KR102258282B1 - An apparatus for determination of body weight of Korean Hanwooo using three-dimensional imaging - Google Patents

Abstract

The present invention relates to an apparatus for measuring the weight of Korean cattle by applying a three-dimensional image acquisition and image processing technology for Korean cattle. It takes a lot of effort to measure the weight of Korean cattle using a conventional scale, and if the livestock is not fixed, there is a risk of injury.
According to the present invention, the ToF (Time Of Flight) camera A for acquiring a three-dimensional image of Korean cattle is connected to the extension arm C by a fixing bracket B. The extension arm (C) is connected to the base arm (D), and a counterweight (E) is used to balance the weight of the base arm (D). The basic arm (D) is supported by a tripod (F). The image acquired by the ToF camera (A) is stored in the computer (G), and the weight is determined by the image processing and weight estimation algorithm.
According to the above configuration, body type information such as the body length, height, bust width, yaw width, etc. of Korean cattle was determined, and the measured Hanwoo's age was substituted with the above information in the regression equation (1) to provide a means to measure the weight of the Korean cattle. . This configuration provided an effective means for measuring the weight of Korean beef over almost all ages.

Description

An apparatus for determination of body weight of Korean Hanwooo using three-dimensional imaging.

The present invention relates to a device for non-contact measuring the weight of live Korean cattle using an image acquisition device and an image processing device.

Prior art prior to the present invention is a photographing unit for photographing an image; A sensor unit for measuring a distance to livestock; And a control unit that calculates the weight of livestock using the captured image and data on the measured distance, and improves the accuracy of the weight calculation through deep learning. .

Another prior art relates to an image processing weight measuring device that measures the weight of a cow, wherein the image processing weight measuring device is installed at a certain place where the cow moves, and has entrances on both sides to allow the cow to enter and exit. The gate and; At least one camera unit installed at a predetermined position of the gate to capture at least one surface of the cow; An image processing unit connected to the camera unit and configured to perform image processing of each of the captured image signals to enable monitoring; A monitor that displays the digital signal processed by the image processing unit so that an operator can visually see it; A technology including a weight calculation unit that is connected to the camera unit, controls and controls the entire system, receives a video bitstream signal according to each angle output from the image processing unit, and calculates the volume and chest size of the cow.

Unexamined Patent Publication 10-2019-0068266 Unexamined Patent Publication No. 2002-0024688

The present invention relates to an apparatus for measuring the weight of Korean beef by applying a three-dimensional image acquisition and image processing technology for Korean beef. It takes a lot of effort to measure the weight of Korean cattle using a conventional scale, and if the livestock is not fixed, there is a risk of injury. If the manure remains on the scale, it is impossible to accurately measure the weight.

In addition, although there is an invention in which the weight of a cow is calculated by image processing in the preceding document, the calculation is complicated by photographing an image using three cameras, and it has not been possible to measure the weight of cows of various ages. In addition, since the body shape of each cow is different, it was impossible to apply the method of measuring the weight of the cow to Korean cattle as it is. In addition, in the prior art, it is described that it can be measured through image processing, but detailed technology is not described, and thus it is not applicable.

In addition, a'simple weight measuring tape for Korean beef' has been developed to predict the weight by using the correlation between body type information and weight, but it is not used properly due to the risk of injury and deterioration in accuracy during the body shape measurement process. There is a need to develop a new measurement technology to replace it.

In order to solve the above problems, the present invention provides the following problem solving means.

ToF camera (A), fixing bracket (B), extension arm (C), basic arm (D), counterweight (E), tripod (F), computer (G) to enable 3D image acquisition and image processing It provides a Hanwoo weight measuring device comprising a.

In addition, it provides a Korean cattle weight measuring apparatus, characterized in that the body length, body height, bust width and maximum width are measured from the obtained 3D image, and the weight of the Korean cattle is measured by substituting the age of the target Korean cattle into the following equation (1). .

