A cattle flying festival hair removal detection device and detection method

Abstract

The present invention relates to the field of intelligent breeding and intelligent devices, and discloses a hock hair removal detection device for cattle, which includes a device main body, an image acquisition unit for collecting images of the hock part of cattle, an array pressure sensing unit for mapping the pressure of the cattle hoof, a time-sharing excitation light source for providing light, a synchronous control unit for synchronizing the image acquisition unit and the time-sharing excitation light source, an analysis and reasoning unit for performing data analysis and reasoning, and a vertical adjustment unit for adjusting the height of the image acquisition unit. The present invention also discloses a hock hair removal detection method for cattle. This method uses the above detection device for data acquisition, can realize quantitative detection of hock hair removal of cattle, and obtain a quantitative result. The detection results of this detection device and measurement method are accurate, the detection efficiency is high, and the practicability is strong.

Description

Technical Field

[0001] The present invention relates to the field of intelligent breeding and intelligent devices, and relates to, but is not limited to, a device and method for detecting hock hair loss in cattle. Background Art

[0002] Hair loss is often an early sign of cattle health problems, which may be caused by factors such as skin diseases, parasite infections, malnutrition, or environmental stress. By detecting hair loss in a timely manner, ranchers can discover and intervene in these potential problems as early as possible, thereby protecting the health of cattle and improving their production performance and reproductive efficiency.

[0003] The hock is a key part that directly touches the ground when the cattle lies down, and bears friction and pressure for up to 12 - 14 hours every day, so it is extremely vulnerable to physical wear. If the bedding material is too hard or the environmental hygiene is poor, the risk of hock hair loss will increase significantly. Compared with other parts, hock hair loss can more intuitively reflect the bedding quality, environmental hygiene, and feeding management level, and because of its clear position and easy observation, it has become an ideal indicator for monitoring and quantifying the hair loss condition. Detecting hock hair loss in cattle plays an irreplaceable role in maintaining productivity, reducing medical costs, protecting individual value, optimizing the feeding environment, and improving management level.

[0004] However, the temperature measurement camera based on thermal imaging can only provide temperature data and cannot comprehensively reflect other key features of the target object. In the application scenario of detecting hock hair loss in cattle, relying solely on temperature features may lead to incomplete information and affect the accuracy of overall judgment. Secondly, in the case of water, dirt, or suppuration in the hock area, the thermal imaging of the temperature sensor may not accurately reflect the real temperature. Finally, the farm environment is complex, and coupled with the fact that the activity areas of livestock are often in outdoor environments, complex and changeable environmental interference sources will significantly reduce the data acquisition ability of the temperature measurement camera, resulting in severely distorted collected data. These external factors will interfere with the measurement results of the sensor, causing the temperature readings to be distorted, and thus leading to misjudgment.

[0005] Therefore, there is an urgent need to provide a detection device and its detection method that can accurately detect the hock hair loss state of cattle. Summary of the Invention

[0006] In view of this, the embodiments of the present invention provide a device and method for detecting hock hair loss in cattle, aiming at the deficiencies of the existing technology, and proposing to improve the determination accuracy of the hock hair loss state through a method of multi-modal data fusion.

[0007] The technical solutions of the embodiments of the present invention are specifically as follows:

[0008] In a first aspect, an embodiment of the present invention provides a cattle hock hair removal detection device, comprising: a device main body in a groove structure; two groups of vertical adjustment units, an image acquisition unit, and a light source group respectively arranged at opposite positions on both sides of the groove; a synchronous control unit and an analysis and reasoning unit arranged on either side of the groove; and an array-type pressure sensing unit arranged at the bottom of the groove; wherein:

[0009] The vertical adjustment unit is used to adjust the acquisition position of the image acquisition unit by sliding up and down;

[0010] The image acquisition unit is deployed in the central area of the vertical adjustment unit and is used to collect multi-dimensional image data of the cattle hock area passing through the detection device in real time in combination with the light source group;

[0011] The light source group includes four time-division excitation light sources respectively deployed around the vertical adjustment unit and is used to supplement light sources when the image acquisition unit captures images;

[0012] The array-type pressure sensing unit is used to collect pressure data of four cattle hooves in real time while the image acquisition unit acquires images and transmit them to the analysis and reasoning unit;

[0013] The synchronous control unit is used to control the exposure timing of the image acquisition unit and synchronously control the turning on or off of the four time-division excitation light sources during image acquisition;

[0014] The analysis and reasoning unit is used to extract texture features from the preprocessed multi-dimensional image data, calculate the hock hair removal rate, and determine the hock hair removal degree in combination with the results of hoof pressure abnormality determination based on the pressure data.

[0015] In some embodiments, the imaging image is processed by a built-in depth vision algorithm in the image acquisition unit to automatically identify the cattle hock area; when it is recognized that the cattle hock area is located at the horizontal center position of the imaging image, the synchronous control unit is evoked.

[0016] In some embodiments, the synchronous control unit, a hardware-level synchronous controller based on FPGA, realizes multi-sensor clock domain synchronization; and adopts a double-buffer DMA architecture to ensure phase locking between image acquisition and light source pulses.

[0017] In some embodiments, the device main body also integrates a network unit for uploading the collected image data, pressure data, and hair removal detection analysis to a database in the cloud in real time; meanwhile, the device main body is also externally connected with an RFID antenna for identifying the individual identity of cattle while detecting hock hair removal.

