An automatic recognition and recording system for abnormal breathing behaviors of rabbits

Abstract

The present invention relates to the technical field of rabbit behavior recognition, and particularly to an automatic recognition and recording system for abnormal rabbit breathing behaviors. Its technical solution includes an image acquisition module, a data transmission module, a data analysis module, and a data storage and interaction module. The image acquisition module is used to obtain the image data and environmental data of the rabbit. The data transmission module is responsible for transmitting the data collected by the image acquisition module to the data analysis module. The data analysis module processes and analyzes the received data to identify the abnormal breathing behaviors of the rabbit. The data storage and interaction module is used to store data and implement interactive operations with users. The present invention optimizes the whole process from image acquisition, data transmission, analysis to storage and interaction, significantly improving the accuracy, reliability, and convenience of recognizing abnormal rabbit breathing behaviors, effectively assisting the refined management and health guarantee of rabbit breeding, and bringing considerable economic benefits and efficient management experiences to breeders.

Description

Technical Field

[0001] The present invention relates to the technical field of rabbit behavior recognition, and particularly to an automatic recognition and recording system for abnormal rabbit breathing behavior. Background Art

[0002] In the process of rabbit breeding, the health status of rabbits plays a crucial role in breeding efficiency. Abnormal breathing is often one of the important signals indicating health problems in rabbits. Timely and accurately identifying the abnormal breathing behavior of rabbits can provide key evidence for early disease diagnosis and treatment, thereby effectively reducing the mortality rate of rabbits and minimizing the economic losses of breeders.

[0003] Traditional rabbit health monitoring mainly relies on manual observation by breeders. However, this method has many limitations. First, it is difficult to maintain continuous and uninterrupted observation. Breeders cannot focus on the breathing status of a single rabbit for a long time, which easily leads to missed detection of abnormal breathing behavior. Second, there are errors in human subjective judgment. Different breeders may have different criteria for judging abnormal breathing, and they are prone to fatigue and negligence after long-term observation, making the judgment results inaccurate and unreliable. Moreover, manual observation can only be carried out when the breeder is present, and it is impossible to remotely monitor the breathing status of rabbits, which is not conducive to refined management in large-scale breeding scenarios.

[0004] With the continuous development of technology, some automated monitoring technologies have begun to be applied in the field of animal health monitoring. For example, some existing technologies use a single ordinary optical camera to collect animal images and infer the breathing situation by analyzing the movement of the animal's body in the images. However, this method has deficiencies in obtaining the depth information of the rabbit's body and is difficult to accurately locate the key areas related to breathing, such as the accurate position and movement amplitude of the chest cavity, thus affecting the accuracy of breathing feature extraction. In addition, in terms of data transmission, the transmission rate of some traditional monitoring systems is low or the stability is poor, and problems such as data loss or transmission delay are likely to occur, resulting in a lag in the recognition of abnormal breathing behavior. In terms of data analysis algorithms, the algorithms used in some existing technologies are relatively simple, the comprehensive judgment of breathing frequency, breathing depth, and breathing rhythm is not precise enough, and there is a lack of effective compensation and adaptation mechanisms for complex environmental factors (such as light changes, temperature and humidity fluctuations, etc.), which cannot meet the requirements of high-precision recognition of abnormal rabbit breathing behavior.

[0005] In summary, there is an urgent need for an automatic recognition and recording system for abnormal rabbit breathing behavior that can overcome the defects of existing technologies and achieve high-precision, automation, remote operation, and stable and reliable performance, so as to improve the health management level and breeding efficiency of rabbit breeding. Summary of the Invention

[0006] The object of the present invention is to address the problem in the background art of the lack of an automatic recognition and recording system for abnormal breathing behaviors of rabbits with high precision and stability. The present invention proposes an automatic recognition and recording system for abnormal breathing behaviors of rabbits.

[0007] The technical solution of the present invention: An automatic recognition and recording system for abnormal breathing behaviors of rabbits, comprising:

[0008] An image acquisition module, used to acquire image data and environmental data of the rabbit;

[0009] A data transmission module, responsible for transmitting the data collected by the image acquisition module to the data analysis module;

[0010] A data analysis module, which processes and analyzes the received data to identify abnormal breathing behaviors of the rabbit;

[0011] A data storage and interaction module, used to store data and implement interaction operations with the user.

