JP2014044078A - Body weight estimation device and body weight estimation method, for animal body - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a device which measures a dimension of an animal body by means of three-dimensional measurement with high accuracy so as to improve accuracy of a body weight estimated from obtained dimension data.SOLUTION: Random light points in an infrared band are projected from three-dimensional measuring instruments 14A and 14B from at least one of upper, left lateral and right lateral directions of a cow body 3 quiescent in a prescribed posture, and reflection points reflected by the cow body 3 are imaged so as to acquire three-dimensional point group data of an upper surface of the cow body 3. With respect to the cow body data, constituted by the three-dimensional group data acquired from the three-dimensional measuring instruments 14, a cross-sectional shape in parallel with a gravity direction and a cross-sectional shape in parallel with a horizontal direction are approximated to an elliptical shape; characteristic dimension data of the cow body 3 is extracted from the approximated elliptical shape; and a body weight of the cow body is estimated by substituting the extracted characteristic dimension data of the cow body 3 in a prescribed body weight formula.

Description

  The present invention relates to a weight estimation apparatus and weight estimation method for an animal body using three-dimensional measurement data.

2. Description of the Related Art Conventionally, as a means for measuring the weight of a cow body, a cow impactor that performs measurement by placing the cow body on a platform is known. However, there is a problem in that it places mental stress on the cow impactor and affects the quality of the meat. For this reason, the actual situation is that cow impactors are not used by livestock farmers as a means of measuring body weight.
For this reason, when measuring the weight of a cow during breeding, a skilled person estimates the weight visually, or a method of estimating the weight from the measured values of local parts of the body is adopted. .
However, the conventional weight estimation (evaluation) method has a problem in measurement accuracy and accuracy. Therefore, development of a measuring instrument that is portable and capable of accurately measuring (estimating) the weight of a cow without contact is eagerly desired.

Patent Document 1, livestock withers and (H) to the projected area (A), but with respect to the weight ^{(W), W = aH b} A are closely related as ^c, further dimensions of the multiple regression equation Since (d = b + 2c) is close to the volume dimension (d = 3), the body height and projected area of livestock are remotely measured using the multiple regression equation, and those values are substituted into the multiple regression equation. Thus, there has been reported a non-contact weight measurement method for livestock that can measure the weight of a large number of livestock in a non-contact manner without depending on the breed and with relatively high accuracy.
In this case, by using an image taken by a camera or a video camera, the projected area of the livestock is obtained, so that a human does not touch the livestock and is remotely measured. By measuring the height (one dimension) and area (two dimensions) of livestock remotely, the volume of livestock (three dimensions) is obtained, and the weight of the livestock is estimated assuming that the specific gravity of the livestock is constant. It is disclosed.

JP 2002-243527 A

In patent document 1, it was comprised so that the projection area of livestock might be measured using the image image | photographed with the camera, and it may substitute for the said multiple regression equation.
However, since the cited document 1 is two-dimensional measurement by image measurement, there is a problem that the accuracy of the estimated weight is not sufficient.
Therefore, a proposal for a weight estimation apparatus and a weight estimation method capable of measuring the dimension of an animal body with high accuracy by three-dimensional measurement and improving the accuracy of the weight estimated from the obtained dimension data is highly desired. Yes.
The present invention has been made in view of the above, and its purpose is to measure the dimensions of a moving object with high accuracy by three-dimensional measurement and to improve the accuracy of weight estimated from the obtained dimension data. It is an object of the present invention to provide a weight estimation apparatus and a weight estimation method that can perform the above-described measurement.

  In order to solve the above-mentioned problem, the invention according to claim 1 is a weight estimation apparatus for an animal body that estimates the weight of the animal body by substituting the dimension data of the animal body into a predetermined weight calculation formula, Random light spots in the infrared band are projected from at least one of the upper, left side, and right side of the moving object stationary in a posture, and the reflected points reflected by the moving object are imaged. A 3D measuring device that acquires 3D point cloud data and an elliptical cross-sectional shape parallel to the gravitational direction and a horizontal parallel shape to the 3D point cloud data acquired by the 3D measuring device. A shape approximating unit for approximating the shape, and a shape extracting unit for extracting characteristic dimension data of the moving object from the elliptical shape approximated by the shape approximating unit, and extracted by the shape extracting unit Of the animal body To estimate the weight of the animal body by substituting symptoms dimensions data to the predetermined weight calculation formula, it is weight estimation apparatus of the animal body characterized by.

  According to the present invention, a random light spot in the infrared band is projected from at least one of the upper, left side, and right side of the moving object stationary in a predetermined posture, and the reflected point reflected by the moving object is obtained. 3D point cloud data of the moving object is acquired by imaging, and the acquired 3D point cloud data is approximated by approximating the cross-sectional shape parallel to the gravity direction and the cross-sectional shape parallel to the horizontal direction to an elliptical shape. By extracting characteristic dimension data of the moving object from the elliptical shape, substituting the extracted characteristic dimension data of the moving object into a predetermined weight calculation formula, and estimating the weight of the moving object, The dimension of the moving object can be measured with high accuracy by three-dimensional measurement, and the accuracy of the weight estimated from the obtained dimension data can be improved.