[0018] In a second aspect, an embodiment of the present invention provides a cattle hock hair removal detection method. The cattle hock hair removal detection device described in the first aspect is deployed in a passage that cattle must pass through, and the detection is carried out through the following steps:

[0019] After the image acquisition unit captures the image of the cow's leg, the synchronization control unit is awakened when the hock area of the cow is recognized by the depth vision algorithm to be at the horizontal center position of the imaging image; the synchronization control unit controls the exposure timing of the image acquisition unit and synchronously controls the turning on or off of four time-sharing excitation light sources to collect multi-dimensional image data; the multi-dimensional image data is preprocessed and then transmitted to the analysis and inference unit; while the image acquisition unit is acquiring images, the array pressure sensing unit real-time collects the pressure data corresponding to the four cow hooves and transmits it to the analysis and inference unit; the analysis and inference unit extracts the texture features from the preprocessed multi-dimensional image data, calculates the hock hair removal rate, and combines the results of the hoof pressure abnormality determination based on the pressure data to determine the degree of hock hair removal.

[0020] In some embodiments, the four time-sharing excitation light sources are respectively a Light Source 1, a Light Source 2, a Light Source 3, and a Light Source 4, which are distributed around the image acquisition unit for light source supplementation from four angles; the process of collecting the multi-dimensional image data is specifically as follows:

[0021] When the hock area of the cow is at the horizontal center position of the imaging image, the synchronization control unit controls all four time-sharing excitation light sources to turn on, triggering the image acquisition unit to acquire a full-light source image; the synchronization control unit turns off the Light Source 2, the Light Source 3, and the Light Source 4, waits for the afterglow to fade, and triggers the image acquisition unit to acquire a first single-light source image; the synchronization control unit turns off the Light Source 1 and turns on the Light Source 2, waits for the afterglow to fade, and triggers the image acquisition unit to acquire a second single-light source image; the synchronization control unit turns off the Light Source 2 and turns on the Light Source 3, waits for the afterglow to fade, and triggers the image acquisition unit to acquire a third single-light source image; the synchronization control unit turns off the Light Source 3 and turns on the Light Source 4, waits for the afterglow to fade, and triggers the image acquisition unit to acquire a fourth single-light source image, and then turns off the Light Source 4.

[0022] In some embodiments, the preprocessed multi-dimensional images include a grayscale full-light source image and four single-light source images; the analysis and inference unit extracts the texture features from the preprocessed multi-dimensional image data, calculates the hock hair removal rate, and combines the pressure data collected by the array pressure sensing unit to determine the degree of hock hair removal, including: calculating the texture features for all pixels in the full-light source image and each single-light source image respectively using the local binary pattern; based on the texture features, calculating the global image difference matrix C of the full-light source image and the image difference matrix C between any two single-light source images respectively ij , where the values of i and j range from 1 to 4; for the global image difference matrix C and all image difference matrices C ijPerform element-wise summation, mark the element points with a summation value greater than or equal to 4 as 1 to represent hair, and mark the element points with a summation value less than 4 as 0 to represent skin, obtaining the fusion matrix Z; calculate the proportion of the element points representing skin in the fusion matrix Z as the hock hair removal rate; determine the pressure difference between the pressure value of the cow's hoof on the side close to the camera and the average pressure of the other three cow's hooves based on the collected pressure data; perform weighted summation on the hock hair removal rate and the pressure difference to obtain a comprehensive score; determine the hock hair removal degree according to a preset determination rule based on the comprehensive score and a specific coefficient range threshold.

[0023] In some embodiments, based on the texture features, calculate the global image difference matrix C of the full-light source image and the image difference matrix C between any two single-light source images respectively ij , including: for the full-light source image and each single-light source image, calculate the local binary pattern (LBP) value of each pixel; perform binarization processing on the LBP value of each pixel in the full-light source image based on a set threshold to generate the global image difference matrix C of the full-light source image; compare the four single-light source images pairwise, calculate the comprehensive difference of the same pixel in any two single-light source images to obtain the corresponding difference image; wherein, the difference image reflects the intensity of shadow change under two light sources; the comprehensive difference is determined by performing weighted summation on the LBP texture difference and the gray-scale difference of the same pixel in the two single-light source images; for each difference image, perform binarization processing through a set threshold to generate the image difference matrix C ij .

[0024] In some embodiments, for each single-light source image, calculate the LBP value of each pixel through the following formula:

[0025] ;

[0026] wherein, ( ) is the sign function, is the gray-scale value of pixel , is pixel the gray-scale value of the k-th pixel in the 3×3 neighborhood of;

[0027] For the full-light source image, calculate the LBP value of each pixel through the following formula:

[0028] ;

[0029] wherein, is the gray-scale value of pixel in the full-light source image, is the gray-scale value of the p-th pixel in the 3×3 neighborhood of.