------ 식(1)

------ Equation (1)

In addition, an absolute encoder (H, not shown) that measures the angle between the extension arm (C) and the base arm (D), and an absolute encoder (I) that measures the angle between the base arm (D) and the tripod (F). , Not shown) and the height of the tripod (F) and the height of the ToF camera to determine the height of the ToF camera, and measure the height of the Korean cattle by using the point closest to the cattle from the ToF camera, and use this as the absolute height. It provides a Hanwoo weight measuring device, characterized in that to set.

In addition, when there is a difference between the body height measured from the obtained 3D image and the absolute body height, the absolute body height is used as a value of the body height in Equation (1).

In addition, there is provided a Hanwoo weight measuring device, characterized in that it automatically reads and uses the month age information of the Korean cattle measured by the RFID value of the Korean cattle by using the RFID attached to the tag of the Korean cattle in order to know the age of the cattle.

In addition, instead of the month-age information, it provides a Korean cattle weight measuring apparatus, characterized in that the accumulated feed intake of the Korean cattle is used by using RFID information attached to the tag of the Korean cattle.

According to the above configuration and operation, the present invention can measure the weight of Korean beef by applying a three-dimensional image acquisition and image processing technology for Korean beef. In addition, it is possible to determine the body type information such as the body length, body height, bust width, and yaw width of Korean cattle.

1 is a configuration diagram of a measuring device of the present invention
Figure 2 is a point cloud image measured using the measuring device of the present invention
3 is an example of the ToF camera of the present invention

In order to measure the weight of Korean cattle based on image information, it is necessary to acquire 3D images and develop image processing technology for Korean cattle. Until now, CCD cameras, stereo vision cameras, or Kinect have been used to acquire images of cows or pigs to analyze their body shape.

When a CCD camera is used, only 2D images of an object are acquired, and thus 3D information cannot be obtained. Meanwhile, a stereo vision camera or Kinect is being used to obtain 3D body shape information. However, in the case of using a stereo vision camera or Kinect, an image processing technique is required to accurately match two images in order to reduce errors. Moreover, Kinect is an interface device developed to enjoy games and entertainment using a human body, and it is easy to be affected by lighting conditions, and problems such as recognizing objects as humans and attempting skeletonization occur. When estimating the three-dimensional information of the large livestock, the error appears large.

Therefore, it is an object of the present invention to obtain a three-dimensional image of Korean cattle using a ToF (time of flight) camera to solve the above problems, and to measure the weight of Korean cattle by developing an index suitable for weight estimation from the obtained image. To develop.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram schematically showing an apparatus according to the present invention. According to the present invention, the ToF (Time Of Flight) camera A for acquiring a three-dimensional image of Korean cattle is connected to the extension arm C by a fixing bracket B. The extension arm (C) is connected to the base arm (D), and a counterweight (E) is used to balance the weight of the base arm (D). The basic arm (D) is supported by a tripod (F). The image acquired by the ToF camera (A) is stored in the computer (G), and the weight is determined by the image processing and weight estimation algorithm.

The ToF camera is provided with a light source that emits light around the image acquisition camera as shown in FIG. 3, and the light is reflected by the object and returned, and the light when the light source is turned off is separately acquired. This is the principle of measuring the distance by using the difference in contrast according to the distance between two signals.

According to the present invention, the weight of Korean cattle is measured in the order of 3D image acquisition, image processing, and weight determination.

A point cloud image (PCI) of the top view of Korean beef acquired using the ToF camera (A) is shown in FIG. 2. Since noise may be included in the point cloud image of Korean cattle, outliers due to noise are removed using the k-nearest neighbors algorithm. Next, the height, length, bust width, and yaw width as well as the maximum width are determined from the image of the point cloud of the Korean cattle.

The noise reduction can be determined as noise if the difference is 5 to 10 mm or more by comparing the average height value of the surrounding 10 points and the height value of the corresponding point using a moving average method,

Although there is a height difference of 5 to 10 mm or more based on the point, if there is a difference within 5 to 10 mm from the points immediately adjacent to the left or right or up and down, a noise removal and boundary line search method, characterized in that the boundary is set, is used.

