WO2023190352A1 - Growing condition evaluation system - Google Patents

Abstract

Provided is a growing condition evaluation system capable of suitably evaluating a development condition of an animal. This growing condition evaluation system 10 comprises: a laser measurement apparatus 13 for receiving laser light that has been emitted toward and reflected from an animal (cow 51), thereby acquiring point group data D1 indicating the appearance of the animal (cow 51) in three-dimensional coordinates; a reference location detection unit 67 for detecting a reference location Pr in the point group data D1; a normalization unit 69 for adjusting the size of the point group data D1 such that the reference location Pr has a uniform size, and generating normalized data D7; an approximation curve calculation unit 74 for calculating an approximation curve Ac that is suitable for the normalized data D7; and an evaluation value calculation unit 75 for calculating, on the basis of the approximation curve Ac, an evaluation value Ev indicating an evaluation of the animal (cow 51).

Description

Growth status evaluation system

The present disclosure relates to a growth status evaluation system.

It has been known to use BCS (body condition score) and the like to judge the growth status of animals such as cows and pigs.

As for the BCS, a health condition estimation device is being considered that can automatically determine the BCS without touching the animal (see, for example, Patent Document 1). The health state estimation device uses a distance image sensor to acquire a group of three-dimensional coordinates that indicate the three-dimensional shape of the animal, and based on the group of three-dimensional coordinates, it calculates feature values that indicate the width of the animal's body and the animal's Feature values indicating the position of the spine are obtained, and the BCS is calculated based on them. Therefore, the conventional health state estimation device can obtain an evaluation based on the BCS with suppressed variations while reducing the burden on the animal.

Patent No. 6777948

By the way, in the conventional health condition estimation device, the evaluation of the breeding status is obtained from the feature amount indicating the degree of concavity of the lumen, based on the three-dimensional shape of the animal reproduced from the three-dimensional coordinate group of the animal.

Here, for animals such as cows and pigs, it is required to be able to judge the state of meat (good or bad) as a growth condition. However, since animals of the same species have different skeletal sizes, it is not possible to determine the state of fleshiness even if the feature quantity that simply indicates the degree of concavity of the lumen is used in conventional health condition estimation devices. It is difficult to make appropriate judgments.

The present disclosure has been made in view of the above circumstances, and aims to provide a growth status evaluation system that can appropriately evaluate the growth status of animals.

In order to solve the above-mentioned problems, the growth condition evaluation system of the present disclosure uses a laser beam that receives reflected light from the animal of the emitted laser beam to obtain point cloud data indicating the external shape of the animal in three-dimensional coordinates. a measuring device, a reference point detection unit that detects a reference point in the point cloud data, and a normalization unit that adjusts the size of the point cloud data so that the reference point has a uniform size to generate normalized data. an approximate curve calculation unit that calculates an approximate curve that matches the normalized data, and an evaluation value calculation unit that calculates an evaluation value indicating the evaluation of the animal based on the approximate curve. shall be.

According to the growth status evaluation system of the present disclosure, it is possible to appropriately obtain an evaluation of the growth status of an animal.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is an explanatory diagram showing the overall configuration of a growth situation evaluation system of Example 1 as an example of a growth situation evaluation system according to the present disclosure. It is a block diagram showing composition of a control system in a growth situation evaluation system. FIG. 2 is an explanatory diagram showing how an imaging device is attached to the growth status evaluation system. FIG. 2 is an explanatory diagram showing an image of a dairy cow taken by an imaging device. It is an explanatory view showing a laser measuring instrument in a growth situation evaluation system. FIG. 2 is a block diagram showing the configuration of a control system in a laser measuring instrument. It is an explanatory view showing point group data acquired by one laser measuring instrument (the 1st). FIG. 7 is an explanatory diagram showing point cloud data acquired by the other laser measuring device (second). FIG. 3 is an explanatory diagram showing composite point cloud data obtained by combining both point cloud data. It is an explanatory view showing virtual animal point cloud data. It is an explanatory view showing animal point group data. It is an explanatory view showing other examples of animal point group data. 13 is an explanatory diagram showing specific part data corresponding to the animal point cloud data of FIG. 12. FIG. FIG. 3 is an explanatory diagram showing how the spine direction in specific part data is made to coincide with the reference axis direction. FIG. 3 is an explanatory diagram showing how a reference point (a pair of hook bones) is detected in specific part data. 16 is an explanatory diagram showing specific part alignment data corresponding to the specific part data of FIG. 15. FIG. It is an explanatory diagram showing specific part alignment data of different sizes, where (a) on the left side shows the appearance seen from the tail side in the reference axis direction, and (b) on the right side shows the appearance seen from above in the vertical direction. . FIG. 3 is an explanatory diagram showing how four slice positions are set on normalized data. 19 is an explanatory diagram showing cross-sectional data, in which data corresponding to each slice position in FIG. 18 are arranged in order from the left. FIG. 19 is an explanatory diagram showing thinned cross-sectional data and calculation data generated from cross-sectional data, and corresponds to the second cross-sectional data from the left in FIG. 19. FIG. 20 is an explanatory diagram showing thinned-out cross-sectional data and calculation data generated from cross-sectional data, and corresponds to the cross-sectional data at the right end of FIG. 19. FIG. 20 is an explanatory diagram showing calculation data, and is shown arranged in order from the left in correspondence with the cross-sectional data of FIG. 19. FIG. FIG. 3 is an explanatory diagram showing how an approximate curve of calculation data is obtained. It is an explanatory view showing the relationship between calculation data of various shapes and approximate curves. It is a flowchart which shows the growth situation evaluation process (growth situation evaluation processing method) performed by the control mechanism of a growth situation evaluation system. This is a table summarizing the relationship between the theoretical cross section and each value obtained using the growth situation evaluation system. FIG. 19 is an explanatory diagram showing approximate curves for each BCS value at the first slice position shown in FIG. 18; FIG. 19 is an explanatory diagram showing approximate curves for each BCS value at the second slice position shown in FIG. 18; FIG. 19 is an explanatory diagram showing approximate curves for each BCS value at the third slice position shown in FIG. 18; FIG. 19 is an explanatory diagram showing approximate curves for each BCS value at the fourth slice position shown in FIG. 18;

Example 1 of a growth situation evaluation system 10 as an embodiment of the growth situation evaluation system according to the present disclosure will be described below with reference to FIGS. 1 to 30. Note that FIGS. 1 and 3 schematically show the area around the drinking fountain 52 (data acquisition area 14) in the cowshed 50, and do not necessarily match the actual appearance of the cowshed 50.

The growth status evaluation system 10 according to the present disclosure automatically evaluates the growth status of animals. The growth status evaluation system 10 of Example 1 evaluates the growth status of a Holstein cow (hereinafter referred to as dairy cow 51) as an example of an animal. As shown in FIGS. 1 to 3, this growth status evaluation system 10 is installed in a cowshed 50 and includes a control mechanism 11, a camera 12, and two laser measuring instruments 13.

A plurality of dairy cows 51 are kept in the cowshed 50, and each dairy cow 51 can be moved. In the cowshed 50, a drinking fountain 52 for the dairy cows 51 is provided. The drinking fountain 52 is configured by placing a water pail 53 in a space in which a plurality of dairy cows 51 can enter. The water pail 53 is long and arranged at one corner of the drinking fountain 52. At this drinking fountain 52, the dairy cow 51 periodically visits the water trough 53 of its own will and stops in front of the water trough 53. For this reason, in the growth status evaluation system 10, the drinking fountain 52 is set as the data acquisition area 14.

As shown in FIG. 2, the control mechanism 11 comprehensively controls the operation of the growth condition evaluation system 10 by deploying a program stored in the storage unit 18 or built-in internal memory 11a on, for example, RAM (Random Access Memory). to control. In the first embodiment, the internal memory 11a is composed of a RAM or the like, and the storage section 18 is composed of a ROM (Read Only Memory), an EEPROM (Electrically Erasable Programmable ROM), or the like. In addition to the above-described configuration, the growth status evaluation system 10 includes a printer that prints measurement results in response to measurement completion signals and instructions from the measurer, an output unit that outputs measurement results to an external memory or server, and an operation controller. An audio output section for notifying the situation etc. is provided as appropriate.

The control mechanism 11 is connected to a camera 12 and two laser measuring instruments 13 (in FIG. 2, one is designated as the first and the other as the second), and controls them as appropriate, and also receives signals (data) from them. It is possible to receive. This connection may be wired or wireless as long as it allows the exchange of signals with the camera 12 and each laser measuring device 13. This control mechanism 11 may be provided at a location different from the cowshed 50, or may be provided within the cowshed 50. An operation section 15 , a display section 16 , a communication section 17 , and a storage section 18 are connected to the control mechanism 11 .

The operation unit 15 is used to operate various settings for evaluating the growth situation, operations and settings of the camera 12 and each laser measuring device 13, and the like. The operation unit 15 may be configured with an input device such as a keyboard and a mouse, or may be configured with software keys displayed on the display screen of the display unit 16 as a touch panel.

The display unit 16 displays the image I (which may be a still image or a moving image (see FIG. 4)) acquired by the camera 12, the point cloud data D1 (see FIGS. 7 and 8) acquired by each laser measuring device 13, and the like. Various data based on the information (see FIGS. 9 to 22, etc.), an evaluation value Ev indicating the evaluation of the animal (dairy cow 51), etc. are displayed. The display section 16 is configured by, for example, a liquid crystal display device (LCD monitor), and is provided in the control mechanism 11 together with the operation section 15. Note that the operation unit 15 and the display unit 16 may be configured by a mobile terminal such as a smartphone or a tablet, and are not limited to the configuration of the first embodiment.

The communication unit 17 communicates with the camera 12, each laser measurement device 13 (its communication unit 45), and external equipment, and drives the camera 12 and each laser measurement device 13, and transmits the image I from the camera 12 and each other. The point cloud data D1 from the laser measuring device 13 can be received. Note that the control mechanism 11 can be configured as a tablet terminal, and in that case, the operation section 15 and the display section 16 can be configured integrally with the control mechanism 11.

The camera 12 is capable of photographing the entire area of the data acquisition area 14, and in the first embodiment, a 4K camera (having a resolution of 3840×2160 pixels) is used. As shown in FIGS. 1 and 3, the camera 12 is attached to an installation plate 54 above the drinking fountain 52, which is the data acquisition area 14, and can photograph the drinking fountain 52 without disturbing the dairy cows 51. It is said that The installation plate 54 is provided on a support 55 of the cowshed 50 for installing the camera 12. The camera 12 constantly photographs the drinking fountain 52 while the growth condition evaluation system 10 is in the operating state, and outputs the photographed image I (its data (see FIG. 4)) to the control mechanism 11.

The two laser measuring instruments 13 are installed at known points, project a pulsed laser beam toward the measurement point, receive the reflected light (pulsed reflected light) of the pulsed laser beam from the measurement point, and measure the pulsed laser beam for each pulse. Performs distance measurement and averages the distance measurement results to perform highly accurate distance measurement. Then, each laser measuring device 13 scans within the set measurement range and sets measurement points evenly over the entire measurement range, thereby determining the surface shape of the object existing in the entire measurement range in three dimensions. A collection of three-dimensional position data (hereinafter referred to as point group data D1) indicated by original coordinates can be obtained. Thereby, each laser measuring device 13 can acquire point cloud data D1 indicating the surface shape of objects existing within the set measurement range.

As shown in FIG. 1, these two laser measuring instruments 13 are located at paired positions across the drinking fountain 52, which is the data acquisition area 14, and at a height that does not interfere with the movement of the dairy cow 51. placed in position. Both laser measuring devices 13 are capable of acquiring point cloud data D1 of at least half of the torso 51a (either the left or right) of the dairy cow 51, regardless of the posture of the dairy cow 51 in the drinking fountain 52 (data acquisition area 14). The positional relationship with respect to the drinking fountain 52 is set as follows. These two laser measuring devices 13 have the same configuration except that they are provided at different positions. Note that the laser measuring device 13 may employ a phase difference measurement method using a light beam modulated at a predetermined frequency, or may employ other methods, and is not limited to the first embodiment. As shown in FIGS. 5 and 6, this laser measuring instrument 13 includes a pedestal 31, a main body part 32, and a vertical rotating part 33.

The pedestal 31 is a part that is attached to the installation base 34. The installation stand 34 is attached to a support 55 of the cowshed 50 for installing the laser measuring device 13. The main body portion 32 is provided on the pedestal 31 so as to be rotatable about a vertical axis relative to the pedestal 31 . The main body portion 32 has a U-shape as a whole, and a vertical rotation portion 33 is provided in the portion between the two. This main body portion 32 is provided with a measuring instrument display section 35 and a measuring instrument operating section 36. The measuring instrument display section 35 is a section that displays various operation icons, settings, etc. for measurement under the control of a measuring instrument control section 43, which will be described later. The measuring device operation section 36 is a place where operations for using and setting various functions in the laser measuring device 13 are performed, and outputs inputted information to the measuring device control section 43. In the main body 32 of the first embodiment, various operation icons displayed on the measuring instrument display section 35 function as a measuring instrument operating section 36. Note that the measuring instrument display section 35 and the measuring instrument operating section 36 can be installed on the installation base 34 of the support column 55 by providing the above-mentioned functions to the operating section 15 and display section 16 in the growth condition evaluation system 10. Remote operation is possible while the device is in the same state. As will be described later, the location and method of installing the laser measuring device 13 can be set as appropriate, as long as it is capable of scanning the data acquisition area 14 (animals such as the dairy cow 51 located there). However, the present invention is not limited to the configuration of the first embodiment.