[0030] The beneficial effects brought by the technical solution provided by the embodiments of the present invention at least include:

[0031] A cattle hock hair loss detection device and detection method of the present invention can quickly determine whether there is hair loss on the hock part through multiple image information and the pressure distribution of cattle hooves. Compared with similar technologies, the present invention has the following advantages:

[0032] (1) Non-invasive design, reducing animal stress: The device can be arranged in the cattle shed passage or milking hall, and the detection is completed when the cattle walk or stand naturally, without additional driving or staying, meeting the needs of animal welfare;

[0033] (2) Strong adaptability to the environment: It is only necessary to collect multiple cattle hock image information through a camera, and it is less affected by environmental factors. With the configuration of cattle hoof pressure distribution detection, the detection effect is good;

[0034] (3) Image + pressure two-dimensional analysis: The combination of image analysis and cattle hoof pressure distribution monitoring forms cross-verification, significantly reducing the misjudgment rate, and is particularly sensitive to early hair loss or local mild symptoms;

[0035] (4) Efficient and automated processing, saving labor costs: Using a deep learning model (such as a convolutional neural network) to analyze the image features in real time, without manual inspection head by head, suitable for large-scale farms;

[0036] (5) Low cost and easy to deploy, with strong compatibility: Using standardized camera and pressure sensor components, supporting the transformation and upgrading of existing cattle shed facilities, without the need to rectify the breeding environment. Description of the Drawings

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings, where:

[0038] Figure 1 It is the front view of the hardware structure of the cattle hock hair loss detection device provided by the embodiments of the present invention;

[0039] Figure 2 It is the right view of the hardware structure of the cattle hock hair loss detection device provided by the embodiments of the present invention;

[0040] Figure 3 It is the top view of the hardware structure of the cattle hock hair loss detection device provided by the embodiments of the present invention;

[0041] Figure 4Schematic flowchart of the hair removal detection method for cattle fly festival provided by the embodiments of the present invention

[0042] Explanation of reference numerals:

[0043] 1 - Main body of the device; 2 - Image acquisition unit; 3 - Array pressure sensing unit; 4 - Light source group; 5 - Synchronous control unit; 6 - Analysis and reasoning unit; 7 - Vertical adjustment unit. Specific embodiments

[0044] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. The following embodiments are used to illustrate the present invention but are not used to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0045] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments and can be combined with each other without conflict.

[0046] It should be noted that the terms "first / second / third" involved in the embodiments of the present invention are only used to distinguish similar objects and do not represent a specific order for the objects. It can be understood that "first / second / third" can be interchanged with a specific order or sequence when allowed, so that the embodiments of the present invention described here can be implemented in an order other than that illustrated or described here.

[0047] Those skilled in the art of the present technology can understand that unless otherwise defined, all terms (including technical terms and scientific terms) used here have the same meaning as the general understanding of those of ordinary skill in the art in the field to which the embodiments of the present invention belong. It should also be understood that terms such as those defined in a general dictionary should be understood to have a meaning consistent with the meaning in the context of the prior art and will not be interpreted with an idealized or overly formal meaning unless specifically defined as here.

[0048] Hock region hair loss in cattle is usually associated with local inflammation or skin lesions, which may cause the cattle to adjust their standing posture to reduce the weight on the affected limb due to pain or discomfort. This compensatory behavior will be directly reflected in the hoof pressure distribution of the cattle, manifested as the pressure data of the affected hoof contacting the ground being significantly lower than that of the healthy hoof. The purpose of the present invention is to establish an association model between physiological abnormalities and behavioral compensation by synchronously analyzing the visual characteristics of the hock hair loss area and the pressure data of the corresponding cattle hoof, so as to break through the limitations of single visual detection. This multi-modal data fusion method can not only improve the accuracy of determining the hock hair loss state, but also provide a quantitative basis for early pain recognition, which has important value for improving animal welfare monitoring technology.

[0049] Figure 1 The following is a schematic diagram of the composition structure of a cattle hock hair loss detection device provided by an embodiment of the present invention, as Figure 1 shown. The detection device includes: a device main body 1 in a groove structure; two groups of vertical adjustment units 7, an image acquisition unit 2 and a light source group 4 respectively arranged at opposite positions on both sides of the groove; a synchronous control unit 5 and an analysis and reasoning unit 6 arranged on either side of the groove; and an array type pressure sensing unit 3 arranged at the bottom of the groove; wherein:

[0050] The vertical adjustment unit 7 is used to adjust the acquisition position of the image acquisition unit by sliding up and down;

[0051] The image acquisition unit 2 is deployed in the central area of the vertical adjustment unit 7 and is used to collect multi-dimensional image data of the cattle leg hock region passing through the detection device in real time in combination with the light source group 4;

[0052] The light source group 4 includes four time-sharing excitation light sources, which are respectively deployed around the vertical adjustment unit 7 and are used to supplement light sources when the image acquisition unit 2 captures images;

[0053] The array type pressure sensing unit 3 is used to collect the pressure data of four cattle hooves in real time while the image acquisition unit 2 acquires images and transmit them to the analysis and reasoning unit 6;

[0054] The synchronous control unit 5 is used to control the exposure timing of the image acquisition unit 2 and synchronously control the turning on or off of the four time-sharing excitation light sources during image acquisition;

[0055] The analysis and reasoning unit 6 is used to extract texture features from the preprocessed multi-dimensional image data, calculate the hock hair loss rate, and determine the hock hair loss degree in combination with the results of hoof pressure abnormality determination based on the pressure data.