In addition, in order to reduce the distance error that may occur as the ToF camera moves away from the vertical lower direction and the vertical direction, the base arm is rotatably installed around the tripod, and an absolute encoder (J, not shown) at the rotating part. By installing the ToF camera, you can measure the size of the Korean cattle, and the measured distance information combines the absolute encoder (j) and the image image, and recognizes that the vertical and downward distance measurement of the camera is the most accurate. It is characterized in that it is possible to calibrate the measured value.

In addition, when the weight of the Korean cattle measured by the image is less than 5% of the normal compared to the age of the month, this information is recorded in the RFID of the Korean cattle so that the breeder can inspect and manage the Korean cattle whose weight is less than 5%. Provides a means.

In addition, if there are many movements of female Korean cattle in the video, there may be health problems or estrus, and thus, a means for notifying the breeder of abnormal conditions through a buzzer connected to the computer or an Internet e-mail transmission function is provided.

Table 1 shows the body shape information including body height and body length measured for Korean beef of 6-109 months of age and the results of the experiment on body weight.

Age (months) Length (m) Body height (m) Bust width (m) Maximum width (m) Weight (kg) 6 0.857 0.711 0.353 0.418 155 6 0.907 0.906 0.311 0.398 159 6 0.849 0.952 0.312 0.437 172 6 0.846 0.831 0.325 0.422 150 6 0.934 0.940 0.311 0.423 165 6 0.789 0.812 0.321 0.423 149 13 1.068 1.132 0.308 0.467 240 13 0.970 0.984 0.262 0.465 190 13 1.111 1.055 0.285 0.488 231 18 1.051 0.940 0.360 0.512 257 18 0.960 0.949 0.370 0.494 260 18 1.032 0.952 0.337 0.544 292 18 1.052 1.061 0.337 0.575 320 18 1.117 1.093 0.397 0.575 361 18 0.931 0.863 0.359 0.454 223 18 1.026 1.114 0.396 0.502 305 21 1.038 1.222 0.322 0.536 315 22 1.040 1.239 0.306 0.532 304 22 1.048 1.243 0.361 0.537 323 22 1.121 1.259 0.352 0.587 403 22 1.091 1.191 0.339 0.506 306 24 1.223 1.163 0.425 0.562 382 24 1.175 1.177 0.356 0.475 300 24 1.256 1.027 0.356 0.499 308 24 1.207 0.968 0.410 0.531 312 25 1.132 1.117 0.393 0.570 341 25 1.314 1.225 0.333 0.518 330 25 1.299 1.264 0.381 0.580 373 25 1.213 1.293 0.337 0.494 316 26 1.026 1.269 0.336 0.525 324 26 1.077 1.187 0.319 0.533 335 27 1.125 1.299 0.432 0.636 423 30 1.119 1.259 0.342 0.575 405 33 1.073 1.052 0.367 0.526 349 33 1.017 1.230 0.355 0.532 344 34 1.061 1.250 0.324 