It is a figure explaining the structure of the weight measurement system 1 applicable to the weight estimation apparatus of the moving body which concerns on 1st Embodiment of this invention. (A) is a figure which shows the external appearance of the three-dimensional measuring device shown in FIG. 1, (b) is a figure which shows the internal structure of the projector 16. FIG. It is a figure which shows the characteristic structure of the personal computer 22 for processing the data from the inclination sensor 12 shown in FIG. 1, and the three-dimensional measuring devices 14A and 14B. (A) (b) is a figure for demonstrating the method of calculating | requiring the distance to a to-be-photographed object with a three-dimensional measuring device. It is a figure for demonstrating the coordinate alignment of a weight estimation apparatus. It is a flowchart for demonstrating operation | movement of the weight measurement system 1 applicable to the weight estimation apparatus which concerns on embodiment of this invention. It is a figure which shows the three-dimensional point cloud data acquired from the upper surface of the cow body. It is a figure which shows the three-dimensional point cloud data acquired from the side of the cow body. It is a schematic diagram when a cow is not located in parallel with the x-axis in the measurement region. It is a schematic diagram which shows that a cow body is located diagonally in the area | region. It is a schematic diagram which shows having approximated the cross-sectional shape which carried out the ring cutting process to the ellipse shape. It is a schematic diagram which shows the approximate straight line 35 which passes through each elliptical center point P1-P6. It is a schematic diagram which shows the result of having rotated the approximate straight line. It is a schematic diagram which shows having approximated the cross-sectional shape of the shape obtained by the rotation process to elliptical shape. It is a figure which shows to the uppermost position of the cow body from the ground. It is a cross-sectional view of the cow body extracted about the cross section (A-A ') shown in FIG. It is a graph which shows the result of the experiment which used the parallel hexahedron mentioned above as a measuring object. (A) (b) is a figure which shows the three-dimensional point cloud data of the synthetic | combination cow. It is a figure which shows the part measured in the torso of a cow body. (A)-(f) is sectional drawing which shows the result measured horizontally every 100 mm. (A)-(d) is sectional drawing which shows the result of having measured the main part perpendicularly | vertically.

  This embodiment uses a Kinect Sensor (registered trademark) (hereinafter referred to as a three-dimensional measuring instrument), which is one of measuring instruments for acquiring three-dimensional data. However, a three-dimensional shape is obtained from a measuring instrument having the same function. If data (three-dimensional point group coordinate data) can be acquired, the present invention is not particularly limited to the above-described kinetic sensor. In this embodiment, the measurement object of the three-dimensional measuring instrument is a bovine body, but it can also be used for large and medium animals such as horses and pigs and other small animals in addition to the bovine body. Hereinafter, in this embodiment, a case where a kinetic sensor is used as one of the three-dimensional measuring devices will be described.

Embodiments of the present invention will be described with reference to the drawings.
<First Embodiment>
With reference to FIG. 1, the structure of the weight measurement system 1 applicable to the weight estimation apparatus of the moving body which concerns on 1st Embodiment of this invention is demonstrated. In the present embodiment, description will be given by taking a cow, which is one of the animals, as an example.
As shown in FIG. 1, the animal body weight measurement system 1 includes a ramen structure 11, a pedestal portion 11a is disposed on the floor surface, and a lower portion of the column portion 11b is joined to the pedestal portion 11a to form the column portion 11b. Extends vertically, and a beam portion 11c extends horizontally from the upper portion of the column portion 11b. The column part 11b and the beam part 11c are a rigid joint structure which is welded and integrated, and a joint may be provided on the beam part 11c in order to improve portability during transportation.
The beam portion 11 c is provided with a three-dimensional measuring device 14 </ b> A, which emits a plurality of spotlights above the cow 3 from the three-dimensional measuring device 14 </ b> A and images a reflection point reflected from the upper surface of the cow 3.
The column part 11b is provided with a three-dimensional measuring instrument 14B, which radiates a plurality of spot lights from the three-dimensional measuring instrument 14B to the left side of the cow body 3 and images the reflection points reflected from the left side surface of the cow body 3. .
The tilt sensor 12 is disposed on the pillar portion 11b, detects the tilt of the cow in the direction of gravity acceleration, and outputs tilt data.
The second three-dimensional measuring instrument 14B is arranged on the side opposite to the left side of the cow body 3 facing, and projects a random light spot on the right side of the cow body 3 and is reflected from the side surface of the cow body 3. You may provide the 3rd measuring device which picturizes the reflected point and acquires the 3D point cloud data of the left side of a cow body.

Here, the three-dimensional measuring instrument 14 will be described.
As shown in FIG. 2A, the three-dimensional measuring instrument 14 includes a projector 16, an RGB color camera 18, and an infrared camera 20.
As shown in FIG. 2B, the projector 16 converts a high-intensity infrared LED 16b that emits light in the infrared band serving as a light source, and infrared light emitted from the high-intensity infrared LED 16b into substantially parallel beam light. A lens 16c, a light diffusing element (diffuser) 16d that suppresses uneven illuminance of the beam light from the lens 16c and adjusts to uniform light, and a plurality of microlenses are arranged, and a plurality of random light spots are formed by the plurality of microlenses. And a lens 16a that projects forward the spot light formed by the micro lens array 16e as a plurality of spot lights having a spread of a predetermined angle β.