[0056] In practice, the detection device can be placed in a passageway that cattle must pass through, with the image acquisition unit 2 equipped with a visible light camera. The image acquisition unit 2 and light source group 4 are mounted on a vertical adjustment unit 7. The light source group 4 is mounted around the visible light camera, illuminating the acquisition area from four angles. The vertical adjustment unit 7 is mounted on the side of a groove in the device body 1 and can be slid up and down to adjust the acquisition position of the image acquisition unit 2. When the cattle pass through the passageway, the image acquisition unit 2 and light source group 4 are controlled by the synchronization control unit 5 to collect image data from multiple angles of the hock area, i.e., multi-dimensional image data, in real time. The processed images are then transmitted to the analysis and reasoning unit 6 for analysis and reasoning. The array pressure sensing unit 3, which has a built-in array pressure sensing device, can send pressure information to the analysis and reasoning unit 6 in real time. The four time-sharing excitation light sources in the light source group 4 are controlled by the synchronization control unit 5 to turn on or off in sequence, providing light to the visible light camera and supplementing the light source when capturing images. The synchronization control unit 5 then controls the appropriate number of time-sharing excitation light sources to synchronize data acquisition with the image acquisition unit 2, ensuring synchronization of light source illumination and image acquisition while ensuring that the dual light source images are free of crosstalk.

[0057] Preferably, the equipment body 1 can be adjusted according to the width of the channel. The main body adopts a steel structure and is sprayed with an anti-corrosion and anti-rust coating on the surface. It can withstand a weight of 2000kg without deformation. A rubber shock-absorbing support is configured at the bottom to effectively reduce the friction between the equipment body and the ground, reduce the wear of the equipment caused by the vibration generated by the passing of cattle, and reduce the shaking of the equipment caused by the cattle.

[0058] Preferably, the image acquisition unit 2 uses a high-speed CMOS image sensor (1920×1080) equipped with an adaptive exposure algorithm to achieve dynamic image capture in the visible light band. It adopts an IP68 waterproof design and integrates a real-time image processing digital signal processing (DSP) chip, transmitting the image data stream to the analysis and reasoning unit 6 via an Ethernet interface.

[0059] Preferably, the array-type pressure sensing unit 3 uses a high-precision pressure sensor, which can achieve low-latency acquisition of pressure sensing information and transmit the pressure information to the analysis and reasoning unit 6 via Ethernet. Furthermore, the array-type pressure sensing unit 3 uses a composite material to effectively control friction and wear resistance, can be disassembled for cleaning, and the pressure sensing unit can be resized.

[0060] Preferably, the light source group 4 is configured with four high-brightness LED high-density light source arrays (color temperature 5600K±200K, illuminance adjustable range 100 - 5000 lux), with a lifespan exceeding 35,000 hours in a room-temperature environment. A protective shell is provided around the light source array, and the protective shell is detachable and washable. The PWM dimming technology is adopted to achieve a response speed at the 0.1ms (millisecond) level. The LED high-density light source array integrates an optical diffuser plate and a polarization filter, supporting the pulse synchronous trigger mode (maintaining a microsecond-level synchronous accuracy with the image sensor), effectively suppressing specular reflection interference.

[0061] Preferably, the synchronization control unit 5 is a hardware-level synchronization controller based on a Field Programmable Gate Array (FPGA), which realizes the synchronization of multiple sensor clock domains. The dual-buffer direct memory access (DMA) architecture is adopted to ensure the phase locking of image acquisition and light source pulses. The synchronization control unit 5 uses RS485 communication and the standard Modbus-RTU communication protocol to control the exposure timing of four time-division excited light sources and the image acquisition unit 2. The exposure delay is less than 2ms, and the trigger accuracy is ±500μs (microseconds).

[0062] Preferably, the analysis and reasoning unit 6 integrates high-performance computing hardware and can deploy a multi-modal data fusion algorithm stack to provide computing power for visual recognition, shadow calculation, pressure calculation, etc. Further, the analysis and reasoning unit 6 can deploy and run a visual recognition model, integrate a hock hair removal calculation program, and can quickly determine hock hair removal.

[0063] In some embodiments, the imaging image is processed by the built-in depth vision algorithm in the image acquisition unit to automatically identify the hock area of the cow; when it is recognized that the hock area is located at the horizontal center position of the imaging image, the synchronization control unit is activated.

[0064] Here, the synchronization control unit can achieve the trigger function through a Pulse-Width Modulation (PWM) control signal, and microcontrollers such as FPGA can be used to design a flexible trigger mechanism.

[0065] In some embodiments, the device body 1 also integrates a network unit for real-time uploading of the collected image data, pressure data, and hair removal detection analysis to a database in the cloud; at the same time, the device body 1 is also externally connected to a Radio Frequency Identification (RFID) antenna for identifying the individual identity of the cow while detecting hock hair removal.

[0066] An embodiment of the present invention also provides a method for detecting hair loss on the hock of cattle. The above-mentioned device for detecting hair loss on the hock of cattle is deployed in a passage that cattle must pass through, such as Figure 4 as shown, and the detection is carried out through the following steps:

[0067] Step S110: After the image acquisition unit captures the image of the cattle's leg, the synchronization control unit is awakened when the hock area of the cattle is recognized by the depth vision algorithm to be located at the horizontal center position of the imaging image.

