 mm. As a digital image pickup camera, a 2 million pixel CCD monochrome camera manufactured by Hitachi Kokusai Electric Inc. (1628 pixels (horizontal) × 1236 pixels (vertical), 1 / 1.8 type CCD, pixel size 4.4 μm × 4.4 μm) ), KP-F202GV, and a lens, HS1614J manufactured by Mutetron Co., Ltd., and a focal length of 16 mm were used. Three-dimensional image measurement was performed from the imaging data by the stereo method, and the first index WMm1 and the second index WMm2 were obtained for each sample as in the above-described embodiment.

  Table 2 shows the obtained first index WMm1 in ascending order for each sample (type). Assuming that the image processing range is 650 × 800 pixels (pixels), the change in the value is large in the portion indicated by the broken line in Table 2. If this is set as the first discrimination threshold (= 61000), crushed sand and crushed stone And was able to be identified with certainty.

  Table 3 shows the obtained second index WMm2 in ascending order for each sample (type). If the processing range of the image is 650 × 800 pixels (pixels), the change in the value is large at the two portions indicated by the broken lines in Table 3, so that one of the lower ones is set as the second discrimination threshold A (= 1650), the crushed stone 2005 and the crushed stone 4020 can be reliably identified, and if the upper one is the second discrimination threshold B (= 24000), the crushed stone 4020 and the crushed stone 8040 can be reliably identified. Could be identified.

  FIG. 37 shows the concrete aggregate type discrimination status in this embodiment, where the horizontal axis is the first index WMm1, that is, the first derivative of the difference between the height image smoothed and corrected and the original height image. Is the area of a region having an absolute value of 0.15 or more, and the vertical axis is the area of the second index WMm2, that is, the region having an absolute value of the first derivative of the height image of 2.0 or more.

  It can be seen that the crushed sand and the crushed stone can be reliably identified by using the first determination threshold (= 61000) for the first index WMm1. Further, by using the second discrimination threshold A (= 1650) for the second index WMm2, the crushed stone 2005 and the crushed stone 4020 can be reliably identified, and further, the second discrimination threshold B (= 24000) is used. By this, it turns out that the crushed stone 4020 and the crushed stone 8040 can be identified reliably.

  Although the above-described embodiments and examples relate to the types of concrete aggregates defined in JIS A 5005, in particular, the invention relates to a type discriminating apparatus for identifying crushed sand, crushed stone 2005, crushed stone 4020, and crushed stone 8040. Is applicable to a device for discriminating the types of other concrete aggregates defined in JIS A 5005, and is also applicable to other types of discriminating devices such as sand and gravel. Furthermore, the present invention can also be applied to discriminating types of various objects other than concrete aggregates that contain particles of different sizes.

  The above-described embodiments and examples are all illustrative and do not limit the present invention, and the present invention can be implemented in various other modifications and changes. Therefore, the scope of the present invention is defined only by the claims and their equivalents.

DESCRIPTION OF SYMBOLS 10 Truck 10a Bottom of loading platform 11 Concrete aggregate 12 Hopper 13 Belt conveyor 14, 14 'Camera 15 Illumination device 16 Concrete aggregate type discriminating device 17 Control panel of concrete aggregate receiving equipment 30 Computer 30a Bus 30b Central processing unit (CPU) )
30c Read only memory (ROM)
30d random access memory (RAM)
30e Hard disk drive (HDD)
30f Communication interface 30g Audio processing unit 30h Image processing unit 30i Disc drive device 30j Input / output interface 30k Speaker 30l Display 30m Blu-ray Disc / Digital Versatile Disc / Compact Disc (BR / DVD / CD)
30n mouse 30o keyboard 40 height image creation means 41 smoothing processing means 42 difference processing means 43 first primary differentiation processing means 44 first area processing means 45 first discrimination means 46 second primary differentiation processing means 47 Second area processing means 48 Second discrimination means 70, 71, 80, 81 Aggregate image 90 Frame line 100, 120, 140, 160 One row of height image (X axis)
180, 200, 220, 240 One line of differential image (X axis)
260, 280, 300, 320 One line of the image representing the absolute value of the first derivative (X axis)

Claims (6)

  Height image creating means for obtaining height image information of an object by three-dimensional image measurement, and smoothing processing means for obtaining smoothed height image information obtained by smoothing height image information obtained from the height image creating means Difference processing means for obtaining difference image information of the height image information and the smoothed height image information obtained from the smoothing processing means, and a difference image obtained by first differentiating the difference image information obtained from the difference processing means The first primary differential processing means for calculating the primary differential information, and the area of the region where the value of the differential image primary differential information obtained from the first primary differential processing means is greater than or equal to the first reference value. First area processing means to be obtained, and first determination means for determining the type of the object by comparing the area obtained from the first area processing means with a first determination threshold value. An object type discrimination device characterized by the above.   Second primary differential processing means for calculating height image primary differential information obtained by performing primary differentiation on the height image information, and height image primary differential information obtained from the second primary differential processing means. A second area processing unit for obtaining an area of a region having a value equal to or greater than a second reference value; and comparing the area obtained from the second area processing unit with a plurality of second discrimination thresholds. The object type discriminating apparatus according to claim 1, further comprising: a second discriminating unit that discriminates the type of the object.   Whether the object is concrete aggregate and whether the concrete aggregate is sand by comparing the area obtained from the first area processing means with the first discrimination threshold by the first discrimination means. The object type discrimination device according to claim 1, wherein the type discrimination unit is a means for discriminating between the two.   The object is concrete aggregate, and the second discriminating unit compares the area obtained from the second area processing unit with a plurality of second discriminating thresholds to classify the concrete aggregate according to size. 3. The object type discriminating apparatus according to claim 2, wherein the object type discriminating means is means for discriminating which kind of gravel is used.   5. The height image creating means is means for obtaining height image information of the object mounted on a conveying means or a conveying belt by three-dimensional image measurement. The object type discriminating device described in 1.   The object according to any one of claims 1 to 5, wherein the height image creating means is means for obtaining height image information of the object by three-dimensional image measurement using a stereo method. Item type identification device.