  The RGB color camera 18 is an RGB color camera in the visible light band having a resolution of several million pixels, and receives light from a subject with an imaging lens 18a, and an optical image of an imaging target imaged by the imaging lens 18a. Various image processing is performed on image signals such as CCD (Charge Coupled Device) and CMOS (Complementary Metal Oxide Semiconductor) that convert analog electrical signals, and frame images that are digital signals by A / D converting analog electrical signals. And an image processing unit including an ISP (Image Signal Processor) or the like.

The infrared camera 20 is a low-resolution infrared-band camera, and receives infrared light, which is spot light reflected from a subject, by the imaging lens 20a, and forms an optical image of the imaging target formed by the imaging lens 20a. Image sensors such as CCD (Charge Coupled Device) and CMOS (Complementary Metal Oxide Semiconductor) that convert analog signals to analog electrical signals, and analog image A / D conversion to various image processing for frame images that are digital signals And an image processing unit composed of an image signal processor (ISP) or the like.
The three-dimensional measuring instrument 14 is an I / F unit that transmits and receives control signals and image data to and from the projector 16, the RGB color camera 18, and the infrared camera 20 via a cable. And so on.

Next, a characteristic configuration of the personal computer 22 for processing data from the tilt sensor 12 and the three-dimensional measuring instruments 14A and 14B shown in FIG. 1 will be described with reference to FIG.
As shown in FIG. 3, the personal computer 22 includes a control unit 22a, a storage unit 22b, a display control unit 22c, and a monitor 22d.
The tilt sensor 12 and the three-dimensional measuring instruments 14A and 14B are connected to the control unit 22a. The control unit 22a stores the data at substantially the same time input from the tilt sensor 12 and the three-dimensional measuring devices 14A and 14B in the storage unit 22b, and performs arithmetic processing on each data read from the storage unit 22b. The calculation result is displayed on the monitor 22d via the display control unit 22c.

Next, how to obtain the distance to the subject will be described with reference to FIGS. As a method for obtaining the distance to the subject, the following well-known method (transistor technology August 2012 P62 to P68) is known.
Here, with respect to the three-dimensional point cloud data obtained by the three-dimensional measuring instrument 14, a small region (Nx × Ny pixels) having a predetermined number of images is determined, and a cross-correlation value is calculated for a horizontally shifted position. The shift amount indicating the highest correlation value is obtained, and the position is calculated from the obtained shift amount.
For example, the closer the subject position is to the screen position, the smaller the horizontal shift amount, and vice versa. Therefore, based on the three-dimensional point cloud data obtained from the infrared camera 20, it is determined how much the original random dot pattern has shifted in the horizontal direction in units of blocks composed of small regions (Nx × Ny pixels). For example, the depth (Z value) of the block can be obtained mathematically.

  As shown in FIG. 4A, the projector 16 and the infrared camera 20 are separated by a distance d. As shown in FIG. 4B, in the camera optical system, assuming that the Lx width is an Nx image at a distance L, the distance Zd to the subject, the original of the partial pattern Assuming that the bit position is x0 and the shifted bit position is x1, the following equation holds from the geometric similarity.

これにより、被写体までの距離Ｚｄを求めることができる。

Thereby, the distance Zd to the subject can be obtained.

Next, with reference to FIG. 5, the coordinate alignment of the weight estimation apparatus will be described.
Before performing the three-dimensional measurement of the cow body 3, it is necessary to match the coordinate axes of the installed three-dimensional measuring devices 14A and 14B. For this purpose, as shown in FIG. 5, four support columns 31 each having a marker (sphere) 30 are provided, and the reference marker 30 is installed at a position where measurement can be performed from all three-dimensional measuring instruments. However, each marker 30 needs to be installed so as not to be on the same plane.
The coordinate system fixed to the three-dimensional measuring instrument 14B is (x, y, z), and the coordinate system fixed to the three-dimensional measuring instrument 14A is (x ′, y ′, z ′).
A coordinate system conversion formula from (x ′, y ′, z ′) to (x, y, z) is as follows in consideration of rotation and movement.

これを展開すると、

If you expand this,

となる。

It becomes.

For the same marker, the positions (x, y, z) and (x ′, y ′, z ′) measured by the respective three-dimensional measuring devices 14A and 14B are substituted. In determining the conversion formula of (1), since there are 12 variables (a11 to a34) to be obtained, four or more combinations of (x, y, z) and (x ′, y ′, z ′) are ( 4) Substituting the equation, 12 variables are obtained from the obtained simultaneous equations. When all variables are obtained, the coordinate system (x ′, y ′, z ′) fixed to the three-dimensional measuring instrument 14A is converted to the coordinate system (x, y, z) fixed to the three-dimensional measuring instrument 14B. An expression is determined.
When the soil condition is poor, the three-dimensional measuring instrument may be tilted with respect to the direction of gravity. Therefore, the coordinate system (x, y, z) is corrected from the value detected by the tilt sensor 12 fixedly arranged on the three-dimensional measuring instrument 14B, and the z axis is corrected to be parallel to the direction of gravity. To do.

