CN110836715A - Moving body weight measurement system and moving body weight measurement method - Google Patents

Abstract

The present invention provides a moving body weight measurement system and a moving body weight measurement method. The moving body weight measurement system measures the weight of the moving body based on an image captured by a camera with respect to a moving body in space, comprising: an image acquisition unit that obtains the image captured by the camera; a recognition unit for recognizing the moving object from the image acquired by the image acquisition unit, and a viewing angle determination unit based on the mapping position of the camera in the image and the movement in the image The position and direction of the body are used to determine whether the viewing angle of the moving body in the image satisfies a predetermined condition; When the angle of view satisfies a predetermined condition, the weight of the moving body is calculated based on at least the size of the moving body in the image.

Description

Moving body weight measurement system and moving body weight measurement method

technical field

The present invention relates to a moving body weight measurement system and a moving body weight measurement method.

Background technique

In aquaculture, large numbers of animals are usually kept in cramped conditions. For example, in pig farms, pigs are raised on a large scale, and the pigs are released when they reach slaughter weight. Therefore, the weight of pigs is a very important indicator for pig farms. During the feeding process, the weight of pigs presents a curve. In the growth stage, the increase in weight is proportional to the amount of feed that is put in. The more feed is put in, the more the weight of pigs increases, but once it reaches the adult stage, Its weight gain becomes slower, and the output is less relative to the feed put in. Therefore, it is very important to judge the most appropriate time to slaughter.

In addition, from the perspective of selecting appropriate feed and adopting appropriate feeding methods, real-time determination of pig body weight also has good guiding significance.

In contrast to the conventional technique of judging whether it is time to slaughter by visual observation of the weight of pigs based on human experience, for example, Patent Document 1 proposes a technique of estimating the weight of live pigs from images obtained by photographing live pigs. Specifically, install the stereo camera on the upper track and make it move, so that the pigs wear wireless RFID tags, and set the RFID card reader at the drinking place of the pigpen. When the camera passes from above, if the RFID card reader detects To RFID, the camera takes a picture, and then, based on the obtained image, the size of the pig is measured and the weight is estimated.

Patent Document 1: CN103884280A

However, in the technique described in Patent Document 1, it is necessary to build rails, so that the system configuration is complicated. Furthermore, since the measurement conditions are strict, the success rate is limited.

SUMMARY OF THE INVENTION

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a moving body weight measurement system and a moving body weight measurement method capable of automatically measuring the weight of a moving body with a simpler structure than conventional ones.

In order to achieve the above object, a moving body measurement system of the present invention measures the weight of a moving body in space based on an image captured by a camera, and includes an image acquisition unit that obtains an image captured by the camera. the image; a moving object recognition unit for recognizing the moving object from the image acquired by the image acquisition unit; a viewing angle determination unit based on the mapping position of the camera in the image and the image The position and direction of the moving body in the image are determined, whether the viewing angle of the moving body in the image satisfies a predetermined condition; When the viewing angle of the moving body satisfies a predetermined condition, the weight of the moving body is calculated based on at least the size of the moving body in the image.

According to the mobile body weight measurement system of the present invention, the weight of the mobile body can be measured without installing another hardware device. Furthermore, by judging whether the camera's viewing angle of a moving object (eg, a pig) satisfies a predetermined condition, the body weight can be calculated at a viewing angle suitable for calculating the body weight, and the accuracy of the measurement of the body weight can be improved.

In addition, the above-mentioned moving body weight measurement system may further include a blocking judgment unit that judges whether the moving body recognized by the moving body recognition unit is blocked; The determination unit calculates the weight of the moving body when it is determined that the viewing angle of the moving body in the image satisfies a predetermined condition and the blocking determination unit determines that the moving body is not blocked.

Thereby, by calculating the weight of the moving body only when the moving body is not blocked, it is possible to further improve the measurement accuracy of the weight.

In addition, the above-mentioned moving body weight measurement system may further include an image processing unit that, when the angle of view of the moving body in the image is determined by the angle of view determination unit to satisfy a predetermined condition, and is determined by the angle of view determination unit When the blocking determination unit determines that the moving body is not blocked, the image is subjected to distortion correction processing according to the positional relationship between the camera and the moving object and the characteristics of the camera; the weight calculation The unit calculates the weight of the moving body based on the image subjected to distortion correction processing by the image processing unit.