[0012] Optionally, the image acquisition module includes:

[0013] An image acquisition unit, which combines a high-resolution, high-frame-rate depth camera and a common optical camera. The resolution of the depth camera is 1280×720 pixels, the frame rate is 60fps, the resolution of the common optical camera is 1920×1080 pixels, the frame rate is 30fps. The camera is equipped with automatic exposure, automatic white balance functions and an 8 - 24mm variable focal length lens, and is installed on a movable robotic arm. The moving range of the robotic arm in the X-axis direction is 0 - 2 meters, in the Y-axis direction is 0 - 1.5 meters, and in the Z-axis direction is 0.5 - 1.2 meters;

[0014] An environmental monitoring unit, including a temperature and humidity sensor with an accuracy of ±0.3°C, a temperature measurement range of -10°C to 40°C and a humidity measurement range of 20% to 90%, and a light sensor with a measurement range of 0 - 20000 lux.

[0015] Optionally, the data transmission module includes:

[0016] A wired transmission unit, which uses a high-speed Ethernet cable for data transmission, follows the TCP / IP protocol, and the transmission rate reaches 1Gbps or more;

[0017] A wireless transmission unit, equipped with a Wi-Fi 6 wireless transmission module, follows the IEEE 802.11ax standard, the theoretical transmission rate is up to 9.6Gbps at most, the transmission frequency band is a dual band of 2.4GHz and 5GHz, and the wired transmission unit and the wireless transmission unit can be automatically switched, and the switching time is 8 - 10 seconds.

[0018] Optionally, the data analysis module includes an image preprocessing unit, and the image preprocessing unit specifically includes:

[0019] An image registration subunit that registers the images collected by the depth camera and the optical camera, and adopts a feature point-based registration algorithm;

[0020] A noise removal subunit that removes noise from the registered image by using a bilateral filtering algorithm.

[0021] Optionally, in the image registration subunit, feature points in the two images are extracted by using the scale-invariant feature transform algorithm. For the feature points in the depth image and the feature points in the optical image , the matching relationship between them is calculated. Let the error threshold for feature point matching be pixels. The matching error is optimized by using the least squares method, so that

[0022] where, is the three-dimensional coordinate of the rd feature point in the depth image, expressed as , where is the depth value, is the two-dimensional coordinate of the st feature point in the optical image, is the error threshold for feature point matching, with the unit of pixel, is the number of pairs of matching feature points, so as to obtain an accurate image registration result.

[0023] Optionally, in the noise removal subunit, the bilateral filtering formula is:

[0024] where, is the intensity value of the pixel after filtering, is the original intensity value of the pixel point , is the pixel 's neighborhood, is the spatial domain Gaussian function, which is used to calculate the weight according to the spatial distance between pixels, with the standard deviation , is the range domain Gaussian function, which is used to calculate the weight according to the pixel intensity difference , with the standard deviation , is the normalization coefficient,

[0025] .

[0026] Optionally, the data analysis module further includes a rabbit breathing feature extraction unit, which includes:

[0027] A breathing area positioning subunit, the seed point , set the growth threshold m, and then starting from the seed point, merge the adjacent pixel points whose depth values differ from the seed point depth value within the growth threshold range into the growth area, and continuously iterate until no new pixel points meet the conditions. The area threshold of the growth area is set to square centimeters, and by comparing with the preset rabbit body model, the accurate breathing area is further determined;

[0028] A breathing frequency calculation subunit analyzes the image sequence of the breathing area, and uses the optical flow method to calculate the movement speed of the breathing area. Let the pixel points in the breathing area of the th frame and the th frame in the image sequence have an optical flow vector of

[0029] where is the image intensity function, indicating the change of the brightness value of a certain point in the image with time ( ) and space . , are the optical flow vector components, respectively representing the movement speeds of the pixel points in the and directions. Set the optical flow speed threshold m / s. When the optical flow speeds of more than 30% of the pixel points in the breathing area exceed the optical flow speed threshold, it is recorded as a breathing action. The breathing frequency , where is the number of breathing actions detected within the time , is the total monitoring time. The normal breathing frequency range of rabbits is 30 - 60 times per minute. If it exceeds this range, the breathing frequency is determined to be abnormal;