[0068] Here, deep learning object detection models such as YOLO and SSD are widely used in object detection, including personnel detection, animal monitoring, etc. These models can be used to train the algorithm for recognizing the hock area of cattle.

[0069] It should be noted that the hock of cattle is located on the hind limb, with the tarsal joint as the anatomical basis, and the angle is moderate, about between 140° and 150°. In image processing, determining whether an image is horizontally centered can be achieved through methods such as reference lines, alignment tools, or intelligent reference lines. These methods can be implemented in the algorithm to determine whether the hock area is located at the center of the image, so as to ensure that the algorithm can accurately recognize the hock area of cattle.

[0070] Step S120: The synchronization control unit controls the exposure timing of the image acquisition unit and synchronously controls the turning on or off of four time-sharing excitation light sources to collect multi-dimensional image data.

[0071] Here, the four time-sharing excitation light sources are around the image acquisition unit and irradiate the acquisition area from four angles. By controlling the exposure timing and the number of light sources, a full-light source image and 4 single-light source images can be obtained.

[0072] Step S130: The multi-dimensional image data is preprocessed and then transmitted to the analysis and reasoning unit.

[0073] Here, image preprocessing includes but is not limited to denoising, contrast enhancement, grayscale conversion, normalization, image registration, etc. In the embodiment of the present invention, the 5 image information collected by the image acquisition unit are T0, T1, T2, T3, and T4 respectively, and the collected images are converted into grayscale images , , , , .

[0074] Step S140: While the image acquisition unit is collecting images, the array pressure sensing unit real-time collects the pressure data corresponding to the four cattle hooves and transmits it to the analysis and reasoning unit;

[0075] Step S150: The preprocessed multi-dimensional image data is subjected to texture feature extraction by the analysis and reasoning unit, the hock hair removal rate is calculated, and the hock hair removal degree is determined by combining the result of hoof pressure abnormality determination based on the pressure data.

[0076] Here, this multi-dimensional analysis method combines image processing and pressure sensing technologies, can provide objective and accurate analysis and detection results, and helps to timely detect and handle the health problems of cows.

[0077] The cow hock hair removal detection method provided by the present invention realizes the quantitative detection of cow hock hair removal through steps such as multi-dimensional image acquisition, image preprocessing, image difference matrix calculation, and hock hair removal rate calculation. Through the above steps, the hair area and skin area can be accurately identified, and the hock hair removal rate can be calculated. Thus, the accurate detection and evaluation of the cow hock hair removal situation can be realized. This detection method has the advantages of non-invasive, strong environmental adaptability, two-dimensional analysis of image + pressure, efficient automatic processing, low cost and easy deployment, is suitable for large-scale farms, can timely detect and intervene in potential health problems, protect the health of cows, and improve their production performance and reproductive efficiency.

[0078] In some embodiments, the four time-sharing excitation light sources are respectively a No. 1 light source, a No. 2 light source, a No. 3 light source, and a No. 4 light source, which are distributed around the image acquisition unit for light source supplementation from four angles; the specific process of collecting the multi-dimensional image data is as follows: when the cow hock area is located at the horizontal center position of the imaging image, the synchronous control unit controls all four time-sharing excitation light sources to be turned on, and triggers the image acquisition unit to collect a full light source image; the synchronous control unit turns off the No. 2 light source, the No. 3 light source, and the No. 4 light source, waits for the afterglow to disappear, and triggers the image acquisition unit to collect a first single light source image; the synchronous control unit turns off the No. 1 light source, turns on the No. 2 light source, waits for the afterglow to disappear, and triggers the image acquisition unit to collect a second single light source image; the synchronous control unit turns off the No. 2 light source, turns on the No. 3 light source, waits for the afterglow to disappear, and triggers the image acquisition unit to collect a third single light source image; the synchronous control unit turns off the No. 3 light source, turns on the No. 4 light source, waits for the afterglow to disappear, and triggers the image acquisition unit to collect a fourth single light source image, and then turns off the No. 4 light source.

[0079] Here, the key points of the collection process include the following aspects. Synchronous control: Ensure the precise synchronization of the turning on and off of the light sources with the operation of the image acquisition unit to avoid interference between the light sources. Wait for the afterglow to disappear: When switching the light sources, wait for a period of time to ensure that the afterglow of the previous light source completely disappears, so as to ensure the accuracy of image acquisition. Multi-angle illumination: Through four light sources at different angles, capture the texture and shadow changes of the hock area under different lighting conditions, and provide rich information for subsequent image analysis.

[0080] This multi-dimensional image data acquisition method can effectively capture the detailed features of the hock area, providing high-quality image data for hair removal detection, thereby improving the accuracy and reliability of the detection.

[0081] In some embodiments, the preprocessed multi-dimensional image includes a grayscale full-light source image and four single-light source images; the texture features are extracted from the preprocessed multi-dimensional image data by the analysis and reasoning unit, the hock hair removal rate is calculated, and the hock hair removal degree is determined by combining the pressure data collected by the array pressure sensing unit, including:

[0082] S121, calculate the texture features for all pixels in the full-light source image and each single-light source image respectively using the local binary pattern.

[0083] Here, LBP can effectively capture the local texture changes in the image, which is very important for identifying the hair removal area.