0.506 344 34 1.206 1.306 0.362 0.513 379 34 1.200 1.287 0.388 0.584 420 34 1.161 1.342 0.441 0.701 478 34 1.158 1.302 0.347 0.578 428 34 1.113 1.265 0.389 0.643 417 34 1.025 1.220 0.352 0.558 384 34 1.109 1.220 0.359 0.609 435 34 1.014 1.212 0.332 0.564 356 34 1.131 1.233 0.347 0.505 362 34 1.116 1.309 0.359 0.543 372 34 1.153 1.271 0.402 0.570 393 34 1.150 1.284 0.316 0.495 333 34 1.110 1.349 0.385 0.554 360 34 1.067 1.190 0.334 0.462 276 35 1.212 1.321 0.425 0.739 541 35 1.123 1.247 0.330 0.514 345 43 1.101 1.246 0.303 0.533 376 43 1.105 1.301 0.391 0.570 437 43 1.102 1.233 0.345 0.547 364 43 1.101 1.196 0.319 0.523 341 44 1.129 1.329 0.353 0.575 425 48 1.290 1.211 0.378 0.605 454 49 1.198 1.037 0.468 0.602 460 49 1.181 1.314 0.414 0.633 475 49 1.243 1.305 0.343 0.634 465 49 1.105 1.327 0.404 0.638 467 50 1.212 1.289 0.398 0.607 451 50 1.155 1.281 0.308 0.467 346 50 1.073 1.273 0.320 0.573 398 55 1.118 1.143 0.400 0.619 477 55 1.154 1.251 0.371 0.525 411 56 1.188 1.253 0.374 0.626 466 56 1.190 1.254 0.349 0.603 395 57 1.110 1.261 0.352 0.563 390 60 1.194 1.061 0.369 0.572 371 60 1.314 1.097 0.342 0.601 408 61 1.113 1.084 0.330 0.584 422 62 1.075 1.084 0.401 0.570 378 62 1.175 1.126 0.440 0.600 487 62 1.181 1.255 0.363 0.586 418 63 1.264 1.320 0.412 0.646 479 63 1.043 1.254 0.328 0.543 388 63 1.137 1.322 0.438 0.644 487 69 1.216 1.152 0.393 0.611 475 69 1.081 1.086 0.325 0.598 414 69 1.191 1.134 0.411 0.581 483 69 1.253 1.103 0.365 0.642 471 70 1.257 1.369 0.441 0.656 588 70 1.174 1.277 0.381 0.609 469 70 1.211 1.316 0.427 0.650 490 70 1.240 1.307 0.440 0.546 475 72 1.207 1.177 0.455 0.647 533 72 1.239 1.157 0.443 0.659 570 72 1.149 1.272 0.364 0.628 473 72 1.222 1.307 0.362 0.590 431 72 1.214 1.347 0.385 0.590 513 72 1.204 1.345 0.474 0.632 546 73 1.301 1.366 0.413 0.678 534 73 1.227 1.233 0.464 0.638 503 80 1.108 1.088 0.384 0.607 423 80 1.218 1.072 0.343 0.581 467 80 1.053 1.146 0.399 0.553 431 81 1.201 1.241 0.378 0.568 458 81 1.149 1.239 0.347 0.604 433 81 1.150 1.281 0.368 0.549 418 83 1.140 1.302 0.402 0.581 522 88 1.236 1.356 0.471 0.696 582 89 1.268 1.067 0.448 0.702 550 90 1.216 1.348 0.374 0.652 556 108 1.244 1.161 0.415 0.700 616 109 1.233 1.350 0.426 0.668 589

The regression relationship between the age, body length, bust width, maximum width and weight of Korean cattle analyzed using the results in Table 1 is shown in Equation (1).

(1)

(One)

)(Coefficient of determination,

)

Table 2 shows the error between the estimated weight of Korean cattle and the measured value using the regression relationship in Equation (1).