[0030] A breathing depth analysis subunit calculates the maximum change in the depth of the chest area during breathing according to the depth image sequence of the breathing area, and sets the depth change threshold m. If the change in the chest depth during breathing exceeds the depth change threshold, it is determined that the breathing depth is abnormal;

[0031] Optionally, the data analysis module further includes an abnormal behavior recognition unit, and the abnormal behavior recognition unit is provided with a breathing rhythm judgment subunit to analyze the time interval sequence , calculate the autocorrelation function of the time interval:

[0032] where is the value of the autocorrelation function at the lag order , which is used to measure the periodicity of the respiratory time interval sequence. The time interval sequence average value of is the total number of the time interval sequence. is the lag order. Set the autocorrelation function threshold . When the autocorrelation function is less than this threshold within the lag order range , it is determined that the respiratory rhythm is abnormal.

[0033] Optionally, the data storage and interaction module includes:

[0034] A data storage unit that stores the original image data using a distributed file system and classifies and stores it according to date, rabbit number, and camera type, and stores the analysis data using a relational database, including rabbit individual information, respiratory feature data, and abnormal behavior records. The data storage time is not less than 1 year.

[0035] A user interaction unit that develops a Web-based user interface. Users can log in through a browser to view the rabbit respiratory monitoring data and abnormal behavior reports. The interface provides a real-time data display function, which displays the rabbit's respiratory rate and respiratory depth changes in the form of charts, and at the same time displays the environmental temperature, humidity, and light intensity information. Users set the alarm threshold parameters on the interface. When an abnormal behavior of the rabbit occurs, the system sends a text message alarm to the administrator through the text message gateway.

[0036] Optionally, the text message content includes the rabbit number, abnormal type, and detailed occurrence time information, and the alarm information is prominently displayed on the user interface so that the administrator can take measures in a timely manner.

[0037] Compared with the prior art, the present application includes at least one of the following beneficial technical effects:

[0038] 1. Through the combination of depth and optical cameras and the environmental perception function, accurate images are provided for respiratory analysis and it adapts to various environments, reducing environmental interference.

[0039] 2. The data transmission is efficient and stable. The wired high-speed transmission and wireless backup switching ensure the timely and accurate transmission of data, avoiding monitoring errors caused by transmission interruptions.

[0040] 3. The data analysis is accurate. Advanced algorithms enable accurate preprocessing of images, accurate extraction of respiratory features, and accurate identification of multi-dimensional abnormal behaviors, improving the timeliness and accuracy of discovering health problems.

[0041] 4. The data storage is reliable and the user interaction is convenient, which is conducive to data management traceability and long-term health tracking, facilitating breeders to monitor the health status of rabbits in real time and flexibly set alarms, thus improving the breeding management efficiency and the survival rate of rabbits.

[0042] The present invention optimizes the whole process from image acquisition, data transmission, analysis to storage and interaction, significantly improving the accuracy, reliability and convenience of identifying abnormal breathing behaviors of rabbits, effectively facilitating the refined management and health guarantee of rabbit breeding, and bringing considerable economic benefits and efficient management experiences to breeders. Brief Description of the Drawings

[0043] Figure 1 It is a principle block diagram of a system for automatically identifying and recording abnormal breathing behaviors of rabbits. Detailed Embodiments

[0044] The following specific examples illustrate the embodiments of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0045] Embodiment: The system for automatically identifying and recording abnormal breathing behaviors of rabbits disclosed in this application consists of an image acquisition module, a data transmission module, a data analysis module, and a data storage and interaction module. The following is a detailed description of each module.

[0046] I. Image Acquisition Module

[0047] Image Acquisition Unit

[0048] A combination of a high-resolution, high-frame-rate depth camera and a common optical camera is adopted. The depth camera has a resolution of 1280×720 pixels and a frame rate of 60fps, which can accurately obtain the three-dimensional information and depth data of the rabbit's body; the common optical camera has a resolution of 1920×1080 pixels and a frame rate of 30fps, which is used to capture the texture and detail features of the rabbit's body surface.