[0084] S122, based on the texture features, calculate the global image difference matrix C of the full-light source image and the image difference matrix C between any two single-light source images respectively ij , where the values of i and j range from 1 to 4.

[0085] Here, these difference matrices reflect the texture changes in the image under different light source conditions.

[0086] S123, perform element-wise summation on the global image difference matrix C and all image difference matrices C ij , mark the element points with the summation value greater than or equal to 4 as 1 to represent hair, and mark the element points with the summation value less than 4 as 0 to represent skin, to obtain the fusion matrix Z.

[0087] Here, the above-obtained 7 image difference matrices 、 、 、 、 、 、 Perform element-wise summation, mark the element points with the summation value greater than or equal to 4 as 1 (representing hair), and mark the element points with the summation value less than 4 as 0 (representing skin), so as to obtain the fusion matrix Z. This fusion matrix synthesizes the texture information under multi-light source conditions and helps to more accurately identify the hair removal area.

[0088] S124, calculate the proportion of the element points representing skin in the fusion matrix Z and use it as the hock hair removal rate.

[0089] Here, this hair removal rate reflects the severity of hair removal in the hock area.

[0090] S125. Determine the pressure difference between the pressure value of the hoof on the side close to the camera and the average pressure of the other three hooves based on the collected pressure data.

[0091] Here, this pressure difference can provide additional information about the force on the hooves, which helps to comprehensively evaluate the degree of hair removal.

[0092] ;

[0093] Among them, represents the pressure difference, is the maximum pressure value of the hoof on the side close to the camera, are the maximum pressure values of the other three hooves respectively.

[0094] S126. Perform a weighted sum of the fetlock hair removal rate and the pressure difference to obtain a comprehensive score.

[0095] Here, this comprehensive score combines the image texture features and pressure data, providing a more comprehensive hair removal assessment. Combining the fetlock hair removal rate and the pressure difference, the following determination formula can be designed:

[0096] ;

[0097] Among them: S is the comprehensive score of fetlock hair removal; and are the weight coefficients, used to adjust the influence of the sum on the final score (to be adjusted according to actual data, usually = 1); R is the fetlock hair removal rate (range: 0 to 1); F is the pressure difference.

[0098] It should be noted that and are obtained through experience with 30 cows in the actual production environment. The influence degree of hair removal recognition in the image is greater, so the proportion should be higher than that of hoof pressure measurement. The specific values are: = 0.7, = 0.3. After verification by actual measurement of 145 cows, the detection accuracy of this device reaches 93.2%, and the detection time per head is less than 3 seconds.

[0099] S127. Determine the degree of fetlock hair removal based on the comprehensive score and the specific coefficient range threshold according to the preset determination rules.

[0100] Here, assuming the comprehensive score S, the following determination rules are set:

[0101] In the case of , determine that the degree of fetlock hair removal has no influence;

[0102] In the case of In the case of, it is determined that the degree of hock hair loss has a mild impact;

[0103] In the case of, it is determined that the degree of hock hair loss has a moderate impact;

[0104] In the case of, it is determined that the degree of hock hair loss has a severe impact.

[0105] Among them, is the determination threshold, which is set through actual measurement and aquaculture disease experience. Preferably, , , . This determination rule can help aquaculturists quickly understand the health status of the hocks of cattle and take corresponding measures.

[0106] In some embodiments, based on the texture features, the global image difference matrix C of the full-light source image and the image difference matrix C between any two single-light source images are calculated respectively ij , including:

[0107] S221. For the full-light source image and each single-light source image, calculate the LBP value of each pixel.

[0108] Here, the LBP value reflects the texture change between a pixel and its neighboring pixels.

[0109] For each single-light source image, calculate the LBP value of each pixel through the following formula:

[0110] ;

[0111] Among them, ( ) is the sign function, is the gray value of pixel , is the gray value of the kth pixel in the 3×3 neighborhood of pixel ;

[0112] For the full-light source image, calculate the LBP value of each pixel through the following formula:

[0113] ;

[0114] Among them, is the gray value of pixel in the full-light source image, is the gray value of the pth pixel in the 3×3 neighborhood of.

[0115] S221, perform binarization processing on the LBP value of each pixel in the full light source image based on a set threshold to generate the global image difference matrix C of the full light source image.

[0116] Here, the grayscale image corresponding to the image T0 collected with all light sources turned on 0, through texture feature calculation and binarization processing: If the LBP value of a pixel is greater than the threshold T, it is considered that the pixel belongs to the hair area and is recorded as 1; otherwise, it is considered to belong to the skin area and is recorded as 0, generating the global image difference matrix C of the full light source image:

[0117] ;

[0118] Among them, is the value of the pixel in the global image difference matrix C, is the pixel 's LBP value.

[0119] S221, compare the four single light source images in pairs, calculate the comprehensive difference of the same pixel in any two single light source images, and obtain the corresponding difference image.

[0120] Here, the difference image reflects the intensity of shadow change under two light sources; the comprehensive difference is determined by weighted summation of the LBP texture difference and grayscale difference of the same pixel in the two single light source images, and the comprehensive difference of the same pixel in any two single light source images is specifically calculated through the following formula:

[0121] ;

[0122] ;

[0123] ;

[0124] Among them, is the grayscale difference of the element in the single light source image i and the single light source image j, is the LBP texture difference of the element in the single light source image i and the single light source image j, and are weight coefficients, satisfying .