Accordingly, even when a camera such as a fisheye camera is used, the weight of the moving body can be measured with high accuracy by calculating the weight of the moving body from the distortion-corrected image.

In addition, the predetermined condition may be that a center point perpendicular to the spine axis of the movable body when a perpendicular line is drawn from the camera mapping position in the image to the spine axis of the movable body as a reference is at within the specified range.

Furthermore, the predetermined condition may be that a vertical foot when a vertical line is drawn from the camera-mapped position in the image to the spine axis of the movable body is a center point of the spine axis of the movable body.

This makes it possible to calculate the body weight using the image of the moving body at a point in time when the moving body is in an appropriate (satisfying predetermined condition) angle of view during the movement of the moving body. Therefore, the accuracy of measuring the body weight can be improved.

Furthermore, when the camera is a monocular camera, the weight calculation unit may calculate the weight based on the pixel amount of the moving object in the image and the actual distance from the camera to the moving object. the weight of the moving body.

Furthermore, when the camera is a binocular camera, the weight calculation unit may calculate the weight of the moving body based on the body length and abdominal diameter of the moving body obtained from the image.

Thereby, the weight of the moving body can be measured more appropriately according to different types of cameras.

The camera is mounted above the center of the space.

Thereby, the success rate with which the weight of a moving body can be measured can be improved.

In addition, in order to achieve the above-mentioned object, the moving body weight measurement system of the present invention measures the weight of the moving body in a plurality of spaces based on images captured by a camera, comprising: an image acquisition unit that measures the weight of the moving body according to the plurality of spaces. In each of the spaces, obtain at least one image captured by the camera installed in the space; the processor calculates the weight of the moving body in the at least one image and outputs it; the processor presses For each image, the following processing is performed: from the image, a plurality of moving objects are identified; for each identified moving object, based on the mapping position of the camera in the image and the the position and direction of the moving object, and determine whether the viewing angle of the moving object in the image satisfies the prescribed condition; and when it is determined that the viewing angle of the moving object in the image satisfies the prescribed condition, at least according to The size of the moving body in the image is calculated, and the weight of the moving body is calculated.

This makes it possible to measure the weight of the moving body for each space even when there are a plurality of spaces in which the moving body exists, without separately installing another hardware device. Furthermore, by judging whether the camera's viewing angle of a moving object (eg, a pig) satisfies a predetermined condition, the body weight can be calculated at a viewing angle suitable for calculating the body weight, and the accuracy of the measurement of the body weight can be improved.

In addition, the present invention can also be implemented as a method for measuring the weight of a moving body, and can achieve the same effects as the above-mentioned system for measuring the weight of a moving body.

Description of drawings

FIG. 1 is a block diagram showing a moving body weight measurement system according to the first embodiment.

FIG. 2 is a flowchart showing a method for measuring the weight of a moving body according to the first embodiment.

FIG. 3 is a diagram showing an example of an image acquired by the moving body weight measurement system.

FIG. 4 is a diagram showing an example of a method for identifying a moving body by the moving body weight measurement system.

FIG. 5 is a diagram showing the principle of viewing angle judgment.

FIG. 6 is a diagram showing an example of viewing angle determination processing by the moving body weight measurement system.

7 is a block diagram showing a moving body weight measurement system according to a second embodiment.

8 is a block diagram showing a moving body weight measurement system according to a third embodiment.

9 is a diagram showing a distortion correction process performed by an image processing unit in the moving body weight measurement system according to the third embodiment.

Fig. 10 is a diagram showing the configuration of a moving body weight measurement system in which a plurality of pigsties are present.

Detailed ways

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

first embodiment

First, the moving body weight measurement system of the present embodiment will be described. FIG. 1 is a block diagram showing a moving body weight measurement system according to the present embodiment. As shown in FIG. 1 , the moving body weight measurement system 1 includes an image acquisition unit 10 , a moving body recognition unit 20 , a viewing angle determination unit 30 , and a weight calculation unit 40 .

The image acquisition unit 10 acquires an image obtained by capturing a space in which a moving object is located by a camera. In this embodiment, for convenience of description, it is assumed that the moving body is a pig, and the space where the moving body is located is a pigpen. In addition, the camera is installed above the center of the pigsty and can photograph the entire interior of the pigsty.