The vertical rotation section 33 is provided in the main body section 32 so as to be rotatable about a rotation axis extending in the horizontal direction. This vertical rotation section 33 has a distance measuring optical section 37 built therein. The distance measuring optical section 37 projects a pulsed laser beam as distance measuring light and receives reflected light (pulsed reflected light) from a measurement point to measure the optical distance to the measurement point.

A horizontal rotation drive section 38 and a horizontal angle detection section 39 are provided in the main body section 32 that allows the vertical rotation section 33 to rotate around the horizontal axis. The horizontal rotation drive section 38 rotates the main body section 32 relative to the pedestal 31 around the vertical axis, that is, in the horizontal direction. The horizontal angle detection section 39 detects (angle measurement) the horizontal angle of the collimation direction by detecting the horizontal rotation angle of the main body section 32 with respect to the pedestal 31. For example, the horizontal rotation drive section 38 can be configured with a motor, and the horizontal angle detection section 39 can be configured with an encoder.

Further, the main body portion 32 is provided with a vertical rotation drive portion 41 and a vertical angle detection portion 42. The vertical rotation driving section 41 rotates the vertical rotation section 33 relative to the main body section 32 around the horizontal axis, that is, in the vertical direction. The vertical angle detection section 42 detects (angle measurement) the vertical angle of the collimation direction by detecting the vertical angle of the vertical rotation section 33 with respect to the main body section 32 . For example, the vertical rotation drive section 41 can be configured with a motor, and the vertical angle detection section 42 can be configured with an encoder.

Furthermore, a measuring instrument control section 43 is built into the main body section 32. The measuring instrument control section 43 controls the operation of the laser measuring instrument 13 in an integrated manner using a program stored in a connected storage section 44 . The storage unit 44 is composed of a semiconductor memory and various storage media, and stores programs such as a calculation program necessary for measurement and a data transmission program that generates and transmits information, as well as setting data and points. Group data D1 is stored as appropriate. The information and data are appropriately transmitted to the control mechanism 11 via the communication section 45 described later and the communication section 17 described above (see FIG. 2). The measuring instrument control section 43 includes a measuring instrument display section 35, a measuring instrument operating section 36, a distance measuring optical section 37, a horizontal rotation drive section 38, a horizontal angle detection section 39, a vertical rotation drive section 41, and a vertical angle detection section 42. , a storage section 44 and a communication section 45 are connected.

The communication unit 45 enables communication between the control mechanism 11 (see FIG. 2) and the measuring instrument control unit 43 via the communication unit 17, and allows communication between each unit stored in the storage unit 44 under the control of the measuring instrument control unit 43. Send information accordingly. This communication section 45 enables exchange of data and the like with the control mechanism 11 (communication section 17). The communication unit 45 may communicate with the communication unit 17 by wire via an installed LAN cable, or may communicate with the communication unit 17 wirelessly.

The measurement device control section 43 receives output values for measurement from the distance measuring optical section 37, the horizontal angle detection section 39, and the vertical angle detection section 42. Based on these output values, the measuring device control section 43 determines the arrival time difference or phase difference between the reference light propagating through the reference optical path provided in the main body section 32 and the reflected light acquired via the distance measuring optical section 37. Calculate the distance from to the measurement point (reflection point). Further, the measuring device control unit 43 measures (calculates) the elevation angle and horizontal angle during the calculated distance measurement. Then, the measuring instrument control unit 43 stores the measurement results in the storage unit 44 and transmits them to the control mechanism 11 (communication unit 17) via the communication unit 45 as appropriate.

The measuring instrument control section 43 controls the driving of the horizontal rotation drive section 38 and the vertical rotation drive section 41 to appropriately rotate the main body section 32 and the vertical rotation section 33 (see FIG. 1). can be directed in a predetermined direction and can scan a predetermined range. The measuring instrument control unit 43 of the first embodiment scans the drinking fountain 52, which is the data acquisition area 14, and sets each position of the drinking fountain 52 including the dairy cow 51 there as a measurement point. Since the positional relationship between both laser measuring instruments 13 and the drinking fountain 52 is known in advance and is constant (does not change), the scanning range can be appropriately adjusted to cover the entire area of the drinking fountain 52. When scanning the drinking fountain 52, the measuring device control unit 43 of the first embodiment sets measurement points at intervals of 12.5 mm on the scanning surface. The scanning plane can be appropriately set within the drinking fountain 52, and in the first embodiment, it is set at a position where the body 51a of the dairy cow 51 is assumed to be located.

The measuring device control section 43 controls the distance measuring optical section 37 to scan the drinking fountain 52 and perform distance measurement (distance measurement) at each set measurement point. At this time, the measuring device control unit 43 measures the three-dimensional coordinate position of each measurement point of the drinking fountain 52 by measuring (calculating) the elevation angle and horizontal angle in the collimation direction. Then, the measuring device control unit 43 generates point cloud data D1, which is a collection of three-dimensional coordinate positions (coordinate data) of each measured point of the drinking fountain 52 (one example is shown in FIG. 7, the other example is shown in FIG. 8) is generated and transmitted to the control mechanism 11 via the communication unit 45 and the communication unit 17 as appropriate. The point group data D1 indicates the surface shape of the drinking fountain 52, and if a dairy cow 51 is present in the drinking fountain 52, the surface shape of the dairy cow 51 is also included.

The control mechanism 11 includes an animal detection section 61, a data acquisition section 62, a data synthesis section 63, an animal extraction section 64, and a specific part cutting section 65, as shown in FIG. Specific part alignment section 66, reference point detection section 67, data alignment section 68, normalization section 69, cut plane extraction section 71, calculation data generation section 72, calculation area extraction section 73, approximate curve calculation section 74, and evaluation value calculation 75.

The animal detection unit 61 detects from the image I acquired by the camera 12 that an animal (in Example 1, the dairy cow 51 (see FIG. 4)) is present at the drinking fountain 52, which is the data acquisition area 14. The animal detection unit 61 recognizes various shapes in the image I based on contrast and the like, and discriminates between equipment such as the water trough 53 in the drinking fountain 52 and the dairy cow 51 based on the recognized shapes and the like. Here, since the camera 12 is disposed at a predetermined position, the animal detection unit 61 knows in advance the image of the equipment such as the water trough 53 in the drinking fountain 52. Not only is it easy to distinguish, it is also easy to obtain data indicating the area in which the dairy cow 51 is displayed. When the dairy cow 51 is shown in the image I, the animal detection unit 61 outputs a signal indicating that the dairy cow 51 has been detected at the drinking fountain 52 to the data acquisition unit 62. Therefore, the animal detection unit 61 functions as an animal detection mechanism that detects the presence of an animal in the data acquisition area 14 in cooperation with the camera 12.

In addition, the animal detection unit 61 of Example 1 identifies the detected animals (dairy cows 51) individually, and generates individual identification data D2 indicating the identified information. First, the animal detection unit 61 identifies which dairy cow 51 among the dairy cows kept in the cow shed 50 is shown based on the image I showing the dairy cow 51 . For example, the animal detection unit 61 recognizes the shape and position of the black and white pattern on the dairy cow 51 based on the contrast etc. from the area where the dairy cow 51 in image I is shown, and the recognized black and white pattern is used in each of the pre-registered areas. The dairy cow 51 is identified by comparing it with the black and white pattern of the dairy cow. Then, the animal detection unit 61 generates individual identification data D2 that identifies each dairy cow 51 according to the identification, and outputs the individual identification data D2 to the data acquisition unit 62. The animal detection unit 61 identifies which of the dairy cows kept in the cowshed 50 the dairy cow 51 shown in the image I, that is, the dairy cow 51 currently in the drinking fountain 52 is, and Any other method may be used, such as using tags, as long as it generates individual identification data D2 that individually identifies each dairy cow 51 based on the method, and is not limited to the configuration of the first embodiment.

When the data acquisition unit 62 receives a signal indicating that the dairy cow 51 has been detected from the animal detection unit 61, the data acquisition unit 62 drives each of the two laser measurement devices 13 to scan the drinking fountain 52 (data acquisition area 14). Each laser measuring device 13 acquires point cloud data D1 (see FIGS. 7 and 8) of the drinking fountain 52, associates individual identification data D2 with the point cloud data D1, and outputs the data to the data synthesis section 63. Here, since there is a dairy cow 51 at the drinking fountain 52 when the data acquisition unit 62 is driven as described above, it is assumed that the point cloud data D1 includes the dairy cow 51 indicated by the individual identification data D2. Become.

The data synthesis unit 63 synthesizes the point cloud data D1 acquired by the two laser measuring instruments 13 to generate combined point cloud data D3 (see FIG. 9). Here, since the two laser measurement devices 13 are provided at paired positions with the drinking fountain 52 in between as described above, they measure the same drinking fountain 52 from different directions. Therefore, by combining the respective point cloud data D1 (so-called point cloud composition), different sides of the drinking fountain 52 (for example, one is the right side and the other is the left side of the dairy cow 51 there). It can be a collection of three-dimensional position data including the surface shape of. The data synthesis unit 63 connects overlapping parts of point clouds (so-called point cloud matching), uses targets as landmarks (so-called tie point method), and uses other known techniques. , the point group data D1 are combined to generate combined point group data D3. Here, in the growth condition evaluation system 10, the drinking fountain 52 is set as the data acquisition area 14, and the positional relationship of the laser measuring device 13 with respect to the drinking fountain 52 is known in advance. For this reason, it is easy to determine the portion where the point clouds in the two point cloud data D1 acquired by the respective laser measuring devices 13 overlap, so that the data synthesis unit 63 can appropriately generate the composite point cloud data D3. The data synthesis unit 63 associates the generated composite point group data D3 with the individual identification data D2 and outputs the data to the animal extraction unit 64.

The animal extraction unit 64 extracts only the three-dimensional coordinate position (coordinate data) corresponding to the dairy cow 51 from the composite point cloud data D3 generated by the data synthesis unit 63, thereby extracting an animal that shows the surface shape of the dairy cow 51. Point cloud data D4 (see FIGS. 11 and 12) is generated. Here, in the growth situation evaluation system 10, the drinking fountain 52 is set as the data acquisition area 14, and the positional relationship of the laser measuring device 13 with respect to the drinking fountain 52 is known in advance. The point cloud data of the drinking fountain 52 to be acquired is also known in advance. Therefore, the animal extraction unit 64 calculates the difference between the composite point cloud data D3 and the point cloud data indicating the drinking fountain 52 without the dairy cow 51, thereby obtaining the temporary animal point cloud data D4' (see FIG. 10). ) is generated. Since both laser measurement devices 13 are arranged as described above, this virtual animal point cloud data D4' always includes at least half of the body 51a, regardless of the posture of the dairy cow 51 in the drinking fountain 52. There is. However, the virtual animal point cloud data D4' includes objects other than the dairy cow 51, such as the water trough 53 and a part of the water therein when the dairy cow 51 is drinking water from the water trough 53. , objects that are very close to the dairy cow 51 may be included. As an example, the virtual animal point group data D4' shown in FIG. 10 includes a point group A indicating a dairy cow 51 and a point group B indicating a part of a water pail 53. Therefore, the animal extraction unit 64 performs clustering processing on the temporary animal point group data D4' to generate animal point group data D4. This clustering process groups items that have similar tendencies, positions, characteristics, etc., and the dairy cow 51 and other items can be grouped into separate groups. In the example of FIG. 10, the point indicating the dairy cow 51 is Two groups are generated: group A and point group B showing a part of the water pail 53. The animal extraction unit 64 of the first embodiment divides the pseudo-animal point cloud data D4' into a plurality of groups by clustering the clusters of point cloud data that are close to each other, and extracts the largest cluster among them (point cloud A in the example of FIG. 10). is extracted as the dairy cow 51, and animal point cloud data D4 (see FIGS. 11 and 12) is generated. As a result, the animal point cloud data D4 always includes at least half of the torso 51a, and objects other than the dairy cow 51 are removed. The animal extraction unit 64 associates the generated animal point cloud data D4 with the individual identification data D2 and outputs the data to the specific region extraction unit 65.

Note that the animal extraction unit 64 may have any other configuration as long as it generates the animal point cloud data D4, and is not limited to the configuration of the first embodiment. For example, the animal extraction unit 64 performs a clustering process on the pseudo animal point cloud data D4' generated as described above, when there is a low possibility that it contains animals other than the target animal (dairy cow 51). Alternatively, the temporary animal point group data D4' may be used as the animal point group data D4. In addition, the animal extraction unit 64 registers in advance three-dimensional coordinate positions (coordinate data) indicating the dairy cows 51 at various angles, sizes, and types, and extracts the animal point group data D4 by comparing them. good. Furthermore, the animal extraction unit 64 extracts a clean plane such as the floor or wall of the drinking fountain 52 from the composite point cloud data D3, and performs threshold processing on a place where the dairy cow 51 is likely to be found using the plane as a reference. The temporary animal point group data D4' and the animal point group data D4 may be generated by extracting locations where there is a high possibility that the animal 51 is present.