[0049] The cameras are equipped with automatic exposure and automatic white balance functions, which can automatically adjust the exposure time and color balance according to the ambient light intensity to ensure stable image quality. The lens focal length of the cameras is an 8 - 24mm variable focal length lens, which can adapt to the shooting requirements at different distances.

[0050] Installed on a movable robotic arm, the movement range of the robotic arm is 0 - 2 meters in the X-axis direction, 0 - 1.5 meters in the Y-axis direction, and 0.5 - 1.2 meters in the Z-axis direction. The camera position can be automatically adjusted through a preset program to comprehensively cover the activity area of the rabbit. A combination of a high-resolution, high-frame-rate depth camera and an ordinary optical camera is adopted. The depth camera can accurately obtain the three-dimensional information and depth data of the rabbit's body, and the ordinary optical camera can capture the texture and detailed features of the rabbit's body surface. The two work together to provide a rich and accurate image data basis for subsequent precise respiratory feature analysis. For example, when positioning the respiratory area of the rabbit's chest, the depth information helps to more accurately determine its position and range, while the texture details can assist in judging the subtle changes on the body surface caused by the respiratory movement.

[0051] The camera is equipped with automatic exposure and automatic white balance functions as well as a variable-focus lens, which can automatically adjust the exposure time and color balance according to the ambient light intensity, adapt to the shooting requirements at different distances, and ensure stable image quality. Regardless of how the lighting conditions in the breeding environment change, clear and accurate images of the rabbit can be obtained, reducing image errors caused by environmental factors and improving the accuracy of identifying abnormal respiratory behaviors.

[0052] Environmental monitoring unit

[0053] Temperature and humidity sensor: A temperature and humidity sensor with an accuracy of ±0.3°C is selected to continuously monitor the temperature and humidity of the rabbit breeding environment. Its measurement range is -10°C to 40°C for temperature and 20% to 90% for humidity.

[0054] Light sensor: A light intensity sensor is adopted, with a measurement range of 0 - 20000 lux, which can sense the ambient light intensity to compensate for the lighting factor during image acquisition. The temperature and humidity sensor and the light sensor in the environmental monitoring unit continuously monitor the temperature, humidity, and light intensity of the rabbit breeding environment. On the one hand, these environmental data can be used to compensate for the lighting factor during image acquisition, reducing the interference of ambient light changes on image analysis; on the other hand, the environmental data combined with the rabbit's respiratory data helps to more comprehensively analyze the potential relationship between the rabbit's abnormal respiratory behavior and environmental factors, providing more in-depth reference for breeders in optimizing the breeding environment and preventing diseases.

[0055] II. Data transmission module

[0056] Wired transmission unit

[0057] Data is transmitted using a high-speed Ethernet cable (CAT6 and above standards), with a transmission rate of up to 1 Gbps and above, stably transmitting the data collected by the image acquisition module to the server or local computer where the data analysis module is located. The TCP / IP protocol is used for data encapsulation and transmission to ensure the integrity and accuracy of the data. The wired transmission unit uses a high-speed Ethernet cable for data transmission, follows the TCP / IP protocol, and has a transmission rate of up to 1 Gbps and above, capable of quickly and stably transmitting a large amount of data collected by the image acquisition module to the server or local computer where the data analysis module is located, ensuring the integrity and accuracy of the data. In a large-scale farming scenario, the image data of multiple rabbits can be transmitted in a timely manner, avoiding data backlog and transmission delays, and providing strong support for real-time identification of abnormal breathing behaviors.

[0058] Wireless transmission unit

[0059] At the same time, a Wi-Fi 6 wireless transmission module is equipped as a backup transmission method and automatically switches when the wired network fails. The Wi-Fi 6 module follows the IEEE 802.11ax standard, with a theoretical maximum transmission rate of up to 9.6 Gbps, and the transmission frequency bands are 2.4 GHz and 5 GHz dual bands, which can automatically switch frequency bands according to signal strength and interference conditions to ensure the stability of data transmission.

[0060] It is worth noting that the wired transmission unit and the wireless transmission unit can be automatically switched, and the switching time is 8 - 10 seconds. The design of combining wired and wireless with an automatic switching function ensures the stability and reliability of data transmission. Even if the wired network fails, the wireless transmission module can quickly take over the data transmission task, and the switching time does not exceed 10 seconds, ensuring that data transmission is not interrupted and avoiding missed detection or misjudgment of abnormal breathing behaviors caused by transmission interruption.