The moving object recognition unit 20 recognizes a pig from the pigpen image acquired by the image acquisition unit 10 . The method of identifying pigs can be any well-known method such as deep learning based on neural network, edge detection, etc. In addition, the pigs identified from the pigpen images can be represented by the corresponding rectangles or by the outlines of the detected pigs.

The angle of view determination unit 30 determines whether the angle of view of the pig in the pigsty image satisfies a predetermined condition based on the mapping position of the camera in the pigsty image and the position and direction of the pig in the pigsty image. Thus, it is determined whether the angle of view of the pig in the pigsty image is suitable for the subsequent body weight calculation. In simple terms, if the perspective of the pig in the pigpen image is that the camera is facing the side of the pig, it is the most suitable situation for weight calculation. If the perspective of the pig in the pigpen image is that the head or tail of the pig is facing the camera, It is the most inappropriate situation for weight calculation. A specific determination method will be described later.

The weight calculation unit 40 calculates the weight of the pig based on at least the size of the pig in the pigpen image when the angle of view determination unit 30 determines that the angle of view to the pig in the pigpen image satisfies the predetermined condition. That is, the body weight of the pig is calculated by the body weight calculation unit 40 only when the angle of view judgment unit 30 judges that the body weight calculation is suitable.

Next, the operation flow of the moving body weight measurement method by the moving body weight measurement system will be described with reference to FIGS. 2 to 6 .

FIG. 2 is a flowchart showing a method for measuring the weight of a moving body according to the present embodiment.

As shown in FIG. 2 , first, the image acquisition unit 10 acquires a pigpen image obtained by photographing a pigpen with a camera (step S1 ). An example of acquiring a pigpen image is shown in FIG. 3 . As shown in Figure 3, the camera is installed above the center of the pigsty and can photograph the entire interior of the pigsty.

Next, the moving object recognition part 20 recognizes a pig from the pigpen image acquired by the image acquisition part 10 (step S2). An example of a method for identifying pigs is shown in FIG. 4 . The deep learning method based on neural network is adopted in Fig. 4. Specifically, a neural network is pre-built to perform feature vector-based deep learning on pigs in various poses, so that pigs can be identified from pigpen images. The specific presentation forms for the identified pigs can be various, which are represented by rectangular boxes in FIG. 4 . In addition, in fact, the image captured by the camera installed above the center of the pigpen and identified the pig is shown on the right side of Fig. 3 .

Next, the angle of view determination unit 30 judges whether the angle of view of the pig in the pigsty image satisfies a predetermined condition based on the mapping position of the camera in the pigsty image and the position and direction of the pig in the pigsty image (step S3 ).

When it is determined that the angle of view of the pig in the pigpen image does not satisfy the predetermined condition (NO in step S3 ), the process ends.

When it is determined that the angle of view to the pig in the pigpen image satisfies the predetermined condition (YES in step S3 ), the process proceeds to step S4 .

Here, the principle of the processing performed by the viewing angle determination unit 30 will be described in detail below.

FIG. 5 is a diagram showing the principle of viewing angle judgment. In Figure 5, the same pig is photographed from different perspectives, and its size is different. The angle of view perpendicular to the center point of the pig's spine axis is the best shooting angle of view, and the corresponding camera is Camera 1. At this time, the pixel quantity of the photographed pig is the largest, which has a high reference value and is considered suitable for weight calculation. On the other hand, if the image is taken from the camera 4 in the direction of the pig's head or the camera 5 in the direction of the pig's tail, the error is relatively large, and it is considered that the calculation of the body weight is not suitable. In addition, although the positions of the camera 2 and the camera 3 are not in the angle of view perpendicular to the center point of the vertebral axis of the pig, if the error is within the allowable range, it can be considered that the calculation of the body weight is suitable.

Based on such a principle, it is possible to judge the angle of view of the pig identified in step S2.