Next, the operation of the weight measurement system 1 applicable to the weight estimation apparatus according to the embodiment of the present invention will be described with reference to the flowchart shown in FIG. When the personal computer 22 is turned on, for example, the operating system OS stored in the storage unit 22b (hard disk) is booted and started on a RAM (not shown) by the control unit 22a (CPU). A plurality of types of application software can be executed on the OS.
The weight measurement system 1 is configured to operate when the control unit 22a executes the program shown in the flowchart shown in FIG. 6 on the OS.
Steps S10 to S20 described later indicate a dedicated application software program for operating the three-dimensional measuring instruments 14A and 14B. It is executed under the management of the OS so as to be processed in parallel by the control unit 22a. On the other hand, steps S30 to S90 to be described later show programs of other application software.

First, in step S10, the control unit 22a performs 3D point cloud data acquisition processing. That is, a dedicated application software program for operating the three-dimensional measuring instrument 14A is executed on the OS.
Specifically, the control unit 22 a transmits a light projection start signal to the projector 16 provided in the three-dimensional measuring instrument 14. The three-dimensional measuring instrument 14 </ b> A that has received the projection start instruction turns on the high-intensity infrared LED 16 b provided in the projector 16. Accordingly, the high-intensity infrared LED 16b emits light, the infrared light from the high-intensity infrared LED 16b is converted into substantially parallel beam light by the lens 16c, and the illuminance unevenness of the beam light from the lens 16c is suppressed by the light diffusion element 16d. And adjust to a homogeneous light. Next, the homogeneous light suppressed by the light diffusing element 16d is incident on the microlens array 16e, a plurality of random light spots are formed by the plurality of microlenses, and the plurality of point lights have a spread of a predetermined angle β. A plurality of spot lights are projected downward from the lens 16a.

Next, the control unit 22a transmits an imaging start signal to the infrared camera 20 provided in the three-dimensional measuring instrument 14A. The three-dimensional measuring instrument 14 </ b> A that has received the imaging start instruction activates the infrared camera 20. At this time, infrared light, which is spot light projected from the projector 16 and reflected from the subject, is received by the imaging lens 20a, and an optical image of the imaging target formed by the imaging lens 20a is converted into an analog electrical signal by the image sensor. The analog electrical signal is A / D converted by the image processing unit, and various image processes are performed on the frame image which is a digital signal by the image processing unit. Next, the image data output from the image processing unit is transmitted to the personal computer 22 via the I / F unit.
Thereby, the image data above the cow output from the three-dimensional measuring instrument 14A can be acquired. Since the infrared light spot light is emitted from the three-dimensional measuring instrument 14 </ b> A, the reflection point on the skin surface of the cow body 3 can be photographed by the infrared camera 20. FIG. 7 is a diagram showing three-dimensional point cloud data acquired from above the cow 3.
The image data acquired from the three-dimensional measuring instrument 14 is used to calculate the three-dimensional point cloud data by obtaining the distances Zd to the individual light spots according to the above formulas (1) and (2). The original point cloud data (A) is stored in the storage unit 22b.

At the same time, in step S20, the control unit 22a performs processing for acquiring three-dimensional point group data, as in step S10. That is, a dedicated application software program for operating the three-dimensional measuring instrument 14B is executed on the OS.
Thereby, the image data of the left side of the cow output from the three-dimensional measuring instrument 14B can be acquired. FIG. 8 is a diagram showing three-dimensional point cloud data acquired from the side of the cow body. The image data acquired from the three-dimensional measuring instrument 14 is used to calculate the three-dimensional point cloud data by obtaining the distances Zd to the individual light spots according to the above formulas (1) and (2). The original point cloud data (B) is stored in the storage unit 22b.

Next, in step S30, the control unit 22a reads the three-dimensional point cloud data (A) (B) acquired from the storage unit 22b, and the two three-dimensional point cloud data (A) (A) ( B) is synthesized, and bovine body data composed of three-dimensional shape data is generated and stored in the storage unit 22b.
The second three-dimensional measuring instrument 14B is arranged on the right side opposite to the left side of the cow body 3 facing, and projects a random light spot from the right side of the cow body 3 and is reflected by the cow body 3. In the case where a third 3D measuring device that captures the reflected points and obtains 3D point cloud data of another side of the cow body is obtained, in step S30, the data is acquired from the first to third 3D measuring devices. You may comprise so that the three-dimensional point cloud data of an upper surface, a left side surface, and a right side surface may be synthesize | combined and the cow data which consists of three-dimensional shape data may be produced | generated.

By the way, since the cow body 3 is a living body, there is a sufficient possibility of movement. For this reason, as shown in FIG. 9, the cow may not be positioned in parallel with the x-axis in the measurement region. FIG. 9 is a schematic diagram of synthesis result data on the (x, y) plane, in which a plane including a body parallel to the (x, y) plane is set.
Here, in order to estimate the circumference, height, and the like of the bovine torso from the acquired three-dimensional point cloud data, it is necessary to arrange the direction of the torso in a global coordinate system.
In general, since the ground is not flat, the inclination of the frame structure 11 with respect to the ground should be considered.
When the ramen structure 11 (FIG. 1) is inclined with respect to the ground, a roll angle (around the x axis) and a pitch angle (around the z axis) exist. Since the weight measurement system 1 includes the inclination sensor 12, the inclination sensor 12 can easily detect the inclination of the roll angle and the pitch angle by detecting the direction of acceleration of gravity.