[0061] III. Data analysis module

[0062] Image preprocessing unit

[0063] Image registration: The images collected by the depth camera and the optical camera are registered using a feature point-based registration algorithm. First, the SIFT (Scale-Invariant Feature Transform) algorithm is used to extract the feature points in the two images. For the feature points in the depth image and the feature points in the optical image , calculate the matching relationship between them. Let the error threshold of feature point matching pixels, and optimize the matching error through the least squares method to make

[0064] where is the The three-dimensional coordinates of the feature points, denoted as , where is the depth value,[[]] is the two-dimensional coordinates of the th feature point in the optical image,[[]] is the error threshold for feature point matching, in pixels,[[]] is the number of pairs of matched feature points, thus obtaining an accurate image registration result.[[]]

[0065] Noise removal: For the registered image, the bilateral filtering algorithm is used to remove noise. The bilateral filtering formula is:[[]]

[0066] where is the intensity value of the pixel after filtering,[[]] is the original intensity value of the pixel point ,[[]] is the neighborhood of the pixel ,[[]] is the spatial domain Gaussian function, used to calculate the weight according to the spatial distance between pixels, with a standard deviation ,[[]] is the range domain Gaussian function, used to calculate the weight according to the pixel intensity difference between pixels, with a standard deviation ,[[]] is the normalization coefficient,[[]]

[0067] The image registration subunit uses a feature point-based registration algorithm, extracts feature points through the SIFT algorithm, and optimizes the matching error using the least squares method. The error threshold is only 0.1 pixel, which can achieve accurate registration of the depth camera and optical camera images, enabling subsequent respiratory feature analysis to be based on more accurate image fusion data. The noise removal subunit uses the bilateral filtering algorithm. By reasonably setting the standard deviation of the spatial domain Gaussian function to 0.5 and the standard deviation of the range domain Gaussian function to 10, it effectively removes the noise in the image, improves the image quality, reduces the interference of noise on respiratory feature extraction, and further enhances the accuracy of the analysis results.[[]]

[0068] Rabbit respiratory feature extraction unit[[]]

[0069] Respiratory area localization: Using the depth image information, the thoracic region of the rabbit is located as the main respiratory monitoring area through the region growing algorithm. First, a seed point in the thoracic region is selected, and the growth threshold meter (representing the depth difference threshold), and then starting from the seed point, adjacent pixel points with a depth value difference within the growth threshold range from the seed point depth value are merged into the growth region, and the iteration continues until no new pixel points meet the conditions. The area threshold of the growth region is set to square centimeters, and by comparing with the preset rabbit body model, the accurate breathing region is further determined.

[0070] Respiratory rate calculation: Analyze the image sequence of the breathing region, and use the optical flow method to calculate the movement speed of the breathing region. Let the optical flow vector of the pixel points in the -th and the +1-th frames of the breathing region be . Solve the optical flow equation through the Horn-Schunck optical flow algorithm:

[0071] where, is the image intensity function, representing the change of the brightness value of a certain point in the image over time ( ) and space , , are the components of the optical flow vector, representing the movement speeds of the pixel points in the and directions respectively. Set the optical flow speed threshold m / s. When the optical flow speed of more than 30% of the pixel points in the breathing region exceeds this threshold, it is recorded as one breathing action. The respiratory rate , where is the number of breathing actions detected within the time (in seconds), is the total monitoring time. The normal breathing rate range of rabbits is 30 - 60 times per minute, and if it exceeds this range, the respiratory rate is determined to be abnormal.

[0072] Respiratory depth analysis: According to the depth image sequence of the breathing region, calculate the maximum change amount of the chest region depth during the breathing process. Set the depth change threshold m. If the chest depth change during the breathing process exceeds this range, the respiratory depth is determined to be abnormal.