The specific region cutting unit 65 removes the vicinity of the head 51b and the tail 51c from the animal point group data D4 (see FIG. 12) representing the dairy cow 51 to generate specific region data D5 (see FIG. 13). In the present disclosure, since it is considered that the vicinity of the body 51a of the dairy cow 51, particularly the vicinity of the hook bone (see reference numeral Bh in FIG. 13), is important in evaluating the breeding status of the dairy cow 51, the animal point cloud data D4 is Specific part data D5 is generated from which the vicinity of the head 51b and the tail 51c are removed. The specific part extraction unit 65 of the first embodiment extracts a predetermined number of points (10,000 points as an example in the first embodiment) from the upper side of the animal point cloud data D4 in the vertical direction, and The first principal component is the direction in which the spine 51d extends (hereinafter referred to as the spine direction Ds (see FIG. 14)). This first principal component has its axis in the direction where the variance is greatest, and indicates the direction in which the extracted upper point group extends. Note that the specific part cutting unit 65 registers in advance three-dimensional coordinate positions (coordinate data) indicating various angles, sizes, and types of the spine 51d, and extracts the spine direction Ds by comparing with these. or other methods may be used. Then, the specific part cutting unit 65 determines that the thinner part at one end in the spinal direction Ds is near the head 51b (including the neck), and also determines that the narrower part at one end in the spinal direction Ds is near the head 51b (including the neck), and the other end in the longitudinal direction. The extremely thin and long part is determined to be the tail 51c. In addition, if the specific part cutting unit 65 is used to determine the head vicinity 51b and the tail 51c in the animal point cloud data D4, for example, the head vicinity 51b can be determined by comparing with the shape and position registered in advance. or the tail 51c, or other methods may be used, and the method is not limited to the method of the first embodiment. The specific part cutting unit 65 associates the generated specific part data D5 with the individual identification data D2 and outputs the data to the specific part aligning unit 66.

The specific region alignment unit 66 aligns the specific region data D5 generated by the specific region cutting unit 65 so that the spine direction Ds coincides with the reference axis direction Db (see FIG. 14). The reference axis direction Db is assumed to extend along the horizontal plane, and the direction along the horizontal plane that is orthogonal to the reference axis direction Db is defined as the width direction Dw, and the direction orthogonal to the horizontal plane is defined as the vertical direction Dh. The specific region alignment unit 66 of the first embodiment moves the specific region data D5 on the horizontal plane so that the spine direction Ds coincides with the reference axis direction Db (see arrow A1). Note that the specific part alignment unit 66 is not limited to the configuration of the first embodiment as long as it aligns the spine direction Ds in the specific part data D5 to match the reference axis direction Db. The specific part alignment unit 66 outputs specific part data D5 with the spine direction Ds aligned with the reference axis direction Db to the reference part detection unit 67.

The reference location detection unit 67 detects a reference location Pr (see FIG. 15) in the specific region data D5 in which the spine direction Ds is matched with the reference axial direction Db. This reference point Pr is used as a reference for alignment by the data alignment unit 68 and a reference for normalization by the normalization unit 69, and is a point that is easy to detect regardless of individual differences and has common characteristics. Set. In the first embodiment, as shown in FIG. 15, a straight line connecting two hook bones (Bh) that form a pair across the reference axis direction Db (backbone 51d) is defined as the reference point Pr. The reference point detection unit 67 first detects the positions of a pair of hook bones. The reference point detection unit 67 of the first embodiment registers in advance three-dimensional coordinate positions (coordinate data) indicating the vicinity of the hook bone (Bh) in the dairy cows 51 of various angles, sizes, and types, and compares them with the three-dimensional coordinate positions (coordinate data). Extract the position of hook bone (Bh) by Note that the reference point detection unit 67 may use any other method as long as it extracts the position of the hook bone (Bh) in the aligned specific part data D5, and is not limited to the configuration of the first embodiment. Thereafter, the reference point detection unit 67 connects the detected pair of hook bones (Bh) (their positions) to generate a reference point Pr (their data). Then, the reference location detecting section 67 outputs the reference location Pr (the data) detected in the specific region data D5 in which the direction in which the spine 51d extends coincides with the reference axis direction Db to the data alignment section 68. Note that the reference point Pr may be set as appropriate as long as it is a point that is easy to detect and has common characteristics regardless of individual differences in the animals whose breeding status is to be evaluated. Not limited.

The data alignment unit 68 generates specific part alignment data D6 (see FIG. 16) by adjusting the position of the specific part data D5 with respect to the reference axis direction Db using the detected reference point Pr as a reference. The data alignment unit 68 generates specific part alignment data D6 (see FIG. 16) by finely adjusting the inclination of the specific part data D5 on the horizontal plane so that the reference part Pr is perpendicular to the reference axis direction Db. Furthermore, the data alignment unit 68 of the first embodiment aligns the direction of the specific part data D5 by positioning the cut end 51e on the side of the tail 51c at the origin in the reference axis direction Db, thereby aligning the specific part data D6. generate. At this time, if the position of the reference point Pr in the reference axis direction Db exceeds half of the maximum value of the specific part data D5 in the reference axis direction Db, the data alignment unit 68 It is determined that the side is directed toward the origin, and the specific part data D5 is rotated 180 degrees to position the end 51e on the tail 51c side at the origin in the reference axis direction Db. Note that the data alignment unit 68 may position the cut end on the side near the head 51b at the origin in the reference axis direction Db, and is not limited to the configuration of the first embodiment. The data alignment section 68 outputs the specific part alignment data D6 to the normalization section 69.

The normalization unit 69 generates normalized data D7 (see FIG. 18) by normalizing the specific part alignment data D6 using the above-described reference position as a reference. This is because animals (dairy cows 51) have different skeletons and fleshiness, so in the specific part alignment data D6, as shown in FIG. 17, even if they are aligned as described above, the position in the vertical direction Dh is In addition to being different, the size of the vertical direction Dh is also different. The normalization unit 69 of the first embodiment uses the reference point Pr (positions of both hook bones) as a standard for normalization, and makes sure that the above-mentioned reference point Pr in each specific part alignment data D6 has a uniform size. , each normalized data D7 (see FIG. 18) is generated by expanding or contracting and adjusting each specific part alignment data D6. This is because, in the present disclosure, the growth status of each dairy cow is determined not by the size of the skeleton including the hook bone (Bh) but by the amount of meat. Note that the normalization unit 69 may appropriately set a location to be used as a standard for normalization, and is not limited to the configuration of the first embodiment. The normalization section 69 outputs the normalized data D7 to the cut plane extraction section 71.

The cutting plane extraction unit 71 generates cross-sectional data D8 (see FIG. 19), which is a cross-section of the dairy cow 51 (its body) cut along a predetermined plane, from the normalized data D7. The cut plane extraction unit 71 of the first embodiment assumes that a predetermined plane is orthogonal to the front-back direction of the dairy cow 51, and cuts the normalized data D7 along a slice position Sp set at an arbitrary position in the front-back direction. By doing so, cross-sectional data D8 is generated. This slice position Sp may be set as appropriate, but in Example 1, it is set at four locations shown in FIG. There is. The slice positions Sp1 to Sp3 are planes perpendicular to the reference axis direction Db. Further, the slice position Sp4 is a plane perpendicular to the width direction Dw.

The slice position Sp1 is an intermediate position in the reference axis direction Db between the hook bone position (reference point Pr) and the end 51e on the tail 51c side (the origin in the reference axis direction Db). The slice position Sp2 is the position of a reference point Pr (hook bone) in the reference axis direction Db. The slice position Sp3 is set to be 1.5 times the distance from the end 51e (origin of the reference axis direction Db) to the slice position Sp2 in the reference axis direction Db. The slice position Sp4 is an intermediate position between the position of one hook bone and the reference axis direction Db in the width direction Dw.

An example of the cross-sectional data D8 is shown in FIG. 19. In FIG. 19, the slice positions Sp are arranged in order from the left side when viewed from the front, corresponding to the numbers at the end of each slice position Sp, and numbers 1 to 4 are attached to the ends. That is, the cross-section data D81 corresponds to the slice position Sp1, the cross-section data D82 corresponds to the slice position Sp2, the cross-section data D83 corresponds to the slice position Sp3, and the cross-section data D84 corresponds to the slice position Sp4. The cut plane extracting unit 71 associates the generated cross-sectional data D8 with the individual identification data D2 and outputs it to the calculation data generating unit 72. Note that the cut plane extraction unit 71 may appropriately set the number and position of the cross section data D8 to be generated for each normalized data D7, and is not limited to the configuration of the first embodiment.

The calculation data generation unit 72 generates calculation data D9 (see FIG. 20) for use in calculation from the cross-sectional data D8. An example of this will be explained using cross-sectional data D82 (hereinafter simply referred to as cross-sectional data D8) corresponding to slice position Sp2 in FIG. 19. First, the calculation data generation unit 72 interpolates the point group of the cross-sectional data D8 (see FIG. 19) using the average value, thereby generating thinned-out cross-sectional data D8' (left side in FIG. 20) for thinning out the number of point groups. generate. In addition, the calculation data generation unit 72 generates a thinned cross section so that the position indicating the spine 51d in the thinned cross section data D8' is located at the origin (center position) in the cutting plane direction (width direction Dw in the example shown in FIG. 20). The data D8' is displaced in the direction of the cutting surface (width direction Dw) (see arrow A2). Then, the calculation data generation unit 72 compares the number and distribution of point groups located on both sides of the origin (center position) in the cross-section direction (width direction Dw) in the thinned-out cross-section data D8'. Select the side with more groups or the side with fewer gaps in the distribution in the cross-section direction (width direction Dw), copy the data (point group) on the selected side to the opposite side, and create calculation data D9. (See right side of FIG. 20). Thereby, the calculation data D9 can reduce the number of missing data points in the width direction Dw while suppressing an increase in the amount of data, and has a symmetrical shape (distribution) in the width direction Dw.

Here, in the cross-sectional data D8, although the original animal point cloud data D4 represents the surface shape of the dairy cow 51 as a very vivid point cloud, the points do not necessarily cover the entire cross section at a predetermined position. There is a possibility that the group is not dispersed. In the cross-sectional data D8, it is unlikely that the point group is largely missing on both sides in the cross-sectional direction. For this reason, the calculation data D9 can suppress an increase in the amount of data while having a point group necessary for obtaining an approximate curve for evaluating the growth status.

Furthermore, when the plane perpendicular to the width direction Dw is set as the slice position Sp, the calculation data generation unit 72 generates the calculation data D9 as follows. An example of this will be explained using cross-sectional data D84 corresponding to slice position Sp4 in FIG. 19. First, the calculation data generation unit 72 interpolates the point group of the cross-sectional data D84 (see FIG. 19) using the average value, thereby generating thinned-out cross-sectional data D84' (left side in FIG. 21) for thinning out the number of point groups. generate. In addition, the calculation data generation unit 72 determines that the right end (the highest position in the example shown in FIG. 21) of the thinned cross-section data D84' is the origin (center position) in the cutting plane direction (the reference axis direction Db in the example shown in FIG. 21). The thinned cross-section data D84' is displaced in the cutting plane direction (reference axis direction Db) so that the thinned-out cross-section data D84' is located at the same position (see arrow A3). Then, since the thinned cross-section data D84' has a plane perpendicular to the width direction Dw as the slice position Sp, a point group exists only on one side of the cut plane direction (reference axis direction Db), and the other side is the empty part of the distribution. is on the smaller side. Therefore, the calculation data generation unit 72 copies the point group (the data) existing on one side in the thinned cross-section data D84' to the opposite side, and calculates the calculation data D9 (the right side in FIG. 21). generate. Note that the calculation data D9 is twice the size of the thinned cross-section data D84' in the direction of the cut plane (reference axis direction Db), but in FIG. 21, the scale in the direction of the cut plane (reference axis direction Db) is changed. It shows.

An example of the calculation data D9 generated in this way is shown in FIG. 22. In FIG. 22, the slice positions Sp are arranged in order from the left side when viewed from the front, corresponding to the numbers at the end of each slice position Sp, and numbers 1 to 4 are attached to the ends. That is, the calculation data D91 corresponds to the cross-section data D81, the calculation data D92 corresponds to the cross-section data D82, the calculation data D93 corresponds to the cross-section data D83, and the calculation data D94 corresponds to the cross-section data D84. The calculation data generation unit 72 associates the generated calculation data D9 with the individual identification data D2 and outputs the generated calculation data D9 to the calculation area extraction unit 73. At this time, when the calculation data generation unit 72 generates a plurality of calculation data D9 for the single individual identification data D2, the calculation data generation unit 72 outputs them to the calculation area extraction unit 73 in association with the slice position Sp. do.