Therefore, in step S40, the control unit 22a corrects the inclination of the cow data, which is the synthesis result data, with respect to the gravitational direction according to the inclination data acquired from the inclination sensor 12, and stores the correction result data in the storage unit 22b. .
The inclination of the cow with respect to the direction of acceleration of gravity is detected by the inclination sensor 12, and the yaw angle around the direction of acceleration of gravity is detected by image processing. FIG. 15 shows the global coordinate system.

Here, with reference to FIGS. 15 and 16, the principle of the yaw angle detection process will be described with reference to the bovine shape data.
In FIG. 15, the height H indicates from the ground up to the top position of the cow body, and the cross section (AA ′) is set to the torso. In addition, the height of the cross section (AA ′) is set to 0.67H so that the cross-sectional shape is included in the body.
FIG. 16 shows a cross-sectional view of the bovine body extracted with respect to the cross-section (AA ′).
In FIG. 16, the torso is tilted by the yaw angle θy about the y-axis.
In order to calculate the angle of inclination, the cross-sectional shape is approximated as an elliptical shape using the least square method. The approximate oval shape is indicated by a curve (broken line) in FIG. The inclination of the cross section of the cow can be detected as an elliptical inclination.
The approximate elliptical shape is estimated at 10 cm intervals around the cross-section AA ′ and the average rotation around the y axis is determined as the amount of rotation of the cow around the y axis. The body of the cow is rotated by this amount, and the control unit 22a aligns the cow body with the global coordinates.

  Here, correction of the bovine data along the global coordinate system is performed for the roll angle (around the x axis), the yaw angle (around the y axis), and the pitch angle (around the z axis). Each correction formula is shown below.

  The initial coordinate system (x, y, z) of the bovine data can be aligned along the global coordinate system (x ′, y ′, z ′) by the following equation.

As described above, the three-dimensional point cloud data forming the bovine data is rotated in the global coordinate system based on the roll angle, the pitch angle, and the yaw angle, and the cow data obtained as a result of the rotation is stored in the storage unit 22b. .
Thereby, when there is an inclination in the measurement environment of the cow body 3, the roll angle and pitch of the second three-dimensional measuring instrument 14B are measured by the inclination sensor 12 arranged in the ramen structure 11 in the vicinity of the second three-dimensional measuring instrument 14B. The angle is detected, the yaw angle is calculated based on the detected roll angle and pitch angle values, and the composition result is calculated based on the detected roll angle and pitch angle values and the calculated yaw angle value. The cow data 3i can be corrected, and more accurate cow data can be acquired.

  Next, in step S50, the control unit 22a performs a rounding process on the synthesis result data into a cross-sectional shape parallel to the direction of gravity, and stores the processed bovine data in the storage unit 22b. That is, as shown in FIG. 10, a plane parallel to the x-axis is calculated for the cross-sectional shape of the bovine body, and each cross-sectional shape is obtained. FIG. 10 is a schematic diagram showing that the cow body is positioned obliquely in the region.

  Next, in step S60, the control unit 22a approximates each cross-sectional shape with an elliptical shape according to the least-square approximation with respect to the cross-sectional shape subjected to the circular cutting process, and stores the processed bovine data in the storage unit 22b. FIG. 11 is a schematic diagram showing that the cross-sectional shape subjected to the ring cutting process is approximated to an elliptical shape.

  Next, in step S70, the control unit 22a obtains the respective elliptical center points P1 to P6 and connects the center points of the respective cross-sectional shapes, that is, approximate straight lines passing through the respective elliptical center points P1 to P6. And the obtained approximate straight line is stored in the storage unit 22b. FIG. 12 is a schematic diagram showing approximate straight lines 35 passing through the respective elliptical center points P1 to P6.

Next, in step S80, the control unit 22a performs a rotation process. That is, the three-dimensional coordinate data of the cow is rotated around the z-axis by the value of the inclination of the approximate straight line, the shape parallel to the x-axis is obtained, and the obtained cow data is stored in the storage unit 22b. . FIG. 13 is a schematic diagram showing the result of rotating the approximate straight line.
In this way, for the bovine data generated by combining the 3D point cloud data acquired from the bovine, a plurality of cross-sectional shapes parallel to the gravity direction are approximated to an elliptical shape, and the center of each elliptical shape is approximated. A straight line connecting the points is calculated, the bovine data is rotated according to the inclination of the straight line, and the cross-sectional shape parallel to the gravity direction and the cross-sectional shape parallel to the horizontal direction after the rotation processing are approximated to an elliptical shape. Thus, even when there is a missing portion in the three-dimensional point cloud data acquired from the bovine, an elliptical shape can be approximated, so that an elliptical shape very close to the cross-sectional shape of the bovine can be estimated.

Next, in step S90, the control unit 22a ellipses the cross-sectional shape parallel to the gravity direction and the cross-sectional shape parallel to the horizontal direction according to the least square approximation with respect to the bovine data parallel to the x-axis obtained by the rotation process. Each ellipse data that is approximated to the shape and becomes the processing result is stored in the storage unit 22b.
FIG. 14 is a schematic diagram showing that the cross-sectional shape of the bovine data 3i obtained by the rotation process is approximated to an elliptical shape 36. Further, with respect to the cow data shown in FIG. 19, an example of the cross-sectional shape in the horizontal direction is shown in FIG. 20, and an example of the cross-sectional shape in the gravity direction is shown in FIG.