[0073] The breathing area positioning subunit uses depth image information and adopts the region growing algorithm to locate the rabbit's thoracic cavity area as the main breathing monitoring area. By setting the depth difference growth threshold of the seed point to 0.2 meters and the growth area threshold to 100 - 300 square centimeters, and combining with the preset rabbit body shape model for secondary screening, it can accurately determine the breathing area and provide an accurate monitoring range for calculating the breathing frequency and depth. The breathing frequency calculation subunit uses the optical flow method to calculate the movement speed of the breathing area, solves the optical flow equation based on the Horn - Schunck optical flow algorithm, sets the optical flow speed threshold to 0.05 m / s, and combines with the standard of the normal rabbit breathing frequency range of 30 - 60 times per minute, it can accurately calculate the breathing frequency and accurately judge the abnormal breathing frequency situation. The breathing depth analysis subunit sets the depth change threshold to 0.03 - 0.1 meters according to the depth image sequence of the breathing area, can accurately calculate the maximum change amount of the thoracic cavity area depth during breathing and judge the abnormal breathing depth. Through the comprehensive and accurate analysis of the breathing frequency and depth, it realizes the comprehensive and accurate monitoring of the rabbit's breathing state.

[0074] Abnormal behavior recognition unit

[0075] Respiratory rhythm judgment: Analyze the time interval sequence between consecutive breathing actions , calculate the autocorrelation function of the time interval:

[0076] where, is the value of the autocorrelation function at the lag order , which is used to measure the periodicity of the breathing time interval sequence, the time interval sequence average value, is the total number of the time interval sequence, is the lag order. Set the autocorrelation function threshold , when the autocorrelation function is less than this threshold within the lag order range , it is determined that the respiratory rhythm is abnormal. The respiratory rhythm judgment subunit of the abnormal behavior recognition unit calculates the autocorrelation function of the time interval sequence between consecutive breathing actions, sets the autocorrelation function threshold to 0.3, and the lag order to 3 - 10, and can accurately judge the abnormal respiratory rhythm. This multi - dimensional breathing feature analysis and abnormal behavior recognition method greatly improves the accuracy and reliability of abnormal breathing behavior recognition compared with traditional single - index judgment or simple algorithm analysis, can detect potential health problems of rabbits earlier, and provides the possibility for timely treatment.

[0077] IV. Data storage and interaction module

[0078] Data storage unit

[0079] Use a distributed file system (such as Ceph) to store the original image data and the processed analysis data. The original image data is classified and stored according to the date, rabbit number, and camera type. The analysis data is stored in a relational database (such as PostgreSQL), including the individual information of the rabbits, respiratory characteristic data, and abnormal behavior records. The database table structure is reasonably designed to ensure the efficient storage and query of data. The data is stored for at least 1 year to meet the needs of long-term health tracking. The data storage unit uses a distributed file system to store the original image data, which is classified and stored according to the date, rabbit number, and camera type, facilitating the management and traceability of data. At the same time, a relational database is used to store the analysis data, including the individual information of the rabbits, respiratory characteristic data, abnormal behavior records, etc. The database table structure is reasonably designed to ensure the efficient storage and query of data. The data is stored for at least 1 year to meet the needs of long-term health tracking, providing historical data reference for the breeders on the respiratory health status of the rabbits, helping to analyze the development trends and patterns of abnormal respiratory behaviors of the rabbits, and thus formulating more scientific breeding management strategies.

[0080] User Interaction Unit

[0081] Develop a Web-based user interface through which users can log in to the system via a browser to view the respiratory monitoring data and abnormal behavior reports of the rabbits. The interface provides a real-time data display function, presenting the respiratory rate and depth changes of the rabbits in the form of charts (such as bar charts and line charts), while also showing the environmental temperature, humidity, and light intensity information. Users can also set alarm threshold parameters on the interface. When an abnormal respiratory behavior of a rabbit occurs, the system sends a text message alarm to the administrator through the SMS gateway. The content of the text message includes the rabbit number, abnormal type, and detailed occurrence time information, and the alarm information is prominently displayed on the user interface for the administrator to take timely measures. The User Interaction Unit develops a Web-based user interface through which users can conveniently log in to the system via a browser to view the respiratory monitoring data and abnormal behavior reports of the rabbits. The interface provides a real-time data display function, presenting the respiratory rate and depth changes of the rabbits in an intuitive chart form, while also showing information such as environmental temperature, humidity, and light intensity, enabling breeders to clearly understand the real-time health status of the rabbits and the breeding environment parameters at a glance. Users can also flexibly set alarm threshold parameters on the interface according to actual needs. When an abnormal respiratory behavior of a rabbit occurs, the system sends a detailed text message alarm to the administrator through the SMS gateway, including information such as the rabbit number, abnormal type, and occurrence time, and the alarm information is prominently displayed on the user interface, facilitating the administrator to take timely measures. This convenient user interaction method allows breeders to real-time control the health dynamics of the rabbits, achieve refined breeding management, and improve the breeding efficiency and the survival rate of the rabbits, whether locally or remotely.