The calculation area extraction unit 73 extracts a starting point Ps and an ending point Pe for calculating an approximate curve Ac, which will be described later, in the curve indicated by the arrangement of points in the calculation data D9. Here, the present applicant has found that when looking at the vicinity of the hook bone of the dairy cow 51 in a cross section orthogonal to the reference axis direction Db, there are roughly convex portions, concave portions, and convex portions from the end, regardless of the position in the reference axis direction Db. , we focused on the fact that there is a tendency to form a curved line that has concave parts and convex parts, with the spine 51d located in the central convex part. Therefore, the applicant thought that by applying a sixth-order Bezier curve using seven control points, a smooth approximated curve Ac that matches the calculation data D9 can be calculated. Therefore, in order to appropriately approximate the calculation data D9 with a sixth-order Bezier curve, the calculation area extraction unit 73 extracts a convex shape at both ends of the external shape (outline of the dairy cow 51) indicated by the calculation data D9, that is, at the center. One of the end points (points of inflection) of the convex part on the outside of the part is extracted as the starting point Ps, and the other as the ending point Pe (see FIG. 22). Then, the calculation area extraction unit 73 outputs the starting point Ps and the ending point Pe in the calculation data D9 to the approximate curve calculation unit 74.

In addition, the present applicant has proposed that when the slice position Sp is a plane orthogonal to the width direction Dw in the vicinity of the hook bone of the dairy cow 51, convex portions, concave portions, convex portions, etc. We also focused on the fact that there is a tendency for the shape to be a curved line. For this reason, as described above, when the plane perpendicular to the width direction Dw is set as the slice position Sp, the calculation data generation unit 72 creates a The thinned cross-sectional data D84' is displaced in the direction of the cut plane and is copied to the opposite side to generate the calculation data D9. As a result, even if the calculation data D9 has the plane perpendicular to the width direction Dw as the slice position Sp, it can be made into a curve that draws a convex portion, a concave portion, a convex portion, a concave portion, and a convex portion from the end. Therefore, even if the calculation data D9 has the plane orthogonal to the width direction Dw as the slice position Sp, the calculation area extraction unit 73 can calculate the external shape (outline of the dairy cow 51) indicated by the calculation data D9 in the same way as described above. One of both ends is set as the starting point Ps, and the other end is extracted as the ending point Pe, and output to the approximate curve calculating section 74.

The approximate curve calculation unit 74 uses the starting point Ps and the ending point Pe extracted by the calculation area extraction unit 73 to calculate an approximate curve Ac (see FIGS. 23 and 24) that matches the calculation data D9. As shown in FIG. 23, in order to appropriately approximate each calculation data D9 with a sixth-order Bezier curve, the present applicant has determined that both ends of the external shape (outline of the dairy cow 51) indicated by the calculation data D9 The starting point Ps and the ending point Pe are set as fixed control points C, and five control points C are set between the starting point Ps and the ending point Pe. In the following, seven control points C will be set including the starting point Ps and the ending point Pe, and when indicated individually, numbers 1 to 7 will be added to the beginning starting from the starting point Ps, and numbers 1 to 7 will be added to the end. A number will be attached. That is, when shown individually, the starting point Ps becomes the first control point C1, the next one becomes the second control point C2, the third control point C3 to the sixth control point C6 start from the next one, and the end point Pe becomes the second control point C2. 7 control point C7. Then, the present applicant provides the second control point C2 next to the first control point C1 fixed at the starting point Ps on the upper side when viewed in the vertical direction Dh with respect to the calculation data D9, and sequentially from the next point The third control point C3 is provided on the lower side, the fourth control point C4 is provided on the upper side, the fifth control point C5 is provided on the lower side, and the sixth control point C6 is provided on the upper side. Then, excluding the fixed first control point C1 (start point Ps) and seventh control point C7 (end point Pe), the remaining five second control points C2 to sixth control point C6 are moved as appropriate on the cutting surface. By doing so, an approximate curve Ac that fits the calculation data D9 is calculated.

Therefore, the approximate curve calculation unit 74 sets the starting point Ps and the ending point Pe in the calculation data D9 as the fixed first control point C1 and the seventh control point C7, and also sets five control points (the first control point C7) between them. The second control point C2 to the sixth control point C6) are set. Then, the approximate curve calculation unit 74 calculates the five curves so that the curve from the starting point Ps (first control point C1) to the ending point Pe (seventh control point C7) overlaps with the calculation data D9 (the external shape indicated by it). An approximate curve Ac is calculated by adjusting the positions of the control points (second control point C2 to sixth control point C6).

Here, in Example 1, the calculation data D9 is generated by selecting one of the cutting plane directions in the cross-sectional data D8 and reversing it. Therefore, in the calculation data D9, regarding the line segment (hereinafter referred to as center line Lc) extending in the vertical direction Dh passing through the intermediate position on the cutting plane, the starting point Ps (first control point C1) and the ending point Pe (first control point C1) The seventh control point C7), the second control point C2 and the sixth control point C6, and the third control point C3 and the fifth control point C5 have a line-symmetrical positional relationship. In the calculation data D9, the fourth control point C4 is provided on the center line Lc on the cut plane. From these facts, the approximate curve calculation unit 74 of the first embodiment calculates the position of either the second control point C2 or the sixth control point C6, the third control point C3, or the fifth control point on the cutting plane. The position of the fifth control point C5 on the center line Lc is adjusted. Then, the approximate curve calculation unit 74 of the first embodiment calculates the other position of the second control point C2 and the sixth control point C6 from one position, and also calculates the position of the other of the second control point C2 and the sixth control point C6. Find the other position from one position. Therefore, the approximate curve calculation unit 74 of the first embodiment substantially adjusts the positions of the three control points C so that the starting point Ps( The approximate curve Ac is calculated by adjusting the way the Bezier curve curves from the first control point C1) to the end point Pe (seventh control point C7).

In this way, the approximate curve calculation unit 74 standardizes the method of obtaining the approximate curve Ac. That is, the calculation area extraction unit 73 extracts the starting point Ps and the ending point Pe in the calculation data D9, and the approximate curve calculating unit 74 sets a total of seven control points C including the starting point Ps and the ending point Pe. Then, the approximate curve calculation unit 74 fixes the starting point Ps (first control point C1) and the ending point Pe (seventh control point C7), and adjusts the remaining five control points C to move from the starting point Ps. An approximated curve Ac is obtained as a sixth-order Bezier curve leading to the end point Pe. In this way, the approximate curve calculation unit 74 adjusts the positions of the seven control points C based on the common characteristics as described above, thereby adjusting the individual differences in the calculation data D9 (dairy cow 51). Regardless, the external shape can be represented by an approximate curve Ac. As a result, the approximate curve calculation unit 74 calculates the calculation data D9, that is, the original cross section (cross section data D8 ), and the coefficients in each approximate curve Ac and the position of each control point C can be changed according to individual differences among dairy cows 51.

An example of the approximate curve Ac obtained in this way is shown in FIG. In FIG. 24, three different calculation data D9 and an approximate curve Ac corresponding thereto are shown side by side. FIG. 24 also shows the positions of seven control points C including the starting point Ps and the ending point Pe for each approximate curve Ac. The approximate curve calculating section 74 outputs an approximate curve Ac (its data) based on a sixth-order Bezier curve including seven control points C to the evaluation value calculating section 75 in association with the individual identification data D2.

The evaluation value calculation unit 75 calculates an evaluation value Ev indicating the evaluation of the breeding status of each dairy cow 51 in the drinking fountain 52 based on each approximate curve Ac from the approximate curve calculation unit 74. The evaluation value calculation unit 75 calculates the evaluation value Ev using the positional relationship of each control point C on the approximate curve Ac. The positional relationship between point C3 (fifth control point C5) and fourth control point C4 is used. This will be explained using FIG. 23. First, as shown in FIG. 23, the interval (its size) between the second control point C2 and the third control point C3 is defined as a first interval a, and the interval between the third control point C3 and the fourth control point C4 ( The distance between the second control point C2 and the fourth control point C4 is defined as a third distance c. Further, as shown in FIG. 23, the interval (its size) between the sixth control point C6 and the fifth control point C5 is a fourth interval d, and the interval (the size) between the fifth control point C5 and the fourth control point C4 ( The distance between the sixth control point C6 and the fourth control point C4 is defined as a sixth distance f.

Here, in the dairy cow 51, the more fleshy the various bones such as the backbone 51d and the hook bone become less noticeable, that is, the unevenness becomes rounded as a whole, and if the dairy cow 51 is not fleshy, the various bones appear angular. It becomes like this. Further, the beefier the dairy cow 51 is, the more the cow 51 has a shape that swells outward. When this is applied to the calculation data D9, the difference in the vertical direction Dh between the second control point C2 (sixth control point C6) and the fourth control point C4 increases, that is, the third interval c (sixth interval As f) becomes larger, the corners of the unevenness become tighter, and there is a tendency for the shape to become less fleshy. For this reason, it is desirable that the evaluation value Ev becomes less solid, that is, the evaluation becomes lower, as the third interval c (sixth interval f) becomes larger. Also, when applied to the calculation data D9, the second control point C2 (sixth control point C6) and the third control point C3 (fifth control point C5) are separated, that is, the first interval a (fourth interval d) As the size increases, the overall shape tends to swell outward and become more fleshy. Therefore, it is desirable that the evaluation value Ev becomes thicker and higher as the first interval a (fourth interval d) becomes larger. In order to satisfy the above two conditions, the second control point C2 (sixth control point C6) and the fourth control point C4 will be separated, and the second interval b (fifth interval e) will become larger. A trend can be seen. From these facts, the evaluation value Ev of the present disclosure is calculated by dividing the third interval c by the sum of the first interval a and the second interval b (evaluation value Ev = third interval c/(first interval a + second interval b). 2 intervals b)), and the larger the distance, the lower the evaluation (poor fleshiness), and the smaller the difference, the higher the evaluation (good fleshiness).

Similarly, the evaluation value Ev of the present disclosure is calculated by dividing the sixth interval f by the sum of the fourth interval d and the fifth interval e (evaluation value Ev=sixth interval f/(fourth interval d+fifth interval e)). Here, the calculation data D9 of Example 1 is based on the starting point Ps (first control point C1), the end point Pe (seventh control point C7), and the second control point C2 with respect to the center line Lc on the cutting plane. The sixth control point C6, the third control point C3, and the fifth control point C5 have a line-symmetrical positional relationship. Therefore, the first interval a and the fourth interval d become equal, the second interval b and the fifth interval e become equal, and the third interval c and the sixth interval f become equal. Therefore, in the first embodiment, the evaluation value Ev calculated using the first interval a, the second interval b, and the third interval c, and the fourth interval d, the fifth interval e, and the sixth interval f are used. Since the evaluation value Ev and the calculated evaluation value Ev are equal to each other, it is sufficient to calculate either one of them.

From this, the evaluation value calculation unit 75 of the present disclosure calculates the approximate curve Ac, that is, the second control point C2 (sixth control point C6), the third control point C3 (fifth control point C5), and the fourth control point. Based on the position (coordinate data) with C4, calculate the above-described first interval a (fourth interval d), second interval b (fifth interval e), and third interval c (sixth interval f). . Then, the evaluation value calculation unit 75 applies the above calculation formula (third interval c/(first interval a+second interval b)) or (sixth interval f/(fourth interval d+fifth interval e)). By fitting, an evaluation value Ev is calculated. The evaluation value calculation unit 75 associates the evaluation value Ev of the dairy cow 51 with the individual identification data D2 and stores it in the storage unit 18 as appropriate.

The control mechanism 11 causes the evaluation value Ev stored in the storage unit 18 to be displayed on the display unit 16 or output to an external device via the communication unit 17 as appropriate. Here, since the evaluation value Ev is associated with the individual identification data D2, it is possible to easily understand which dairy cow 51 the breeding status indicates. Thereby, the control mechanism 11 can notify the evaluation value Ev of the dairy cow 51.

Next, a growing situation evaluation process (a growing situation evaluation control method) as an example of evaluating the growing situation of the dairy cow 51 using the growing situation evaluation system 10 will be described using FIG. 25. This growth status evaluation process is executed by the control mechanism 11 based on a program stored in the storage unit 18 or the internal memory 11a. Each step (each process) of the flowchart of FIG. 25 will be explained below. The flowchart of FIG. 25 is started when the growth condition evaluation system 10 is activated, the browser or application is launched, the camera 12 is driven, and both laser measuring devices 13 are placed in a standby state.

In step S1, it is determined whether or not the dairy cow 51 exists in the drinking fountain 52 (data acquisition area 14). If YES, proceed to step S2; if NO, step S1 is repeated. In step S1, the animal detection unit 61 analyzes the image I acquired by the camera 12 to determine whether or not there is a dairy cow 51 at the drinking fountain 52, and if there is a dairy cow 51, it sends a signal to that effect as data. It is output to the acquisition unit 62. In addition, in step S1 of the first embodiment, when a dairy cow 51 is detected, individual identification data D2 that identifies each dairy cow 51 is generated, and the individual identification data D2 is output to the data acquisition unit 62.