[0125] S223, for each difference image, perform binarization processing through a set threshold to generate the image difference matrix C ij .

[0126] Here, according to actual measurement, a threshold corresponding to the difference image is obtained Perform binarization processing. If the difference value at a certain pixel position , it is considered that the shadow change at this position is significant, and it is a hair area, denoted as 1; otherwise, it is considered a skin area, denoted as 0.

[0127] Image difference matrix C ij can be expressed as:

[0128] ;

[0129] where is the value of the pixel ij in the image difference matrix C , and is the comprehensive difference value of the pixel .

[0130] Through the above steps, the global image difference matrix C of the full-light source image and the image difference matrix C between any two single-light source images can be calculated respectively ij . These matrices provide important texture feature information for subsequent hock hair removal rate calculation and hair removal degree determination. This multi-dimensional analysis method combines image processing and pressure sensing technologies, can provide objective and accurate detection and analysis results, and helps to timely discover and handle the health problems of cows.

[0131] The hock hair removal detection device and detection method provided by the embodiments of the present invention realize precise detection of the hock hair removal situation of cows through a variety of sensors and an advanced control unit. It not only uses image data for intuitive analysis, but also combines pressure data for comprehensive judgment, and can provide scientific health monitoring means for breeders. In the future, with the further development of technology, this device can also be optimized in terms of intelligence, automation and environmental adaptability to better serve the health management of the livestock industry.

[0132] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification only illustrates the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

[0133] It should be understood that the "one embodiment" or "an embodiment" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present invention. Therefore, the "in one embodiment" or "in an embodiment" that appears throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present invention, the magnitudes of the sequence numbers of the above processes do not mean the sequence of execution. The execution sequence of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention. The sequence numbers of the embodiments of the present invention above are only for description and do not represent the advantages or disadvantages of the embodiments.

[0134] It should be noted that in this text, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, article or device including the element.

[0135] In several embodiments provided by the present invention, it should be understood that the disclosed methods can be implemented in other ways. The methods disclosed in several method embodiments provided by the present invention can be arbitrarily combined without conflict to obtain new method embodiments. The features disclosed in several method embodiments provided by the present invention can be arbitrarily combined without conflict to obtain new method embodiments.

[0136] The above is only the implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claimed rights.

Claims

1. A cattle flying festival hair removal detection device, characterized in that, Comprising: A device main body in a groove structure; two sets of vertical adjustment units, an image acquisition unit, and a light source group respectively arranged at opposite positions on both sides of the groove; a synchronous control unit and an analysis and reasoning unit arranged on either side of the groove; and an array pressure sensing unit arranged at the bottom of the groove; wherein: The vertical adjustment unit is used to adjust the acquisition position of the image acquisition unit by sliding up and down; The image acquisition unit is deployed in the central area of the vertical adjustment unit and is used to collect multi-dimensional image data of the hock joint area of the cow passing through the detection device in real time in combination with the light source group; The light source group includes four time-sharing excitation light sources, which are respectively deployed around the vertical adjustment unit and are used to supplement the light source when the image acquisition unit captures the picture; The array pressure sensing unit is used to collect the pressure data of the four cow hooves in real time while the image acquisition unit acquires the image and transmit it to the analysis and reasoning unit; The synchronous control unit is used to control the exposure timing of the image acquisition unit and synchronously control the turning on or off of the four time-sharing excitation light sources during image acquisition; The analysis and reasoning unit is used to extract texture features from the preprocessed multi-dimensional image data, calculate the hock joint hair removal rate, and determine the hock joint hair removal degree in combination with the result of hoof pressure abnormality determination based on the pressure data; Wherein, the preprocessed multi-dimensional image data includes a full light source image and four single light source images that have been grayscaled; the analysis and reasoning unit is specifically configured to perform the following steps to determine the non-joint hair removal degree: Calculate the texture features of all pixels in the full-light source image and each of the single-light source images respectively using the local binary pattern; based on the texture features, calculate the global image difference matrix C of the full-light source image and the image difference matrix C between any two single-light source images respectively ij , i, j The value range of is from 1 to 4; perform element-wise summation on the global image difference matrix C and all the image difference matrices C ij ; mark the element points with the summation value greater than or equal to 4 as 1 to represent hair, and mark the element points with the summation value less than 4 as 0 to represent skin, obtaining the fusion matrix Z; calculate the proportion of the element points representing skin in the fusion matrix Z as the hock hair removal rate; determine the pressure difference between the pressure value of the hoof on the side close to the camera and the average pressure of the other three hooves based on the collected pressure data; perform weighted summation on the hock hair removal rate and the pressure difference to obtain a comprehensive score; determine the hock hair removal degree according to the preset determination rule based on the comprehensive score and the specific coefficient range threshold 2. The cattle fly festival hair removal detection device according to claim 1, wherein, Process the imaging image through the depth vision algorithm built in the image acquisition unit to automatically identify the cow hock joint area; when it is recognized that the cow hock joint area is located at the horizontal center position of the imaging image, the synchronous control unit is awakened.

3. The cattle fly festival hair removal detection device according to claim 1, characterized in that The synchronous control unit, based on a hardware-level synchronous controller of a field programmable gate array (FPGA), realizes multi-sensor clock domain synchronization; and adopts a double-buffer direct memory access (DMA) architecture to ensure phase locking of image acquisition and light source pulses.