[0082] The above specific embodiments are only several alternative embodiments of the present invention. Based on the technical solution of the present invention and the relevant revelations of the above embodiments, those skilled in the art can make various alternative improvements and combinations to the above specific embodiments.

Claims

1. An automatic recognition and recording system for abnormal breathing behaviors of rabbits, characterized in that, Including: An image acquisition module, which is used to acquire the image data and environmental data of the rabbit; A data transmission module, which transmits the data collected by the image acquisition module to the data analysis module; A data analysis module, which processes and analyzes the received data to identify the abnormal breathing behavior of the rabbit; A data storage and interaction module, which is used to store data and implement interactive operations with users; The data analysis module further includes a rabbit breathing feature extraction unit, and the rabbit breathing feature extraction unit includes: Respiratory region localization subunit, seed point P s =(x s , y s , z s ), set the growth threshold T g = 0.2 m, and then starting from the seed point, merge adjacent pixel points whose depth value differs from the seed point depth value within the growth threshold range into the growth region, and continuously iterate until no new pixel points meet the conditions. The area threshold of the growth region is set to A t = 100 - 300 square centimeters, and further determine the respiratory region by comparing with a preset rabbit body shape model; A breathing frequency calculation sub-unit, which analyzes the image sequence of the breathing area, calculates the movement speed of the breathing area by using the optical flow method, sets the optical flow vector of the pixel point P=(x,y) in the breathing area of the t-th frame and the (t+1)-th frame in the image sequence as (u,v), and solves the optical flow equation through the Horn-Schunck optical flow algorithm: Among them, I is the image intensity function, representing the variation of the brightness value of a certain point in the image with time (t) and space (x, y). u and v are the components of the optical flow vector, respectively representing the movement speeds of the pixel point in the x and y directions. The optical flow speed threshold V is set. t = 0.005 m / s. When the optical flow speeds of more than 30% of the pixel points in the breathing area exceed the optical flow speed threshold, it is recorded as one breathing action. The breathing frequency where n is the number of breathing actions detected within time t, and t is the total monitoring time. The normal breathing frequency range of rabbits is 30 - 60 times per minute. If it exceeds this range, the breathing frequency is determined to be abnormal. Respiratory depth analysis subunit, based on the depth image sequence of the respiratory region, calculates the maximum change in the depth of the thoracic cavity region during respiration, and sets a depth change threshold D t = 0.3 to 0.1 meters. If the change in thoracic cavity depth during respiration exceeds the depth change threshold, it is determined that the respiratory depth is abnormal.

2. The automatic recognition and recording system for abnormal respiratory behaviors of rabbits according to claim 1, wherein The image acquisition module includes: An image acquisition unit, which combines a high-resolution, high-frame-rate depth camera and an ordinary optical camera. The resolution of the depth camera is 1280×720 pixels, the frame rate is 60fps, the resolution of the ordinary optical camera is 1920×1080 pixels, the frame rate is 30fps. The camera is equipped with automatic exposure, automatic white balance functions and an 8-24mm zoom lens, and is installed on a movable robotic arm. The movement range of the robotic arm in the X-axis direction is 0-2 meters, in the Y-axis direction is 0-1.5 meters, and in the Z-axis direction is 0.5-1.2 meters; An environmental monitoring unit, which includes a temperature and humidity sensor with an accuracy of ±0.3°C, a temperature measurement range of -10°C to 40°C and a humidity measurement range of 20% to 90%, and a light sensor with a measurement range of 0-20000 lux.