In step S2, point cloud data D1 of the drinking fountain 52 is acquired, and the process proceeds to step S3. In step S2, when the data acquisition unit 62 receives a signal indicating that the dairy cow 51 has been detected from the animal detection unit 61, the data acquisition unit 62 drives the two laser measurement devices 13 to scan the drinking fountain 52 (data acquisition area 14). , the point cloud data D1 of the drinking fountain 52 is acquired. Then, in step S2, upon receiving the point cloud data D1 from both laser measuring devices 13, the data acquisition section 62 associates the individual identification data D2 with each point cloud data D1 and outputs the data to the data synthesis section 63.

In step S3, composite point group data D3 is generated, and the process proceeds to step S4. In step S3, the data synthesis unit 63 synthesizes the point cloud data D1 acquired by both laser measuring devices 13, generates composite point cloud data D3, and adds individual identification data D2 to the generated composite point cloud data D3. The data is associated and output to the animal extraction unit 64.

In step S4, animal point cloud data D4 is generated, and the process proceeds to step S5. In step S4, the animal extraction unit 64 extracts only the three-dimensional coordinate position (coordinate data) corresponding to the dairy cow 51 from the composite point cloud data D3 generated by the data synthesis unit 63 to generate animal point cloud data D4, The generated animal point cloud data D4 is associated with the individual identification data D2 and outputted to the specific part cutting section 65.

In step S5, specific part data D5 is generated, and the process proceeds to step S6. In step S6, the specific part cutting unit 65 generates specific part data D5 by removing the vicinity of the head 51b and the tail 51c of the dairy cow 51 from the animal point cloud data D4 generated by the animal extraction unit 64. The specific part data D5 is associated with the individual identification data D2 and outputted to the reference part detection section 67.

In step S6, the specific part data D5 is arranged, and the process proceeds to step S7. In step S6, the specific part alignment unit 66 extracts the first principal component in the specific part data D5 generated by the specific part cutting unit 65 as the spine 51d of the dairy cow 51, and extracts the first principal component (in the direction in which the spine 51d extends). ) on the horizontal plane so as to match the reference axis direction Db. Then, in step S6, the specific part alignment unit 66 outputs to the reference part detection unit 67 the specific part data D5 in which the spine 51d is aligned so that the direction in which it extends coincides with the reference axis direction Db.

In step S7, the position of the reference point Pr in the specific part data D5 is detected, and the process proceeds to step S8. In step S7, the reference point detection unit 67 extracts the hook bone positions from the specific part data D5 aligned by the specific part alignment unit 66, and generates a reference point Pr that connects the extracted hook bone positions. Then, the reference point Pr (that data) is output to the data alignment section 68.

In step S8, specific part alignment data D6 is generated, and the process proceeds to step S9. In step S8, the specific part alignment data D6 is generated by finely adjusting the inclination of the specific part data D5 on the horizontal plane so that the reference part Pr detected by the reference part detection section 67 is orthogonal to the reference axis direction Db. Then, in step S8, the data alignment unit 68 outputs the generated specific part alignment data D6 to the normalization unit 69.

In step S9, normalized data D7 is generated, and the process proceeds to step S10. In step S9, the normalization unit 69 enlarges or reduces each specific part alignment data D6 so that the reference points Pr in each specific part alignment data D6 generated by the data alignment unit 68 have a uniform size. generated data D7. Then, in step S9, the normalization unit 69 outputs the generated normalized data D7 to the cut plane extraction unit 71.

In step S10, cross-sectional data D8 is generated, and the process proceeds to step S11. In step S10, the cut plane extraction unit 71 cuts the normalized data D7 generated by the normalization unit 69 along the slice position Sp set at an arbitrary position to generate cross-section data D8. Then, in step S10, the cut plane extraction section 71 outputs each generated section data D8 to the calculation data generation section 72 in association with the individual identification data D2.

In step S11, calculation data D9 is generated, and the process proceeds to step S12. In step S11, the calculation data generation unit 72 generates thinned-out cross-sectional data D8' by thinning out the number of points in the cross-sectional data D8 generated by the cross-sectional plane extraction unit 71, and generates thinned-out cross-sectional data D8' in which there is less space in the distribution in the direction of the cut plane. One side is selected and copied to the other side to generate calculation data D9. Then, in step S11, the calculation data generation section 72 outputs the generated calculation data D9 to the calculation area extraction section 73.

In step S12, the starting point Ps and ending point Pe in the calculation data D9 are extracted, and the process proceeds to step S13. In step S12, the calculation area extraction unit 73 extracts the starting point Ps and the ending point Pe for calculating the approximate curve Ac from the calculation data D9, and approximates the positions (the data) of the starting point Ps and the ending point Pe. It is output to the curve calculation section 74.

In step S13, an approximate curve Ac is calculated, and the process proceeds to step S14. In step S13, the approximate curve calculation unit 74 sets the starting point Ps and the ending point Pe as fixed control points C for the calculation data D9, and also sets five control points C between them. The approximate curve Ac is calculated by adjusting the positions of the five control points C. In step S13 of the first embodiment, the approximate curve calculation unit 74 calculates the position of one of the second control point C2 and the sixth control point C6, the third control point C3 and the fifth control point on the cutting plane. An approximate curve that matches the calculation data D9 (contour on the outer surface side) by adjusting the position of either one of the control points C5 and the position of the fifth control point C5 on the center line Lc. Calculate Ac. Then, in step S13, the approximate curve calculation section 74 outputs the calculated approximate curve Ac to the evaluation value calculation section 75 in association with the individual identification data D2 together with each control point C (its data).

In step S14, an evaluation value Ev is calculated, and the process proceeds to step S15. In step S14, the evaluation value calculation unit 75 calculates an evaluation value Ev indicating the evaluation of the breeding status of the dairy cow 51 that was in the drinking fountain 52 based on each approximate curve Ac from the approximate curve calculation unit 74, and The individual identification data D2 is associated with Ev and stored in the storage unit 18 as appropriate.

In step S15, it is determined whether or not the growing situation evaluation process has been completed. If YES, this growing situation evaluation process is ended, and if NO, the process returns to step S1. In step S15, when the growth situation evaluation system 10 is stopped or an operation to end the growth situation evaluation process is performed on the operation unit 15, it is determined that the growth situation evaluation process is ended.

Next, a description will be given of how the growth situation is evaluated using the growth situation evaluation system 10.
When the growth status evaluation system 10 detects that the dairy cow 51 is present at the drinking fountain 52 via the camera 12, it acquires point cloud data D1 of the drinking fountain 52 using both laser measuring devices 13 (steps S1 and S2). Thereafter, the growth situation evaluation system 10 generates each specific part data D5 based on each point cloud data D1 (steps S3 to S5), and arranges each of the specific part data D5 to generate specific part alignment data D6 ( Steps S6 to S8). Then, the growth condition evaluation system 10 generates each normalized data D7 based on the reference point Pr from each specific part alignment data D6 (step S9), thereby evaluating the state of fatness of the dairy cow 51 regardless of its size. Each section data D8 is generated from each normalized data D7 (step S10).

After that, the growth situation evaluation system 10 interpolates the point group of each cross-section data D8 using the average value to generate thinned-out cross-section data D8', selects the side with fewer gaps in the distribution in the direction of the cross-section, and copies it to the other side. (copy) to generate calculation data D9 (step S11). Thereby, the growth condition evaluation system 10 can appropriately judge the state of fleshiness even when one of the point clouds in the direction of the cut plane is largely missing, and can suppress the amount of data.

Thereafter, the growth condition evaluation system 10 applies a sixth-order Bezier curve in order to calculate a smooth approximate curve Ac that fits the calculation data D9. For this purpose, the growth situation evaluation system 10 determines the starting point Ps and the ending point Pe, which are both ends of the external shape (outline of the dairy cow 51) that becomes the convex part, concave part, convex part, concave part, and convex part indicated by each calculation data D9. is set as a fixed control point C, and five control points C are set between the starting point Ps and the ending point Pe (steps S12, S13). The growth situation evaluation system 10 then determines the position of one of the second control point C2 and the sixth control point C6, and the position of one of the third control point C3 and the fifth control point C5. At the same time, the position of the fifth control point C5 on the center line Lc is adjusted to calculate an approximate curve Ac that matches the calculation data D9 (outer shape) (step S13). That is, the growth situation evaluation system 10 calculates the approximate curve Ac that fits the calculation data D9 by adjusting the positions of the five control points C between the starting point Ps and the ending point Pe. Thereafter, the growth situation evaluation system 10 determines the position of the second control point C2 (sixth control point C6), the position of the third control point C3 and (fifth control point C5), and the position of the fifth control point C5. An evaluation value Ev is calculated based on , (step S14) and stored in the storage unit 18 as appropriate. The growth status evaluation system 10 can notify the evaluation value Ev by appropriately displaying the evaluation value Ev on the display unit 16 or outputting the evaluation value Ev to an external device via the communication unit 17 as appropriate.

Here, the evaluation value Ev obtained as described above using the growth condition evaluation system 10 for the actual dairy cow 51 will be explained with reference to FIG. 26. First, as a visual indicator for evaluating growth conditions, a cross-section (hereinafter referred to as a theoretical cross-section Dt) is exemplified in the literature (Journal of Dairy Science Vol. 72, No. 1, 1989). FIG. 26 shows a table summarizing the relationship between the theoretical cross section Dt and each value obtained by the growth condition evaluation system 10. Here, it is not easy to find an individual that matches the theoretical cross section Dt from among the actual dairy cows 51. For this reason, FIG. 26 shows two cross-sections whose shape is close to the theoretical cross-section Dt exemplified in the above-mentioned literature from the acquired data of many dairy cows 51. One of them is close to the definite depression example in the literature, and the other is close to the slight depression example in the literature. For this reason, in FIG. 26, the left column is an example with less flesh, and the right column is an example with good flesh. In FIG. 26, the two theoretical cross sections Dt described above are shown in the top row. In the following, each data of an example that is not fleshy is indicated with a suffix a, and each data of a fleshy example is indicated with a suffix b. That is, an example with a poor thickness on the left side is defined as a theoretical cross section Dta, and an example with a good thickness on the right side is defined as a theoretical cross section Dtb. This also applies to each data in the row below the theoretical cross section Dt.

Furthermore, in FIG. 26, actual cross-sectional data Da of the dairy cow 51 is shown in the row below the theoretical cross-section Dt as data of the dairy cow 51 that was actually obtained. This actual cross-section data Da is created as data indicating a cross-section from the above-mentioned animal point group data D4 (specific part data D5). This is because it is difficult to obtain a cross section of the actual dairy cow 51. Here, the animal point cloud data D4 (specific site data D5) indicates a three-dimensional coordinate position (coordinate data) corresponding to the dairy cow 51 in the acquired composite point cloud data D3. Therefore, it is considered that the actual cross-sectional data Da depicts the external shape of the cross-section of the actual dairy cow 51 extremely accurately. Therefore, in FIG. 26, the actual cross-section data Da is used as the data of the dairy cow 51, and the left side shows the actual cross-section data Daa of the dairy cow 51 with a shape close to the theoretical cross-section Dta, and the right side shows the actual cross-section data Daa of the dairy cow 51 with a shape close to the theoretical cross-section Dtb. Actual cross-sectional data Dab of a dairy cow 51 is shown. As shown in FIG. 26, each actual section data Da is extremely close to the corresponding theoretical section Dt.

In FIG. 26, the calculation data D9, the approximate curve Ac, and the evaluation value Ev are shown in order from the top as the values obtained from the dairy cow 51, which is the source of the actual cross-section data Daa and the actual cross-section data Dab, by the growth condition evaluation system 10. It is listed. Here, since the theoretical cross-section Dt and the actual cross-section data Da are extremely close, the calculation data D9, the approximate curve Ac, and the evaluation value Ev obtained for the actual cross-section data Da are substantially the same as the actual cross-section data Da. It is approximately equal to the calculation data D9, the approximate curve Ac, and the evaluation value Ev obtained for the theoretical cross section Dt. Therefore, if both calculation data D9, both approximate curves Ac, and both evaluation values Ev are appropriate for both theoretical cross sections Dt, the growth situation evaluation system 10 can appropriately evaluate the growth situation of the dairy cow 51. It can be considered that the evaluation is successful.

First, it can be seen from FIG. 26 that the calculation data D9a has a shape extremely similar to the actual cross-sectional data Daa that is the source thereof. Furthermore, it can be seen that the calculation data D9b also has a shape extremely similar to the actual cross-sectional data Dab that is the source thereof. From this, it can be seen that the growth status evaluation system 10 is able to appropriately represent the external shape of the actual cross section of the dairy cow 51, which is the source, of each calculation data D9 used for calculation.