FIG. 6 is a diagram showing an example of viewing angle determination processing. In Figure 6, it is assumed that the camera installed above the center of the pigpen captured three pigs. At this time, since the camera is installed above the center of the pigsty, the mapping position of the camera in the pigsty image is the center O of the screen, which is hereinafter referred to as the camera mapping position O. For example, in the case of Pig 1, it is assumed that the vertical foot R1 when the vertical line OR1 is drawn from the camera mapping position O to the spine axis of Pig 1 is the center point of the spine axis. This case is considered to be a case where the viewing angle of the pig satisfies the condition, and it is considered that the subsequent body weight calculation can be performed. However, the case where the condition is satisfied is not limited to this, and it can be considered that the condition is satisfied as long as the perpendicularity when the vertical line is drawn from the camera mapping position to the pig's spine axis is within a predetermined range based on the center point of the spine axis. The specified range can be appropriately set according to accuracy requirements and experience. Here, the viewing angle in the case of Pig 1 is the most preferable viewing angle, and the calculated body weight is the most accurate at this time.

In addition, for example, in the case of pig 2 and pig 3, it is assumed that the distance between the vertical feet R2 and R3 from the center point of the spine axis when the vertical lines OR2 and OR3 are drawn from the camera mapping position O to the respective spine axes of pig 2 and pig 3 is large. , is not within the specified range, so it is determined that the viewing angles of pigs 2 and 3 do not satisfy the specified conditions, and the subsequent body weight calculation is not performed.

As described above, the viewing angle determination unit 30 determines that the predetermined condition is satisfied when the vertical line when the vertical line is drawn from the camera-mapped position in the image to the spine axis of the pig is within the predetermined range based on the center point of the spine axis. Preferably, the vertical foot when drawing a vertical line from the camera-mapped position to the vertebral axis of the pig is the center point of the vertebral axis. In addition, when the center point of the vertical axis of the spine as a reference is outside the predetermined range, it is determined that the predetermined condition is not satisfied.

In this way, the body weight can be calculated using the image of the pig at a time point when the pig is in a certain suitable (satisfying the prescribed condition) viewing angle during the movement of the pig. Therefore, the accuracy of measuring the body weight can be improved. Moreover, by installing the camera above the center of the pigpen, the possibility that the angle of view of the photographed pigs satisfies the conditions is improved. Therefore, the success rate at which the body weight can be measured becomes high.

Returning to FIG. 2 , in step S4 , the weight calculation unit 40 calculates the weight of the pig based on at least the size of the pig in the pigpen image.

For example, when the camera is a monocular camera, the weight M can be calculated by the following (Equation 1).

Among them, E is an empirical constant, which can be set by statistical experiments or experience. P is the number of pixels occupied by pigs in the pigsty image, which reflects the size of the pigs in the pigsty image. d is the actual distance from the camera to the pig, which can be obtained from the positional relationship between the camera and the pig in the pigpen image and the installation position of the camera.

In addition, when the camera is a binocular camera, the weight M can be calculated by the following (Equation 2).

M=Xld ² ......(Formula 2)

Among them, X is an empirical constant, which can be set by statistical experiments or experience. l is the length of the pig, reflecting the size of the pig, and d is the diameter of the abdomen (abdominal thickness).

In this way, the weight of the pig can be calculated from different kinds of cameras.

As described above, in this embodiment, it is not necessary to provide additional hardware equipment, and by judging whether the angle of view of the camera to the pig satisfies the predetermined condition, the body weight can be calculated at the angle of view suitable for calculating the body weight, and the measurement of body weight can be improved. accuracy.

Second Embodiment

Hereinafter, a moving body weight measurement system according to a second embodiment of the present invention will be described with reference to FIG. 7 .

7 is a block diagram showing a moving body weight measurement system according to a second embodiment.

As shown in FIG. 7 , in the present embodiment, the moving body weight measurement system 2 is different from the first embodiment in that it includes an occlusion determination unit 50 . In addition, the moving body weight measurement system 2 of this embodiment is the same as that of Embodiment 1 unless otherwise specified, and the same reference numerals are assigned to the same configuration, and the detailed description of the configuration is omitted.

The blocking determination unit 50 determines whether or not the pig recognized by the moving object recognition unit 20 is blocked. Here, the case of not being blocked includes not only the case of not being blocked at all, but also the case of being partially blocked but not affecting the calculation of the weight. For example, when only the legs are occluded.

Then, the body weight calculation unit 40 calculates the body weight when it is determined by the angle of view determination unit 30 that the camera's angle of view of the pig satisfies the predetermined condition and the occlusion determination unit 50 determines that the pig is not obscured.