Here, the performance of the weight measurement system will be described.
In order to evaluate the performance of the weight measurement system, a rectangular parallelepiped having a known dimension is used as a measurement object. The distance from the three-dimensional measuring instrument 14B to the object is changed from 1.0 m to 2.5 m every 0.5 m. FIG. 17 shows the results of an experiment using the above-described parallelepiped as a measurement object. It can be understood that the RMS error (mean square error) is less than 0.003 m, and the system performance is sufficient for bovine measurement.

Next, the measurement result of the bovine body will be described.
The target cow's body age is 18 months, the weight measured by the cow impactor is 533 kg, the cross height (HH) is 131.0 cm, the body length (BL) is 216.0 cm, and the chest circumference (CG) is It is 196.0 cm and the width is 50.0 cm.
The three-dimensional point cloud data detected by the respective three-dimensional measuring devices 14A and 14B are shown in FIGS. Each three-dimensional point cloud data is synthesized by the synthesis process in step S30.
FIG. 18 is a diagram showing the three-dimensional point cloud data of the synthesized bovine. The number of data is about 250,000, and each three-dimensional point cloud data is composed of (x, y, z) coordinates and RGB color data on the surface of the cow.

FIG. 19 is a diagram showing a portion measured on the body of the cow body. The cross-sectional shape (H1-H6) shown in FIG. 20 is a cross-sectional view showing the result of horizontal measurement every 100 mm. Also, the four main parts (V1-V4) shown in FIG. 21 are selected vertically.
Once the torso data is created, the main features required to evaluate the cow can be estimated by the personal computer 22. In particular, chest circumference (V2, V3), waist height and trunk length, etc. are important parameters for evaluating the growth of the cow.
According to this embodiment, for example, H5 is used to estimate the body length (BL), and the highest position of V1 is used to estimate the cross height (HH).
A circle close to the chest circumference (V2, V3) is used to estimate the chest circumference (CG). A circle (broken line) in FIG. 21 is a circle joined around the chest trunk.
The error between the estimation processing result and the manual measurement result according to the present embodiment was less than 5% for the chest circumference (CG) and the body length (BL).
On the other hand, for the cross height (HH), the error in computer image using manual light and manual measurement was less than 7%. One of the causes of the difference between the two results is due to the bovine torso hair.

Next, in step S100, the control unit 22a obtains the chest circumference (CG), the body length (BL), the cross height (HH), etc. from the respective ellipse data (see FIGS. 20 and 21) as the processing result in step S90. Dimension data is extracted, and each dimensional data as a processing result is stored in the storage unit 22b.
Next, in step S110, the control unit 22a estimates the weight data of the bovine by substituting the dimension data extracted in step S100 into the weight calculation formula.
Examples of weight (BW) calculation formulas are chest circumference (CG), body length (BL), and cross height (HH).
BW (kg) = 3.74CG + 2.4BL + 2.3HH-862.2 (9)
It becomes. The unit is cm.
Here, the weight calculation formula (9) is the general formula (10): BW = A × CG + B × BL + C × HH−D (10)
It is obtained by acquiring as much actual measurement data as possible with a bull impactor or measure, substituting it into CG, BL, and HH and calculating the counts A, B, C, and D by the least square method.
Substituting the dimensional data such as chest circumference (CG), body length (BL), cross height (HH) into the weight calculation formula (9) to estimate the weight data of the cow body, storing it in the storage unit 22b, The weight data is displayed on the monitor 22d via the display control unit 22c.

  As described above, a random light spot in the infrared band is projected from at least one of the upper, left, and right sides of the moving object stationary in a predetermined posture, and the reflected point reflected by the moving object is imaged. The three-dimensional point cloud data of the moving object is obtained, and the obtained three-dimensional point cloud data is approximated by approximating the cross-sectional shape parallel to the gravity direction and the cross-sectional shape parallel to the horizontal direction to an elliptical shape. By extracting characteristic dimension data of the moving object from the ellipse shape and substituting the extracted characteristic dimension data of the moving object into a predetermined weight calculation formula, The dimension of the moving object can be measured with high accuracy by the original measurement, and the accuracy of the weight estimated from the obtained dimension data can be improved.

<Modification 1>
The animal body weight estimation apparatus according to the first embodiment of the present invention has been described as applied to a configuration including the three-dimensional measuring devices 14A and 14B as the configuration of the weight measurement system 1 shown in FIG. Is not limited to the case where two three-dimensional measuring instruments are used.
That is, as a first modification, one first three-dimensional measuring instrument 14A is disposed above the facing cow body 3, and projects a random light spot in the infrared band from above facing the cow body 3, You may comprise so that the reflective point reflected by the cow body 3 may be imaged and the three-dimensional point cloud data of the upper surface of the cow body 3 may be acquired.
When the three-dimensional measuring instrument 14A acquires bovine data from the upper surface of the bovine, a plurality of cross-sectional shapes parallel to the direction of gravity are approximated to elliptical shapes (V1 to V4 in FIG. 21) with respect to the bovine data. To do.
At this time, when each of the approximated elliptical shapes (gravity direction) is seen on a cross section parallel to the horizontal direction, several points are scattered. Therefore, the above-mentioned several points are regarded as a part of the bovine data, and the cross-sectional shape parallel to the horizontal direction is obtained with respect to the points included in each elliptical shape (gravity direction) and the bovine data. What is necessary is just to approximate an ellipse shape (H1-H6 of FIG. 20).
Thereby, characteristic dimension data of the cow can be extracted from the approximated elliptical shape.