Next, from FIG. 26, it can be seen that the approximate curve Aca has a shape extremely close to the calculation data D9a that is the source thereof. Furthermore, it can be seen that the approximate curve Acb has a shape extremely close to the calculation data D9b that is the source thereof. From this, it can be seen that the growth situation evaluation system 10 appropriately reflects the calculation data D9a, that is, the actual cross-sectional data Da that is the basis of the evaluation value EV obtained from the approximate curve Ac. Based on each approximate curve Ac, the growth status evaluation system 10 calculated 0.211 as the evaluation value EV of the example with poor fleshiness, and calculated 0.186 as the evaluation value EV of the example with good fleshiness. Therefore, the growth situation evaluation system 10 calculated 0.211 as the evaluation value EV for the theoretical cross section Dta of the example with poor fleshiness, and calculated 0.186 as the evaluation value EV for the theoretical cross section Dta of the example with good fleshiness. It can be thought of as a thing. Here, the evaluation value EV indicates that the smaller the value, the better the meatiness. Therefore, the growth situation evaluation system 10 can evaluate the growth situation of the dairy cow 51 in the same way as the visual indicators for evaluation of the growth situation shown in the above-mentioned literature. It can be seen that it represents properly.

Next, the relationship between the BCS and the evaluation value Ev for the approximate curve Ac will be explained using FIGS. 27 to 30. 27 to 30, four approximate curves Ac based on the cross-sectional data D8 obtained at each of the four slice positions Sp (see FIG. 18) are shown by numerical value in the conventional BCS. 27 shows each approximate curve Ac at slice position Sp1, FIG. 28 shows each approximate curve Ac at slice position Sp2, FIG. 29 shows each approximate curve Ac at slice position Sp3, and FIG. 30 indicates each approximate curve Ac at the slice position Sp4. In FIGS. 27 to 30, four approximate curves Ac are sample Sa with BCS of 2.50, sample Sb with BCS of 2.75, sample Sc with BCS of 3.00, and sample Sa with BCS of 2.75. The specimen Sd is 3.25. Here, in FIGS. 27 to 29, in order to easily understand the difference in shape of the approximate curve Ac of each sample S, that is, the calculation data D9, in the approximate curve Ac of each sample S, the first It is assumed that the control point C1, the seventh control point C7 serving as the end point Pe, and the vertex Cv located therebetween are substantially coincident. Further, in FIGS. 27 to 29, other control points C are omitted to avoid complicating the diagrams and making it difficult to understand each one.

Here, as described above, the slice position Sp4 is a plane perpendicular to the width direction Dw, and a group of points exists only in one side of the cutting plane direction (reference axis direction Db). Therefore, as described above, by copying (copying) one data in which a point group exists to the other, the above calculation data D9 can be obtained by copying (copying) one data in which a point group exists to the other slice position Sp. , a concave portion, and a convex portion. However, in the actual dairy cow 51, the cross section obtained at the slice position Sp4 is only one as described above, so in FIG. It is a curved line that depicts a convex portion. Therefore, in FIG. 30, the approximate curve Ac for each specimen S is a curve connecting the first control point C1, which is the starting point Ps, and the vertex Cv. In FIG. 30, in order to easily understand the difference in shape of the approximate curve Ac of each sample S, that is, the calculation data D9, in the approximate curve Ac of each sample S, the first control point C1, which is the starting point Ps, and the apex It is assumed that Cv and Cv are approximately the same. In FIG. 30, similar to FIGS. 27 to 29, other control points C are omitted to avoid complicating the diagram and making it difficult to understand each one.

As shown in FIGS. 27 to 30, regardless of the slice position Sp, the approximate curve Ac of the specimen Sd is located at the outermost side, the approximate curve Ac of the specimen Sc is located inside it, and the approximate curve Ac of the specimen Sb is located inside it. A curve Ac is located, and an approximate curve Ac of the sample Sa is located inside the curve Ac. Therefore, for each approximate curve Ac, that is, for the dairy cow 51 that is the source of the approximate curve Ac, the sample Sd with a BCS of 3.25 is the most fleshy, the sample Sc with a BCS of 3.00 is the next most fleshy, and the sample Sd with a BCS of 2. It can be seen that the sample Sb with a BCS of 75 is the next most fleshy, and the sample Sa with a BCS of 2.50 is the least fleshy.

The applicant calculates an evaluation value EV using the growth condition evaluation system 10 for each approximate curve Ac shown in FIGS. 27 to 30, that is, for each cross section at each slice position Sp of the dairy cow 51 that is the source did. In the example of FIG. 27, which is a cross section at the slice position Sp1, the evaluation value EV is such that the approximate curve Ac of the sample Sa is 0.403, the approximate curve Ac of the sample Sb is 0.395, and the approximate curve Ac of the sample Sc is was 0.374, and the approximate curve Ac of the sample Sd was 0.354. In this way, it can be seen that the evaluation value Ev becomes smaller as the thickness becomes better, that is, the BCS becomes larger.

Similarly, in the example of FIG. 28 where the evaluation value EV is a cross section at the slice position Sp2, the approximate curve Ac for the sample Sa is 0.233, the approximate curve Ac for the sample Sb is 0.231, and the approximate curve Ac for the sample Sc is 0.233. The curve Ac was 0.221, and the approximate curve Ac of the sample Sd was 0.207. In addition, in the example of FIG. 29 where the evaluation value EV is a cross section at the slice position Sp3, the approximate curve Ac of the sample Sa is 0.411, the approximate curve Ac of the sample Sb is 0.396, and the approximate curve Ac of the sample Sc is Ac was 0.370, and the approximate curve Ac of the sample Sd was 0.340. Furthermore, in the example of FIG. 30 which is a cross section at slice position Sp4, the approximate curve Ac of the sample Sa is 0.543, the approximate curve Ac of the sample Sb is 0.511, and the approximate curve Ac of the sample Sc is Ac was 0.468, and the approximate curve Ac of the sample Sd was 0.413. From these results, it can be seen that the evaluation value Ev becomes smaller as the meat becomes thicker, that is, the BCS becomes larger, regardless of the slice position Sp.

Therefore, although the numerical range of the evaluation value Ev changes depending on the slice position Sp, it can be seen that, like the BCS, it appropriately indicates the quality of the fleshiness. From this, it can be seen that the growth status evaluation system 10 appropriately represents the evaluation value Ev calculated from the approximate curve Ac based on each calculation data D9, appropriately representing the actual growth status of the dairy cow 51. Then, the growth situation evaluation system 10 calculates an approximate curve Ac for each calculation data D9 using a sixth-order Bezier curve, and calculates an evaluation value Ev from the approximate curve Ac. Even if the three-dimensional shapes are different, the evaluation value Ev can be easily and appropriately calculated, and the evaluation of a large number of animals can be automated.

The growth status evaluation system 10 can represent the growth status of the dairy cow 51 by indicating the numerical value calculated as the evaluation value Ev as described above. By the way, in the growth situation evaluation system 10, as described above, the numerical range of the evaluation value Ev changes depending on the slice position Sp. Therefore, the growth status evaluation system 10 can represent the growth status of the dairy cow 51 by indicating the evaluation value Ev together with the reference value at the evaluated slice position Sp. Furthermore, the growth status evaluation system 10 may convert the numerical range of each evaluation value Ev according to the evaluated slice position Sp into a unified value regardless of the difference in the slice position Sp. In this case, for example, each evaluation value Ev may be normalized by the maximum value at each slice position Sp, and each normalized evaluation value Ev at a plurality of different slice positions Sp may be calculated. Furthermore, the growth situation evaluation system 10 may convert each evaluation value Ev corresponding to the evaluated slice position Sp into a value corresponding to the BCS by multiplying it by a conversion coefficient for converting into the BCS. In this case, for example, as in the approximate curve Ac shown in FIGS. 27 to 30, the conversion coefficient is determined by determining both the BCS and the evaluation value Ev (the range in which each changes) for the same approximate curve Ac. be able to. Even in these cases, the growth status evaluation system 10 only normalizes the calculated evaluation value Ev using a predetermined method or converts it into a value corresponding to the BCS using a predetermined conversion coefficient. Therefore, evaluation of a large number of animals can be automated.

In addition, as described above, since there is a proportional relationship between the numerical value of the evaluation value Ev and the quality of meat, the growth status evaluation system 10 can also be used to express the growth status of the dairy cow 51 as a graded rank. good. For example, using the above numerical values, at each slice position Sp, the evaluation value Ev for which the BCS is 3.25 or more is ranked A with the highest fleshiness, and the evaluation value Ev for which the BCS is less than 3.25 and is 3.00 or more is ranked A. The evaluation value Ev that becomes the next most fleshy rank is B, and the evaluation value Ev that is 2.75 or more when the BCS is less than 3.00 is the next most fleshy rank C, and the BCS is less than 2.75 and is 2.50. The evaluation value Ev that is above can be set as a rank D that is not very fleshy, and the evaluation value Ev that is less than 2.50 can be set as a rank E that is not very fleshy. By indicating the graded ranks in this way, it is possible to judge whether the fleshiness is good or bad as a uniform evaluation regardless of the difference in the slice position Sp, and it is also possible to judge whether the fleshiness is good or bad intuitively rather than showing it numerically as the evaluation value Ev. can be made to understand. Even in this case, the growth status evaluation system 10 only allocates the evaluation value Ev calculated as described above to predetermined ranks, so that evaluation of a large number of animals can be automated. Note that the number of ranks described above, the numerical value of each evaluation value Ev applied to each rank, the evaluation comment of each rank, etc. may be set as appropriate, and are not limited to the above example.

Here, the growth status evaluation system of the prior art uses a distance image sensor to obtain a group of three-dimensional coordinates of the animal. Since the distance image sensor has a limit in increasing the accuracy and resolution of the three-dimensional coordinate group to be obtained, it is difficult for the three-dimensional coordinate group to appropriately represent the outline of the actual animal. For this reason, in the conventional health condition estimating device, it is difficult to obtain an appropriate evaluation by BCS even if a three-dimensional coordinate group is used.

In addition, since conventional health state estimation devices use distance image sensors, the outline of the animal indicated by the three-dimensional coordinate group differs from the actual animal, such as with jagged irregularities, so it is difficult to determine the appropriate outline shape. It becomes difficult to express it with a mathematical formula (approximate curve). For this reason, in the conventional health condition estimating device, it is difficult to obtain a mathematical formula that appropriately represents the contour shape, and it is difficult to appropriately evaluate the growth status using the contour.

Conventionally, veterinarians and the like have generally determined the BCS by grasping the shape of a dairy cow by feel, etc. Then, in BCS, the person seeking it (veterinarian, etc.) determines where the shape based on bones and fleshiness fits depending on their own appearance, so there is variation depending on the person seeking it, and the appropriate It is difficult to obtain a good evaluation. That is, in the conventional BCS, it is difficult to improve the accuracy of evaluation due to individual differences among the requesters and fluctuations in judgment even when requested by the same person. In addition, in BCS, being touched to understand the shape of the cow causes stress to the cow.

Furthermore, in the conventional health state estimation device, the breeding status is evaluated from the feature amount indicating the degree of concavity of the lumen, based on the three-dimensional shape of the animal reproduced from the animal's three-dimensional coordinate group. Here, in animals such as cows and pigs, it is required to be able to judge the state of fleshiness (good or bad) as a growth condition, but even animals of the same type have different skeletal sizes. For this reason, conventional health condition estimating devices use feature quantities that simply indicate the degree of concavity of the lumen, regardless of differences in skeletal size, making it difficult to appropriately judge the state of fleshiness. This is because animals with large skeletons and animals with small skeletons may naturally have different fleshiness even if they have the same concavity (size).

In addition, conventional health state estimation devices reproduce the three-dimensional shape of the animal by appropriately connecting the three-dimensional coordinates of the animal, and based on that three-dimensional shape, characteristics that indicate the degree of concavity of the lumen are calculated. The cultivation status is evaluated based on the quantity. Here, animals such as cows and pigs have different three-dimensional shapes depending on their skeletons, fleshiness, etc. even if they are of the same type. For this reason, with conventional health condition estimation devices, it is necessary to determine the feature amount that indicates the degree of concavity of the lumen for various three-dimensional shapes. There are a wide variety of methods, etc., making it difficult to automate the evaluation of a large number of animals.

In contrast, the growth status evaluation system 10 uses a laser measuring device 13 to obtain a group of three-dimensional coordinates of an animal (dairy cow 51 in Example 1). The laser measuring device 13 is capable of acquiring a three-dimensional coordinate group (point cloud data D1) with an accuracy level used for precise surveying, and has an extremely high resolution (in Example 1, the distance is 12.5 mm). ), the point cloud data D1 can represent the outline of the actual dairy cow 51 with extreme fidelity. Here, the laser measuring device 13 of the first embodiment can, for example, make the accuracy of the reachable distance of the pulsed laser beam an error of 3.5 mm or less, and make the accuracy of the scanning plane an error of 2.0 mm or less. This allows the accuracy of angle measurement to be less than 6 seconds (angle) both vertically and horizontally. Further, the laser measuring device 13 of the first embodiment can be set to a low output mode, but even in that case, the accuracy of the reachable distance of the pulsed laser beam must be kept within an error of 4.0 mm. , and others can have the above-mentioned accuracy. In this way, by using the laser measuring device 13, the growth status evaluation system 10 can obtain a three-dimensional coordinate group with extremely high accuracy and can also achieve extremely high resolution. Therefore, the growth situation evaluation system 10 can appropriately obtain an evaluation of the growth situation by using the point cloud data D1 from the laser measuring device 13.