In addition, the respective determination processes of the viewing angle determination unit 30 and the occlusion determination unit 50 may be performed in parallel, or may be performed sequentially. For example, the viewing angle determination unit 30 may first perform the viewing angle determination processing, and the blocking determination unit 50 may determine that the moving object identification unit 20 recognizes the pig only when the viewing angle determination unit 30 determines that the angle of view of the pig satisfies a predetermined condition. whether the pig is occluded.

According to the moving body weight measurement system of the present embodiment, it is possible to select a state more suitable for weight calculation by judging whether or not the pig is blocked. Furthermore, the measurement accuracy of the body weight can be further improved.

Third Embodiment

Hereinafter, a moving body weight measurement system according to a third embodiment of the present invention will be described with reference to FIG. 8 .

8 is a block diagram showing a moving body weight measurement system according to a third embodiment.

As shown in FIG. 8 , in the present embodiment, the moving body weight measurement system 3 is different from the second embodiment in that it includes an image processing unit 60 . The moving body weight measurement system 3 of the present embodiment is the same as that of the second embodiment unless otherwise specified, and the same components are assigned the same reference numerals, and detailed descriptions of the components are omitted.

When the image processing unit 60 determines that the angle of view of the pig in the pigpen image satisfies the predetermined condition, and the occlusion judging unit 50 determines that the pig is not occluded by the angle of view judgment unit 30, the image processing unit 60 determines the positional relationship between the camera and the pig according to the positional relationship between the camera and the pig. The characteristics of the camera are used to correct the distortion of the image.

Then, the weight calculation unit 40 calculates the weight of the pig based on the image subjected to the distortion correction processing by the image processing unit 60 .

FIG. 9 is a diagram showing a distortion correction process performed by an image processing unit.

As shown in FIG. 9 , in the original image captured by some cameras, such as a fisheye camera, the distortion is usually relatively small in the middle part, and the distortion is more serious as it goes to the edge. These distortions become factors that affect the measurement accuracy, so they need to be corrected. As a distortion correction method, a specific area of interest is first determined, and then the distortion is restored by detecting the feature points in the area, so as to obtain a corrected image.

In the present embodiment, by correcting the distortion in the pigsty image, even when a wide-angle lens is used due to the low ceiling of the pigsty, the body weight can be calculated using the corrected image with the distortion removed, so it is possible to further Improve weight measurement accuracy.

In addition, in the present embodiment, a configuration in which an image processing unit is added to the second embodiment has been described. However, it is not limited to this, and an image processing unit may be added to the first embodiment. At this time, the image processing unit performs a distortion correction process on the image when the angle of view determination unit determines that the angle of view of the pig in the pigsty image satisfies a predetermined condition.

Other implementations

In the above-described first to third embodiments, for the convenience of description, the case where there is only one pigsty and the angle of view of only one pig in the pigsty image satisfies the predetermined conditions has been described. However, it is not limited to this, and the above-described first to third embodiments can of course be applied to the case where there are a plurality of pig pens, and there are many pigs in each pig pen.

Fig. 10 is a diagram showing the configuration of a moving body weight measurement system in which a plurality of pigsties are present. As shown in FIG. 10 , when there are multiple pigsties, cameras are installed in each pigsty, and the first to One of the body weight measurement processes in the third embodiment outputs the result. For example, in the case of the body weight measurement process of the first embodiment, multiple pigs are identified for each image (in the case of a video, each frame of the image constituting the video), and the angle of view is determined for each identified pig. Process, and calculate the body weight of the pig if the viewing angle meets the specified conditions. Here, the multiple pigs in the pigpen can be identified by any well-known method such as marking the pig's body, or the cooperation of the tag and the card reader, so as to be able to identify the pigs in the pigpen and the output of the system. result to match.

In addition, in each of the above-mentioned embodiments, the camera was described as being attached to the upper center of the pigpen. However, it is not limited to this, as long as the entire interior of the pigpen can be photographed, it can be installed at any position at any angle. At this time, the mapping position of the camera on the screen is not in the center of the screen, so it is necessary to measure the mapping position of the camera on the screen separately.

In addition, in each of the above-mentioned embodiments, the pig was described as an example. However, it is not limited to this, and other moving objects such as animals may be used.