3. The automatic recognition and recording system for abnormal breathing behavior of rabbits according to claim 1, characterized in that, The data transmission module includes: A wired transmission unit, which uses a high-speed Ethernet cable for data transmission, follows the TCP / IP protocol, and the transmission rate reaches 1Gbps or more; A wireless transmission unit, which is equipped with a Wi-Fi 6 wireless transmission module, follows the IEEE 802.11ax standard, the theoretical transmission rate is up to 9.6Gbps, the transmission frequency band is a dual-band of 2.4GHz and 5GHz, and the wired transmission unit and the wireless transmission unit can be automatically switched, and the switching time is 8-10 seconds.

4. The automatic recognition and recording system for abnormal breathing behaviors of rabbits according to claim 1, characterized in that, The data analysis module includes an image preprocessing unit, and the image preprocessing unit specifically includes: An image registration sub-unit, which registers the images collected by the depth camera and the optical camera, and uses a feature point-based registration algorithm; A noise removal sub-unit, for the registered image, uses a bilateral filtering algorithm to remove noise.

5. The automatic recognition and recording system for abnormal breathing behavior of rabbits according to claim 4, characterized in that In the image registration subunit, feature points in two images are extracted by the Scale-Invariant Feature Transform (SIFT) algorithm. For the feature point P d =(x d , y d , z d ) in the depth image and the feature point P o =(x o , y o ) in the optical image, the matching relationship between them is calculated. Let the error threshold E t of feature point matching be 0.1 pixel, and the matching error is optimized by the least squares method to make Among them, is the three-dimensional coordinate of the i-th feature point in the depth image, expressed as (x d , y d , z d ), where z d is the depth value, is the two-dimensional coordinate of the i-th feature point in the optical image, E t is the error threshold for feature point matching, in pixels, and n is the number of pairs of matched feature points.

6. The automatic recognition and recording system for abnormal breathing behavior of rabbits according to claim 5, characterized in that, In the noise removal sub-unit, the bilateral filtering formula is: Among them, I p is the intensity value after filtering the pixel p, and I q is the original intensity value of the pixel point q. S is the neighborhood of the pixel p, is the spatial domain Gaussian function, which is used to calculate the weight according to the spatial distance (||p - q||) between pixels, and the standard deviation σ s = 0.5, is the range domain Gaussian function, which is used to calculate the weight according to the pixel intensity difference (||I p - I q ||), and the standard deviation σ r = 10, W p is the normalization coefficient.

7. The automatic recognition and recording system for abnormal breathing behaviors of rabbits according to claim 1, characterized in that The data analysis module further includes an abnormal behavior recognition unit, and the abnormal behavior recognition unit is provided with a breathing rhythm judgment subunit, which analyzes the time interval sequence {t i} between consecutive breathing actions and calculates the autocorrelation function of the time interval: Among them, ACF(k) is the value of the autocorrelation function at the lag order k, which is used to measure the periodicity of the respiratory time interval sequence. is the average value of the time interval sequence {t i}, N is the total number of the time interval sequence, k is the lag order. Set the autocorrelation function threshold ACF t = 0.

3. When the autocorrelation function is less than this threshold within the lag order range k = 3 to 10, it is determined that the respiratory rhythm is abnormal.

8. An automatic recognition and recording system for abnormal respiratory behaviors of rabbits according to claim 1, characterized in that, The data storage and interaction module includes: A data storage unit, which uses a distributed file system to store the original image data and classifies and stores it according to the date, rabbit number and camera type, and uses a relational database to store the analysis data, including rabbit individual information, breathing feature data, and abnormal behavior records; The user interaction unit develops a Web-based user interface. Users can log in through a browser to view the rabbit's breathing monitoring data and abnormal behavior reports. The interface provides a real-time data display function, showing the rabbit's breathing frequency and the change in breathing depth in the form of charts, and at the same time displaying the environmental temperature, humidity, and light intensity information. Users set alarm threshold parameters on the interface. When abnormal rabbit behavior occurs, the system sends a text message alarm to the administrator through the SMS gateway.

9. The automatic recognition and recording system for abnormal breathing behaviors of rabbits according to claim 8, characterized in that, The content of the text message includes the rabbit number, abnormal type, and detailed occurrence time information, and the alarm information is prominently displayed on the user interface.

Images