In addition, since the growth condition evaluation system 10 uses the laser measuring device 13, the external shape of the point cloud data D1 indicating the dairy cow 51 can represent the outline of the actual dairy cow 51 extremely faithfully. Approximation can be performed using a smooth curve, and an approximated curve Ac that appropriately represents the outline of the actual dairy cow 51 can be obtained. Therefore, in the growth status evaluation system 10, the outline of the dairy cow 51 can be represented by an approximate curve Ac, which is a standardized mathematical formula, and it is possible to obtain a uniform evaluation of the growth status. That is, the growth status evaluation system 10 can express the body shape of the dairy cow 51 using a mathematical formula by approximating the external shape based on the point cloud data D1 to obtain an approximate curve Ac, and can evaluate the growth status. Since it can be calculated by calculation, the evaluation can be made objective and appropriate.

Furthermore, the growth situation evaluation system 10 detects a reference point Pr from each specific part alignment data D6 generated based on each point cloud data D1, and arranges each specific part alignment data so that the reference point Pr has a uniform size. Each normalized data D7 is generated by enlarging or reducing D6. The growth situation evaluation system 10 also generates each cross-sectional data D8 from each of the normalized data D7, generates calculation data D9 from each cross-section data D8, and generates an approximate curve Ac adapted to the calculation data D9. The evaluation value Ev is calculated from. For this reason, the growth status evaluation system 10 can express the state of fleshiness with respect to the skeleton as an approximate curve Ac, and calculates the evaluation value Ev from the approximate curve Ac, thereby greatly suppressing the influence of differences in the size of the skeleton. It is possible to obtain the evaluation value Ev, and to appropriately judge the appearance of fleshiness. In addition, the growth status evaluation system 10 can calculate the evaluation value Ev based on the cross-sectional data D8 of the common slice position Sp determined based on the characteristic parts using bones etc., so that the growth status can be easily compared between animals. be able to appropriately evaluate

Even if the three-dimensional shape of each individual is different depending on the skeleton, fleshiness, etc., the growth condition evaluation system 10 draws convex, concave, convex, concave, convex parts as an overall tendency, and draws the convex part in the middle. The focus is on the curved line in which the spine 51d is located. The growth situation evaluation system 10 applies a sixth-order Bezier curve in order to calculate a smooth approximate curve Ac that fits the calculation data D9 based on each point group data D1. For this reason, the growth situation evaluation system 10 sets the starting point Ps and the ending point Pe, which are both ends of the external shape indicated by the calculation data D9, as a fixed control point C, and also sets the starting point Ps and the ending point Pe as fixed control points C. Set one control point C. Then, the growth situation evaluation system 10 calculates an approximate curve Ac adapted to the calculation data D9 by adjusting the positions of the five control points C, and calculates an evaluation value Ev from the approximate curve Ac. Therefore, the growth status evaluation system 10 can express the calculation data D9 by an approximate curve Ac using a sixth-order Bezier curve, even if the three-dimensional shape of each individual is different depending on the skeleton, fleshiness, etc. The evaluation value Ev can be calculated from the approximate curve Ac. Thereby, the growth situation evaluation system 10 can standardize the calculation of the approximate curve Ac and the evaluation value Ev, and can easily automate the evaluation of a large number of animals.

When the slice position Sp is a plane orthogonal to the width direction Dw, the growth condition evaluation system 10 sets the thinned cross-sectional data D84' in the cutting plane direction so that the right end of the thinned cross-sectional data D84' is located at the origin in the cutting plane direction. The calculation data D9 is generated by displacing it to the opposite side and copying it to the opposite side. For this reason, the growth condition evaluation system 10 calculates the calculation data D9 in which the plane orthogonal to the reference axis direction Db is the slice position Sp, in the same manner as in the case where the plane orthogonal to the width direction Dw is the slice position Sp. , it is possible to generate calculation data D9 that can be represented by an approximate curve Ac using a sixth-order Bezier curve.

By standardizing the method of determining the approximate curve Ac, the growth status evaluation system 10 can indicate individual differences in the outline of the dairy cow 51 by the position of each control point C on the approximate curve Ac. Therefore, the growth situation evaluation system 10 can generate evaluation values Ev based on changes in the position of each control point C, etc., making it possible to judge the growth situation with more uniformity.

Since the growth situation evaluation system 10 generates the evaluation value Ev using the approximate curve Ac obtained from the point cloud data D1 acquired by the laser measuring device 13, the evaluation value Ev can be notified as an evaluation of the growth situation. For this reason, the growth status evaluation system 10 can prevent individual differences and fluctuations in judgment by those seeking the growth from being reflected in the evaluation of the growth status, and can also eliminate differences due to differences in location, facility, time, etc. can. Thereby, the growth status evaluation system 10 can generate the evaluation value Ev based on a unified standard, and can evaluate the dairy cow 51 (animal) as a quantitative value.

Only when the presence of the dairy cow 51 in the drinking fountain 52 is detected via the camera 12, the growth condition evaluation system 10 acquires point cloud data D1 of the drinking fountain 52 using both laser measuring devices 13, and calculates an evaluation value based on the point cloud data D1 of the drinking fountain 52. Ev can be generated. Therefore, the growth status evaluation system 10 can obtain the evaluation value Ev without touching the dairy cow 51 while taking advantage of the fact that the dairy cow 51 comes to the drinking fountain 52 of its own will. Therefore, in order to generate the evaluation value Ev, the growth status evaluation system 10 makes it possible to appropriately judge the growth status of the dairy cow 51 without causing stress such as touching it or leading it to a place against its will. do. Furthermore, since the growth status evaluation system 10 automatically obtains the evaluation value Ev using the natural behavior of the dairy cow 51, it can eliminate time restrictions and can be managed all day long (24 hours). Furthermore, since the growth condition evaluation system 10 does not acquire the point cloud data D1 using both laser measuring devices 13 when the dairy cow 51 is not present at the drinking fountain 52, it is possible to prevent the acquisition and accumulation of unnecessary data. and can be operated efficiently. In addition, since the growth status evaluation system 10 is provided with laser measuring devices 13 arranged in pairs across the data acquisition area 14, the evaluation value can be obtained regardless of the animal's position, posture, facing direction, etc. The possibility of acquiring the point cloud data D1 necessary for generating Ev can be greatly increased.

The growth situation evaluation system 10 of Example 1 of the growth situation evaluation system according to the present disclosure can obtain the following effects.
The growth status evaluation system 10 is a laser measuring device that receives reflected light from an animal (dairy cow 51) of an emitted laser beam (pulsed laser beam) to obtain point cloud data D1 indicating the external shape of the animal in three-dimensional coordinates. 13. The growth situation evaluation system 10 also includes a reference point detection unit 67 that detects the reference point Pr in the point cloud data D1, and a standard point detection unit 67 that adjusts the size of the point cloud data D1 so that the reference point Pr has a uniform size. A normalization unit 69 that generates normalized data D7 is provided. The growth condition evaluation system 10 includes an approximate curve calculation unit 74 that calculates an approximate curve Ac that fits the normalized data D7, and an evaluation value Ev that indicates the evaluation of the animal (dairy cow 51) based on the approximate curve Ac. and an evaluation value calculation unit 75. For this reason, the growth status evaluation system 10 can represent the state of fleshiness with respect to the skeleton as an approximate curve Ac, and calculates the evaluation value Ev from the approximate curve Ac, thereby significantly suppressing the influence of differences in the size of the skeleton. This allows you to appropriately judge the appearance of flesh.

The growth condition evaluation system 10 also generates alignment data (specific part alignment data D6) by aligning the orientation of the point cloud data D1 so that the reference point Pr faces in a predetermined direction with respect to the set reference axis direction Db. A normalizing unit 69 adjusts the size of the aligned data so that the reference portion Pr has a uniform size, and generates normalized data D7. For this reason, the growth condition evaluation system 10 makes the reference point Pr a uniform size for the aligned data that is arranged using the reference point Pr as a reference, so that the normalized data D7 can be generated easily and appropriately. can.

Furthermore, the growth condition evaluation system 10 performs specific region cutting to generate specific region data D5 in which the vicinity of the head 51b and the tail 51c of the animal (dairy cow 51) are removed from the point cloud data D1 acquired by the laser measuring device 13. A reference point detection section 67 detects a reference point Pr from the specific part data D5. For this reason, the growth situation evaluation system 10 detects the reference point Pr from the specific part data D5, which is extracted from the point cloud data D1, only the points necessary for the evaluation of the growth situation. The number of man-hours required for data processing can be reduced, and the reference point Pr can be detected more appropriately and in a shorter time.

The growth situation evaluation system 10 includes a specific part alignment unit 66 that aligns the specific part data D5 so that the spine 51d in the specific part data D5 matches the reference axis direction Db, and a reference part detection unit 67 that The reference location Pr is detected from the specific location data D5 arranged by the specific location alignment section 66. For this reason, the growth situation evaluation system 10 detects the reference point Pr for the specific part data D5 aligned so that the spine 51d coincides with the reference axis direction Db, so the data processing required to detect the reference point Pr is performed. The number of man-hours can be further reduced, and the reference point Pr can be detected more appropriately and in a shorter time.

In the growth situation evaluation system 10, the approximate curve calculation unit 74 fits a single Bezier curve having a plurality of control points C to the contour of a cut surface obtained by cutting the normalized data D7 along a predetermined slice position Sp. Then, the approximate curve Ac is calculated, and the evaluation value calculation unit 75 calculates the evaluation value Ev using the positional relationship of the control point C on the approximate curve Ac. Therefore, the growth condition evaluation system 10 can calculate the evaluation value Ev using only the positional relationship of the control point C in the approximate curve Ac, which is a single Bezier curve, so the calculation of the evaluation value Ev is simplified. can.

In the growth situation evaluation system 10, the evaluation value calculation unit 75 calculates the evaluation value Ev using the ratio of the intervals between the control points C in the approximate curve Ac. Therefore, the growth situation evaluation system 10 can calculate the evaluation value Ev by simply finding the interval between each control point C on the approximate curve Ac, and can further simplify the calculation of the evaluation value Ev.

Therefore, in the growth situation evaluation system 10 as an example of the growth situation evaluation system according to the present disclosure, it is possible to appropriately obtain an evaluation of the growth situation of the animal (dairy cow 51).

Although the growth status evaluation system of the present disclosure has been described above based on Example 1, the specific configuration is not limited to Example 1, and the gist of the invention according to each claim is described below. Changes and additions to the design are permitted as long as they do not deviate.

For example, in the first embodiment, the drinking fountain 52 is set as the data acquisition area 14. However, the data acquisition area 14 may be set as appropriate and is not limited to the configuration of the first embodiment. Here, the data acquisition area 14 is set as a place, such as a feeding area, that animals regularly visit on their own will, similar to the drinking fountain 52, so that the breeding situation can be monitored without causing stress for the animals. Appropriate evaluation can be obtained. In addition, the data acquisition area 14 is set in a place where the animals can stop of their own accord for the time necessary to complete scanning by each laser measurement device 13, thereby further reducing stress on the animals and monitoring the breeding status. can be evaluated appropriately. Here, the above-mentioned required time depends on the scanning speed of each laser measuring device 13, and can be shortened by increasing the number of pulsed laser beams emitted at once, and the time required for stopping is, for example, Even if the length is shortened, the point cloud data D1 can be appropriately acquired.

Furthermore, in Example 1, an evaluation value Ev indicating the evaluation of the breeding status of the dairy cow 51 is generated as an example of the animal. However, the object for which the evaluation value Ev is generated (target for evaluating the breeding situation) may be any animal whose breeding situation is required to be evaluated, and is not limited to the configuration of the first embodiment.

Furthermore, in Example 1, the approximate curve Ac and the evaluation value Ev are calculated from the calculation data D9 corresponding to a single slice position Sp. However, for each animal (dairy cow 51), the approximate curve Ac and the evaluation value Ev are calculated from the calculation data D9 corresponding to a plurality of slice positions Sp, and the average value of the evaluation values Ev for each approximate curve Ac is calculated. It may be calculated. Here, the growth status evaluation system 10 weights each evaluation value Ev appropriately, since the degree to which the growth status of the dairy cow 51 is reflected may change depending on the difference in the slice position Sp, and then calculates the average value. It may also be something to seek. In this way, the growth condition evaluation system 10 can obtain the average value of the evaluation value Ev indicating the fleshiness at a plurality of slice positions Sp, and even if the fleshiness is uneven depending on the location, the growth condition evaluation system 10 can obtain the average value of the evaluation value Ev indicating the fleshiness at a plurality of slice positions Sp. You can get evaluation.