In addition, each component in each of the above-described embodiments may be configured by dedicated hardware, may be implemented by a software program that executes each component, or may be implemented by a combination of hardware and software. Further, each of the constituent elements may be realized by a program execution unit such as a CPU or a processor reading and executing a software program recorded on a recording medium such as a hard disk or a semiconductor memory.

Claims (10)

1. A moving body weight measurement system, for a moving body in space, according to an image captured by a camera, to measure the weight of the moving body, characterized in that, comprising: an image acquisition unit that acquires the image captured by the camera; a moving object recognition unit for recognizing the moving object from the image acquired by the image acquisition unit; A viewing angle determination unit for determining whether the viewing angle of the moving object in the image satisfies a predetermined condition based on the mapping position of the camera in the image and the position and direction of the moving object in the image ;as well as A body weight calculation unit, when the angle of view determination unit determines that the angle of view of the moving object in the image satisfies a predetermined condition, calculates the The weight of the moving body. 2. The moving body weight measurement system according to claim 1, wherein Also provided with an occlusion determination unit, the occlusion determination unit determines whether the moving body identified by the moving object recognition unit is blocked; the weight calculation unit when it is determined by the viewing angle determination unit that the viewing angle of the moving body in the image satisfies a predetermined condition, and the blocking determination unit determines that the moving body is not blocked , calculate the weight of the moving body. 3. The moving body weight measurement system according to claim 2, wherein and further comprising an image processing unit that is determined by the viewing angle determination unit to satisfy a predetermined condition when the viewing angle of the moving object in the image satisfies a predetermined condition and determined by the occlusion determination unit that the moving object does not In the case of being blocked, perform distortion correction processing on the image according to the positional relationship between the camera and the moving object and the characteristics of the camera; The weight calculation unit calculates the weight of the moving body based on the image subjected to distortion correction processing by the image processing unit. 4. The moving body weight measurement system according to any one of claims 1 to 3, wherein The predetermined condition is that a vertical line drawn from the camera-mapped position in the image to the spine axis of the moving body is within a predetermined range based on the center point of the spine axis of the moving body. 5. The moving body weight measurement system according to any one of claims 1 to 3, wherein The predetermined condition is that the vertical foot when drawing a vertical line from the camera-mapped position in the image to the spine axis of the moving body is the center point of the spine axis of the moving body. 6. The moving body weight measurement system according to any one of claims 1 to 3, wherein When the camera is a monocular camera, the weight calculation unit calculates the weight of the moving body based on the pixel amount of the moving body in the image and the actual distance from the camera to the moving body. 7. The moving body weight measurement system according to any one of claims 1 to 3, wherein When the camera is a binocular camera, the weight calculation unit calculates the weight of the moving body based on the length and abdominal diameter of the moving body obtained from the image. 8. The moving body weight measurement system according to any one of claims 1 to 3, wherein The camera is mounted above the center of the space. 9. A system for measuring the weight of a moving body, measuring the weight of the moving body according to an image captured by a camera for a moving body in a plurality of spaces, characterized in that it comprises: an image acquisition unit for acquiring, for each of the plurality of spaces, at least one image captured by the camera installed in the space; a processor to calculate and output the weight of the moving body in the at least one image; The processor performs the following processing for each image: from the image, identifying a plurality of moving objects; For each identified moving object, according to the mapping position of the camera in the image and the position and direction of the moving object in the image, determine whether the viewing angle of the moving object in the image satisfies stipulated conditions; and When it is determined that the viewing angle of the moving body in the image satisfies a predetermined condition, the weight of the moving body is calculated based on at least the size of the moving body in the image. 10. A method for measuring the weight of a moving body, for a moving body in space, measuring the weight of the moving body according to an image captured by a camera, characterized in that it comprises: an image acquisition step of acquiring the image captured by the camera; a moving body recognizing step for identifying the moving body from the image obtained by the image obtaining step; The angle of view judgment step is to judge whether the angle of view of the moving body in the image satisfies a predetermined condition according to the mapping position of the camera in the image and the position and direction of the moving body in the image ;as well as A weight calculation step of calculating the body weight based on at least the size of the moving object in the image when it is determined by the viewing angle determination step that the viewing angle of the moving object in the image satisfies a predetermined condition The weight of the moving body.

Images