Age (months) Weight (kg) error (%) Measured value Predicted value 6 155 149.7 3.4 6 159 153.4 3.5 6 172 178.8 4.0 6 150 157.5 5.0 6 165 175.3 6.3 6 149 149.5 0.3 13 240 249.2 3.9 13 190 202.6 6.6 13 231 246.1 6.5 18 257 274.6 6.8 18 260 260.9 0.3 18 292 285.6 2.2 18 320 320.8 0.2 18 361 352.3 2.4 18 223 218.5 2.0 18 305 303.0 0.7 21 315 317.4 0.8 22 304 312.8 2.9 22 323 336.8 4.3 22 403 372.1 7.7 22 306 306.9 0.3 24 382 381.8 0.1 24 300 302.6 0.9 24 308 303.3 1.5 24 312 330.1 5.8 25 341 363.2 6.5 25 330 339.7 3.0 25 373 398.6 6.9 25 316 327.7 3.7 26 324 328.2 1.3 26 335 320.4 4.4 27 423 443.7 4.9 30 405 372.9 7.9 33 349 325.0 6.9 33 344 344.3 0.1 34 344 325.5 5.4 34 379 363.0 4.2 34 420 412.0 1.9 34 478 505.4 5.7 34 428 392.1 8.4 34 417 437.2 4.8 34 384 359.4 6.4 34 435 399.3 8.2 34 356 354.0 0.6 34 362 336.7 7.0 34 372 372.8 0.2 34 393 402.5 2.4 34 333 328.0 1.5 34 360 393.7 9.4 34 276 294.4 6.7 35 541 525.2 2.9 35 345 338.3 2.0 43 376 350.3 6.8 43 437 411.8 5.8 43 364 372.3 2.3 43 341 343.6 0.8 44 425 409.1 3.7 48 454 439.3 3.2 49 460 439.6 4.4 49 475 475.6 0.1 49 465 455.0 2.2 49 467 470.6 0.8 50 451 454.9 0.9 50 346 332.3 4.0 50 398 392.4 1.4 55 477 442.6 7.2 55 411 393.0 4.4 56 466 460.2 1.3 56 395 437.3 10.7 57 390 410.1 5.1 60 371 405.7 9.3 60 408 428.5 5.0 61 422 397.1 5.9 62 378 412.1 9.0 62 487 458.1 5.9 62 418 440.5 5.4 63 479 511.8 6.8 63 388 392.0 1.0 63 487 509.1 4.5 69 475 465.6 2.0 69 414 413.5 0.1 69 483 449.6 6.9 69 471 471.0 0.0 70 588 545.0 7.3 70 469 475.2 1.3 70 490 524.9 7.1 70 475 468.4 1.4 72 533 516.9 3.0 72 570 520.0 8.8 72 473 481.1 1.7 72 431 468.2 8.6 72 513 481.5 6.1 72 546 537.6 1.5 73 534 555.6 4.0 73 503 525.8 4.5 80 423 459.1 8.5 80 467 435.6 6.7 80 431 435.3 1.0 81 458 463.9 1.3 81 433 469.5 8.4 81 418 449.6 7.5 83 522 486.4 6.8 88 582 603.1 3.6 89 550 563.1 2.4 90 556 541.9 2.5 108 616 589.8 4.3 109 589 600.6 2.0

The present invention is not limited to the disclosed embodiments, and it is obvious to those of ordinary skill in the art that various modifications and variations can be made without departing from the spirit and scope of the present invention. Therefore, such modified examples or modified examples will have to belong to the claims of the present invention.

A: ToF camera
B: Fixing bracket
C: expanded lava
D: Basic arm
F: tripod
G: computer

Claims (2)

ToF camera (A), fixing bracket (B), extension arm (C), base arm (D), counterweight (E), tripod (F), computer (G) to enable 3D image acquisition and image processing ; And
In the Hanwoo weight measuring apparatus, characterized in that the body length, body height, bust width and maximum width are measured from the obtained 3D image, and the weight of the Korean cattle is measured by substituting the age of the target Korean cattle into the following equation (1),
Since noise may be included in the point cloud image of the top view of the Korean cattle obtained using the ToF camera (A), the outliers due to noise are removed using the k-nearest neighbors algorithm, and then the point of the Korean cattle. Determine the height, length, bust width, yaw width and maximum width from the cloud image,
The noise removal of the k-nearest neighbors algorithm uses a moving average method to compare the average height value of 10 points and the height value of the corresponding point, and if the difference is 5 to 10 mm or more, it is determined as noise. Hanwoo weight measurement characterized by using a noise removal and boundary line search method set as a boundary when there is a height difference of 5 ~ 10 mm or more based on a point, but there is a difference within 5 ~ 10 mm from the adjacent dots left and right or up and down. Way.

------ Equation (1) delete

Images