<Modification 2>
The animal body weight estimation apparatus according to the first embodiment of the present invention has been described as applied to a configuration including the three-dimensional measuring devices 14A and 14B as the configuration of the weight measurement system 1 shown in FIG. Is not limited to the case where two three-dimensional measuring instruments are used.
That is, as a second modification, one first three-dimensional measuring instrument 14B is disposed on the side of the cow body 3 facing, and a random light spot in the infrared band is projected from the side facing the cow body 3. And you may comprise so that the reflective point reflected by the cow body 3 may be imaged and the three-dimensional point cloud data of the upper surface of the cow body 3 may be acquired.
When the three-dimensional measuring instrument 14B acquires the bovine data from the side of the bovine, a plurality of cross-sectional shapes parallel to the horizontal direction are approximated to elliptical shapes (H1 to H6 in FIG. 20) with respect to the bovine data. To do.
At this time, when each of the approximated oval shapes (horizontal direction) is seen on a cross section parallel to the direction of gravity, several points are scattered. Therefore, the above-mentioned several points are regarded as a part of the bovine data, and the cross-sectional shape parallel to the gravitational direction is set for the points included in each elliptical shape (horizontal direction) and the bovine data as approximated results. What is necessary is just to approximate an ellipse shape (gravity direction) (V1-V4 of FIG. 21).
Thereby, characteristic dimension data of the cow can be extracted from the approximated elliptical shape.

<Modification 3>
The animal body weight estimation apparatus according to the first embodiment of the present invention has been described as applied to a configuration including the three-dimensional measuring devices 14A and 14B as the configuration of the weight measurement system 1 shown in FIG. Is not limited to the case where two three-dimensional measuring instruments are used.
That is, as a third modification, the second three-dimensional measuring instrument 14B is arranged on a side different from the side of the cow 3 facing the random, and the infrared band is random from another side facing the cow 3. A third 3D measuring device that projects a light spot, images the reflection point reflected by the cow body 3, and obtains 3D point cloud data on another side surface of the cow body 3 may be provided.
In step S30 described above, in the synthesis process, the upper surface, the side surface, and the 3D point cloud data of the side surface different from the side surface obtained from the first to third 3D measuring instruments are combined to form the cow body composed of the 3D shape data. Data 3i may be generated.
Thereby, since the more exact cow body data 3i can be produced | generated, the weight of a cow body can be estimated more correctly.

<Modification 4>
In the first embodiment of the present invention, description has been given by taking a cow that is one of the animal bodies as an example, but the present invention is not limited to the cow body, and can be applied to other animal bodies. is there. For 3D point cloud data acquired from other moving objects, the cross-sectional shape parallel to the gravitational direction and the cross-sectional shape parallel to the horizontal direction are approximated to an elliptical shape. It is only necessary to extract simple dimension data and substitute the extracted characteristic dimension data of the moving object into a weight calculation formula appropriate for the moving object to estimate the weight of the moving object.

  DESCRIPTION OF SYMBOLS 1 ... Weight measurement system, 11 ... Ramen structure, 11a ... Base part, 11b ... Column part, 11c ... Beam part, 12 ... Tilt sensor, 14 ... Three-dimensional measuring instrument, 14A ... Three-dimensional measuring instrument, 14A, 14B ... Three-dimensional measuring instrument, 14B ... Three-dimensional measuring instrument, 16 ... Projector, 16a ... Lens, 16b ... High brightness infrared LED, 16c ... Lens, 16d ... Light diffusing element, 16e ... Micro lens array, 18 ... RGB color camera, 18a DESCRIPTION OF SYMBOLS ... Imaging lens, 20 ... Infrared camera, 20a ... Imaging lens, 22 ... Personal computer, 22a ... Control part, 22b ... Memory | storage part, 22c ... Display control part, 22d ... Monitor, 30 ... Marker, 31 ... Support | pillar, 35 ... Approximation Straight line, 36 ... elliptical shape, 3i ... cow data

Claims (12)