4. The cattle fly festival hair removal detection device according to any one of claims 1 to 3, characterized in that The device main body also integrates a network unit, which is used to upload the collected image data, pressure data, and hair removal detection analysis results to the database in the cloud in real time; At the same time, the device main body is also externally connected to a radio frequency identification (RFID) antenna, which is used to identify the individual identity of the cow while detecting hock joint hair removal.

5. A method for detecting hair removal during the Niufei Festival, characterized in that, Deploy the cow hock joint hair removal detection device according to any one of claims 1 to 4 in the passage that the cow must pass through, and perform detection through the following steps: After the image acquisition unit captures the picture of the cow's leg, when it is recognized through the depth vision algorithm that the cow hock joint area is located at the horizontal center position of the imaging image, the synchronous control unit is awakened; The synchronous control unit controls the exposure timing of the image acquisition unit and synchronously controls the turning on or off of the four time-sharing excitation light sources to collect multi-dimensional image data; Transmit the preprocessed multi-dimensional image data to the analysis and reasoning unit; While the image acquisition unit acquires the image, the array pressure sensing unit collects the pressure data corresponding to the four cow hooves in real time and transmits it to the analysis and reasoning unit; The texture features of the preprocessed multi-dimensional image data are extracted by the analysis and reasoning unit, the hock hair removal rate is calculated, and the hock hair removal degree is determined by combining the result of hoof pressure abnormality determination based on the pressure data; Among them, the preprocessed multi-dimensional image data includes a grayscale full-light source image and four single-light source images; the following steps are performed by the analysis and reasoning unit to determine the non-joint hair removal degree: Calculate the texture features of all pixels in the full-light source image and each of the single-light source images respectively using the local binary pattern; based on the texture features, calculate the global image difference matrix C of the full-light source image and the image difference matrix C between any two single-light source images respectively ij , i, j ranges from 1 to 4; perform element-wise summation on the global image difference matrix C and all the image difference matrices C ij ; mark the element points with the summation value greater than or equal to 4 as 1 to represent hair, and mark the element points with the summation value less than 4 as 0 to represent skin, obtaining the fusion matrix Z; calculate the proportion of the element points representing skin in the fusion matrix Z as the hock hair removal rate; determine the pressure difference between the pressure value of the hoof on the side close to the camera and the average pressure of the other three hooves based on the collected pressure data; perform weighted summation on the hock hair removal rate and the pressure difference to obtain a comprehensive score; determine the hock hair removal degree according to a preset determination rule based on the comprehensive score and a specific coefficient range threshold 6. The cattle fly festival hair removal detection method according to claim 5, wherein The four time-sharing excitation light sources are the No. 1 light source, the No. 2 light source, the No. 3 light source, and the No. 4 light source, which are distributed around the image acquisition unit and used for light source supplementation from four angles; the acquisition process of the multi-dimensional image data is specifically as follows: When the hock area of the cow is located at the horizontal center position of the imaging image, the synchronous control unit controls all four time-sharing excitation light sources to be turned on, and triggers the image acquisition unit to acquire the full-light source image; The synchronous control unit turns off the No. 2 light source, the No. 3 light source, and the No. 4 light source, waits for the afterglow to fade, and triggers the image acquisition unit to acquire the first single-light source image; The synchronous control unit turns off the No. 1 light source and turns on the No. 2 light source, waits for the afterglow to fade, and triggers the image acquisition unit to acquire the second single-light source image; The synchronous control unit turns off the No. 2 light source and turns on the No. 3 light source, waits for the afterglow to fade, and triggers the image acquisition unit to acquire the third single-light source image; The synchronous control unit turns off the No. 3 light source and turns on the No. 4 light source, waits for the afterglow to fade, and triggers the image acquisition unit to acquire the fourth single-light source image, and then turns off the No. 4 light source.

7. The cattle fly festival hair removal detection method according to claim 5 or 6, characterized in that, Based on the texture features, calculate the global image difference matrix C of the full-light source image and the image difference matrix C between any two single-light source images respectively ij , including: For the full-light source image and each single-light source image, the local binary pattern (LBP) value of each pixel is calculated; Based on a set threshold, the LBP value of each pixel in the full-light source image is binarized to generate the global image difference matrix C of the full-light source image; The four single-light source images are compared pairwise, and the comprehensive difference of the same pixel in any two single-light source images is calculated to obtain the corresponding difference image; among them, the difference image reflects the intensity of shadow change under two light sources; the comprehensive difference is determined by weighted summation of the LBP texture difference and the gray difference of the same pixel in the two single-light source images; For each difference image, perform binarization processing through a set threshold to generate an image difference matrix C ij .

8. The cattle fly festival hair removal detection method according to claim 7, characterized in that For each single-light source image, the LBP value of each pixel is calculated by the following formula: ； wherein, is the sign function, is the pixel gray value, is the pixel in the 3×3 neighborhood of the k gray value of the th pixel; For the full-light source image, the LBP value of each pixel is calculated by the following formula: ； Wherein, is the gray value of the pixel in the full-light source image , is the gray value of the -th pixel in the 3×3 neighborhood of the pixel p .

Images