As described above, the growth condition evaluation system 10 changes the numerical range of the evaluation value Ev depending on the slice position Sp. Each slice position Sp may be standardized by the maximum value of a plurality of evaluation values Ev, and the average value of the normalized evaluation values Ev of the plurality of different slice positions Sp may be calculated for each animal (dairy cow 51). Furthermore, as described above, the average value may be calculated after converting into a value corresponding to the BCS. By doing so, the growth condition evaluation system 10 can suppress the difference in influence on the evaluation of fleshiness due to the difference in the slice position Sp, and can obtain a more appropriate evaluation of the growth condition.

In Example 1, the approximate curve calculation unit 74 calculates the approximate curve Ac using a sixth-order Bezier curve. However, if the approximate curve calculation unit 74 uses a single Bezier curve to calculate an approximate curve that matches the calculation data D9 based on the point cloud data D1, the order (number of control points C) of the approximate curve may be set appropriately, and is not limited to the configuration of the first embodiment.

In the first embodiment, the reference point detection section 67 detects the reference point Pr in the point cloud data D1, and the normalization section 69 adjusts the size of the point cloud data D1 so that the reference point Pr has a uniform size. Normalized data D7 is generated, and an approximate curve Ac and an evaluation value Ev are calculated from calculation data D9 based on the normalized data D7. However, since the evaluation value Ev is calculated using the ratio of the interval between the control points C in the approximate curve Ac, the point group data D1 whose size is not adjusted so that the reference point Pr has a uniform size The approximate curve Ac and the evaluation value Ev may be calculated from the calculation data D9 based on the calculation data D9, and the configuration is not limited to the first embodiment.

In the first embodiment, the animal detection unit 61 functions as an animal detection mechanism that detects the presence of an animal in the data acquisition area 14 based on the image from the camera 12. However, the animal detection mechanism is not limited to the configuration of the first embodiment as long as it detects the presence of an animal in the data acquisition area 14. As an example of this, for example, instead of the camera 12, it is possible to use something that uses infrared rays, such as an infrared scanner or an infrared thermograph. In this case, animals can be detected more reliably even in dark conditions. Further, as other examples, for example, a pressure sensor may be provided at the faucet of the drinking fountain 52, a device for detecting weight may be provided at the data acquisition area 14, or a sensor may be provided at the entrance/exit of the data acquisition area 14. can.

In the first embodiment, the laser measuring devices 13 are provided in pairs with the data acquisition area 14 in between. However, the laser measuring device 13 may be provided as a single device or three or more as long as it obtains the point cloud data D1 of the animal existing in the data acquisition area 14, and is limited to the configuration of the first embodiment. Not done.

By the way, conventional health state estimation devices reproduce the three-dimensional shape of the animal by appropriately connecting the three-dimensional coordinates of the animal, and based on the three-dimensional shape, feature quantities indicating the degree of concavity of the lumen are calculated. The training status has been evaluated by (Problem to be solved by the invention)

Here, animals such as cows and pigs have different three-dimensional shapes depending on their skeletons, fleshiness, etc., even if they are the same type of animal. Then, with conventional health condition estimation devices, it is necessary to find the feature quantity that indicates the degree of concavity of the lumen for various three-dimensional shapes. etc., and it is difficult to automate the evaluation of a large number of animals. (Problem to be solved by the invention)

The present disclosure has been made in view of the above circumstances, and aims to provide a growth condition evaluation system that can automate and appropriately obtain evaluations of animal growth conditions. (Problem to be solved by the invention)

In order to solve the above-mentioned problems, the growth condition evaluation system of the present disclosure uses a laser beam that receives reflected light from the animal of the emitted laser beam to obtain point cloud data indicating the external shape of the animal in three-dimensional coordinates. a measuring device; an approximate curve calculation unit that calculates an approximate curve by fitting a single sixth-order Bezier curve having seven control points to the point group data; An evaluation value calculation unit that calculates an evaluation value indicating . (Means for solving problems)

According to the growth status evaluation system of the present disclosure, it is possible to automate and appropriately obtain an evaluation of the growth status of an animal. (Effect of the invention)

[1]
a laser measuring device that obtains point cloud data indicating the external shape of the animal in three-dimensional coordinates by receiving reflected light from the animal of the emitted laser beam;
an approximate curve calculation unit that calculates an approximate curve by fitting a single sixth-order Bezier curve having seven control points to the point cloud data;
A growth situation evaluation system comprising: an evaluation value calculation unit that calculates an evaluation value indicating evaluation of the animal based on the approximate curve.

[2]
a cutting plane extraction unit that generates cross-sectional data that is a cross-section obtained by cutting the point cloud data acquired by the laser measuring device along a predetermined slice position;
In the cross-sectional data, compare the number of point clouds located on both sides from the center position in the direction of the cutting plane, select the side with the largest number of point clouds, and copy the data on the selected side to the opposite side to obtain calculation data. a calculation data generation unit that generates;
The growth situation evaluation system according to [1], wherein the approximate curve calculation unit calculates the approximate curve by adapting it to the calculation data based on the point group data.

[3]
The growth situation evaluation system according to [2], wherein the evaluation value calculation unit calculates the evaluation value using a positional relationship of the control points in the approximate curve.

[4]
The growth situation evaluation system according to [3], wherein the evaluation value calculation unit calculates the evaluation value using a ratio of intervals between the control points in the approximate curve.

[5]
The seven control points are, in order from one end, a first control point, a second control point, a third control point, a fourth control point, a fifth control point, a sixth control point, and a seventh control point,
The interval between the second control point and the third control point is a first interval, the interval between the third control point and the fourth control point is a second interval, and the interval between the second control point and the fourth control point is a second interval. The interval between the points is a third interval, the interval between the sixth control point and the fifth control point is a fourth interval, the interval between the fifth control point and the fourth control point is a fifth interval, The interval between the sixth control point and the fourth control point is a sixth interval,
The evaluation value calculation unit calculates the evaluation value by dividing the third interval by the sum of the first interval and the second interval, or divides the sixth interval by the sum of the fourth interval and the second interval. The growth situation evaluation system according to [3] or [4], wherein the evaluation value is calculated by dividing by a value obtained by adding 5 intervals.

[6]
The cutting plane extraction unit generates a plurality of cross-sectional data that are cross-sections obtained by cutting the point cloud data acquired by the laser measuring device along a plurality of mutually different slice positions,
The calculation data generation unit generates the calculation data for each of the plurality of cross-sectional data,
The approximate curve calculation unit calculates the approximate curve for each of the plurality of calculation data,
The growth situation evaluation system according to [5], wherein the evaluation value calculation unit calculates an average value of the evaluation values for each of the approximate curves.

[7]
The evaluation value calculation unit calculates the evaluation values of the approximate curves for the plurality of animals for each of the plurality of different slice positions, and normalizes the evaluation values by the maximum value of the plurality of evaluation values for each of the plurality of different slice positions. , the growth status evaluation system according to [6], characterized in that the average value of the normalized evaluation values of the plurality of slice positions different for each animal is calculated.

The growth status evaluation system 10 is a laser measuring device that receives reflected light from an animal (dairy cow 51) of an emitted laser beam (pulsed laser beam) to obtain point cloud data D1 indicating the external shape of the animal in three-dimensional coordinates. 13. The growth condition evaluation system 10 also includes an approximate curve calculation unit 74 that calculates an approximate curve Ac by adapting a single sixth-order Bezier curve having seven control points C to the point cloud data D1; and an evaluation value calculation unit 75 that calculates an evaluation value Ev indicating the evaluation of the animal (dairy cow 51) based on the evaluation value Ev of the animal (dairy cow 51). For this reason, even if the three-dimensional shape of each individual is different depending on the skeleton, fleshiness, etc., the growth condition evaluation system 10 can generate an approximate curve adapted to the point cloud data D1 by adjusting the positions of the seven control points C. Ac can be calculated. Thereby, the growth status evaluation system 10 can standardize the calculation of the approximate curve Ac and the evaluation value Ev, and can easily automate the evaluation of the growth status of a large number of animals (dairy cows 51).

The growth condition evaluation system 10 also includes a cutting plane extraction unit 71 that generates cross-sectional data D8, which is a cross-section obtained by cutting the point cloud data D1 acquired by the laser measuring instrument 13 along a predetermined slice position Sp; , compare the number of point clouds located on both sides from the center position (center line Lc) in the cutting plane direction, select the side with the largest number of point clouds, and copy the data on the selected side to the opposite side. and a calculation data generation unit 72 that generates calculation data D9. Then, in the growth situation evaluation system 10, the approximate curve calculation unit 74 calculates the approximate curve Ac by adapting it to the calculation data D9 based on the point cloud data D1. For this reason, the growth situation evaluation system 10 can generate calculation data D9 in which the number of missing data points in the width direction Dw is reduced while suppressing an increase in the amount of data. In addition, the growth situation evaluation system 10 substantially adjusts the positions of the three control points C so that the starting point Ps (the first control point C1 ) to the end point Pe (seventh control point C7), the approximate curve Ac can be calculated, and the approximate curve Ac and the evaluation value Ev can be more easily calculated.

Further, in the growth situation evaluation system 10, the evaluation value calculation unit 75 calculates the evaluation value Ev using the positional relationship of the control point C on the approximate curve Ac. Therefore, the growth condition evaluation system 10 can calculate the evaluation value Ev using only the positional relationship of the control point C in the approximate curve Ac, which is a single Bezier curve, so the calculation of the evaluation value Ev is simplified. can.

In the growth situation evaluation system 10, the evaluation value calculation unit 75 calculates the evaluation value Ev using the ratio of the intervals between the control points C in the approximate curve Ac. Therefore, the growth situation evaluation system 10 can calculate the evaluation value Ev by simply finding the interval between each control point C on the approximate curve Ac, and can further simplify the calculation of the evaluation value Ev.

The growth situation evaluation system 10 sets the seven control points C as, in order from one end, a first control point C1, a second control point C2, a third control point C3, a fourth control point C4, and a fifth control point C5. , a sixth control point C6, and a seventh control point C7. In addition, the growth condition evaluation system 10 sets the interval between the second control point C2 and the third control point C3 as a first interval a, and the interval between the third control point C3 and the fourth control point C4 as a second interval b. , the interval between the second control point C2 and the fourth control point C4 is the third interval c, the interval between the sixth control point C6 and the fifth control point C5 is the fourth interval d, and the interval between the fifth control point C5 and the fifth control point C5 is the third interval c. Let the interval between the fourth control point C4 be a fifth interval e, and the interval between the sixth control point C6 and the fourth control point C4 be a sixth interval f. Then, in the growth situation evaluation system 10, the evaluation value calculation unit 75 calculates the evaluation value Ev by dividing the third interval c by the sum of the first interval a and the second interval b, or calculates the evaluation value Ev by dividing the third interval c by the sum of the first interval a and the second interval b. The evaluation value Ev is calculated by dividing f by the sum of the fourth interval d and the fifth interval e. Therefore, the growth situation evaluation system 10 can more easily calculate the evaluation value Ev.

Therefore, in the growth situation evaluation system 10 as an example of the growth situation evaluation system according to the present disclosure, it is possible to automate and appropriately obtain evaluation of the growth situation of the animal (dairy cow 51).

[Cross reference to related applications]
This application has priority based on patent application No. 2022-054130 filed with the Japan Patent Office on March 29, 2022 and patent application No. 2022-054131 filed with the Japan Patent Office on March 29, 2022. , the entire disclosure of which is hereby incorporated by reference in its entirety.

Claims (6)

1. a laser measuring device that obtains point cloud data indicating the external shape of the animal in three-dimensional coordinates by receiving reflected light from the animal of the emitted laser beam;
   a reference point detection unit that detects a reference point in the point cloud data;
   a normalization unit that generates normalized data by adjusting the size of the point cloud data so that the reference location has a uniform size;
   an approximate curve calculation unit that calculates an approximate curve that matches the normalized data;
   A growth situation evaluation system comprising: an evaluation value calculation unit that calculates an evaluation value indicating evaluation of the animal based on the approximate curve.
2. comprising a data alignment unit that generates alignment data by aligning the orientation of the point cloud data so that the reference location faces in a predetermined direction with respect to a set reference axis direction;
   2. The normalization unit generates the normalized data by adjusting the size of the alignment data based on the point cloud data so that the reference location has a uniform size. growth status evaluation system.
3. comprising a specific part cutting unit that generates specific part data by removing the vicinity of the head and the tail of the animal from the point cloud data acquired by the laser measuring device;
   The growth situation evaluation system according to claim 2, wherein the reference point detection unit detects the reference point from the specific part data based on the point cloud data.
4. a specific part alignment unit that aligns the specific part data so that the spine in the specific part data coincides with the reference axis direction;
   4. The growth situation evaluation system according to claim 3, wherein the reference point detection section detects the reference point from the specific section data arranged by the specific section alignment section.
5. The approximate curve calculation unit calculates the approximate curve by fitting a single Bezier curve having a plurality of control points to a contour of a cut plane obtained by cutting the normalized data along a predetermined slice position,
   The growth situation evaluation system according to claim 4, wherein the evaluation value calculation unit calculates the evaluation value using a positional relationship of the control points in the approximate curve.
6. The growth situation evaluation system according to claim 5, wherein the evaluation value calculation unit calculates the evaluation value using a ratio of intervals between the control points in the approximate curve.
   

Images