A weight estimation device for an animal body that estimates the weight of the animal body by substituting the dimensional data of the animal body into a predetermined weight calculation formula,
A random light spot in the infrared band is projected from at least one of the upper, left side, and right side of the moving object stationary in a predetermined posture, and the reflected point reflected by the moving object is imaged. A three-dimensional measuring instrument for acquiring three-dimensional point cloud data of an animal body;
Shape approximation processing means for approximating the cross-sectional shape parallel to the gravity direction and the cross-sectional shape parallel to the horizontal direction to an elliptical shape for the three-dimensional point cloud data acquired by the three-dimensional measuring instrument,
Shape extraction means for extracting characteristic dimension data of the moving object from the elliptical shape approximated by the shape approximation processing means,
A weight estimation apparatus for an animal body, wherein the weight data of the animal body is estimated by substituting characteristic dimension data of the animal body extracted by the shape extraction means into the predetermined weight calculation formula. A random light spot in the infrared band is projected from another direction different from the one direction, and the reflection point reflected by the moving object is imaged to obtain three-dimensional point cloud data of the moving object. With other 3D measuring instruments
The animal body weight estimation apparatus according to claim 1, further comprising: a combining unit that combines the three-dimensional point cloud data acquired from the three-dimensional measuring instrument and the other three-dimensional measuring instrument. Detecting means for detecting a roll angle and a pitch angle of the three-dimensional measuring instrument;
A yaw angle calculation means for calculating a yaw angle based on the roll angle and the pitch angle detected by the detection means;
And a correction unit that corrects the three-dimensional point cloud data based on the roll angle and pitch angle detected by the detection unit and the yaw angle calculated by the yaw angle calculation unit. Item 1. An apparatus for estimating the weight of an animal body according to Item 1.   The shape approximation processing means approximates a plurality of cross-sectional shapes parallel to the gravity direction to an elliptical shape with respect to the three-dimensional point cloud data, calculates a straight line connecting the center points of the respective elliptical shapes, The cross-sectional shape parallel to the gravity direction and the cross-sectional shape parallel to the horizontal direction after the rotation processing are approximated to an elliptical shape by rotating the three-dimensional point cloud data according to an inclination. Animal body weight estimation device.   When the three-dimensional measuring instrument acquires the three-dimensional point cloud data from above the moving object, the shape approximation processing means has a plurality of cross sections parallel to the gravitational direction with respect to the three-dimensional point cloud data. The shape is approximated to an elliptical shape, and the cross-sectional shape parallel to the horizontal direction is approximated to an elliptical shape with respect to the points included in each of the approximated elliptical shapes and the three-dimensional point cloud data. The animal body weight estimation apparatus according to claim 1.   When the three-dimensional measuring instrument acquires the three-dimensional point cloud data from the left or right side of the moving object, the shape approximation processing means is parallel to the three-dimensional point cloud data in the horizontal direction. Approximating a plurality of cross-sectional shapes to an elliptical shape, and approximating a cross-sectional shape parallel to the gravitational direction to an elliptical shape for the points included in each of the approximated elliptical shapes and the three-dimensional point cloud data The weight estimation apparatus for an animal body according to claim 1. A method of estimating the weight of an animal body by substituting the dimensional data of the animal body into a predetermined weight calculation formula to estimate the weight of the animal body,
A random light spot in the infrared band is projected from at least one of the upper, left side, and right side of the moving object stationary in a predetermined posture, and the reflected point reflected by the moving object is imaged. A shape that approximates the cross-sectional shape parallel to the gravity direction and the cross-sectional shape parallel to the horizontal direction to an elliptical shape with respect to the three-dimensional point cloud data acquired by the three-dimensional measuring instrument that acquires the three-dimensional point cloud data of the moving object An approximation processing step;
A shape extraction step of extracting characteristic dimension data of the moving object from the elliptical shape approximated by the shape approximation processing step,
A weight estimation method for a moving body, characterized in that the weight of the moving body is estimated by substituting characteristic dimension data of the moving body extracted in the shape extraction step into the predetermined weight calculation formula.   A random light spot in the infrared band is projected from the three-dimensional measuring instrument and another direction different from the one direction, and the reflection point reflected by the moving object is imaged to capture the moving object. 8. The method of estimating the weight of an animal body according to claim 7, further comprising a synthesis step of synthesizing each three-dimensional point cloud data acquired from another three-dimensional measuring instrument that acquires the three-dimensional point cloud data. A yaw angle calculating step for calculating a yaw angle based on the roll angle and the pitch angle detected by the detecting means for detecting the roll angle and the pitch angle of the three-dimensional measuring instrument;
And a correction step of correcting the three-dimensional point cloud data based on the roll angle and pitch angle detected by the detection means and the yaw angle calculated by the yaw angle calculation step. Item 8. The method for estimating the weight of an animal body according to Item 7.   The shape approximation processing step approximates a plurality of cross-sectional shapes parallel to the gravity direction to an elliptical shape with respect to the three-dimensional point cloud data, calculates a straight line connecting the center points of the respective elliptical shapes, 8. The three-dimensional point cloud data is rotated according to an inclination, and a cross-sectional shape parallel to the gravity direction and a cross-sectional shape parallel to the horizontal direction after the rotation processing are approximated to an elliptical shape. Method for estimating the weight of an animal body.   When the three-dimensional measuring instrument acquires the three-dimensional point cloud data from above the moving object, the shape approximation processing step includes a plurality of cross sections parallel to the gravitational direction with respect to the three-dimensional point cloud data. The shape is approximated to an elliptical shape, and the cross-sectional shape parallel to the horizontal direction is approximated to an elliptical shape with respect to the points included in each of the approximated elliptical shapes and the three-dimensional point cloud data. The weight estimation method for an animal body according to claim 7.   When the three-dimensional measuring instrument acquires the three-dimensional point cloud data from the left or right side of the moving object, the shape approximation processing step is parallel to the three-dimensional point cloud data in the horizontal direction. Approximating a plurality of cross-sectional shapes to an elliptical shape, and approximating a cross-sectional shape parallel to the gravitational direction to an elliptical shape for the points included in each of the approximated elliptical shapes and the three-dimensional point cloud data The method for estimating the weight of an animal body according to claim 7.

Images