Intelligent ewe estrus state monitoring system and method based on multi-mode ear thermal imaging

Abstract

The invention discloses an ewe estrus state intelligent monitoring system and method based on multi-mode ear thermal imaging, and belongs to the technical field of intelligent animal husbandry. The system comprises a hardware sensing module, a data analysis module and an output application module; the method comprises the following steps: S1, constructing a front-end preprocessing process; s2, constructing a cloud analysis platform; s3, designing a state judgment rule; s4, designing a user interface; therefore, the temperature information of the ear of the ewe is comprehensively captured and analyzed through multi-mode ear thermal imaging; meanwhile, advanced data analysis and artificial intelligence technologies are combined to intelligently process and interpret thermal imaging data, so that the estrus condition of the ewe is accurately judged, and the accuracy and reliability of estrus monitoring are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent animal husbandry technology, and in particular to an intelligent monitoring system and method for the estrus status of ewes based on multimodal ear thermal imaging. Background Art

[0002] In modern animal husbandry, monitoring estrus in ewes is a critical step in reproductive management, directly impacting conception rates and overall production efficiency. Currently, however, estrus monitoring still faces numerous technical drawbacks: traditional rectal thermometry produces a significant stress response (causing a 0.8-1.2°C increase in body temperature) and is inefficient (a single measurement takes 5-8 minutes); existing electronic ear tag systems can only monitor basal body temperature (with an error of ±0.5°C) and fail to capture ear microcirculatory characteristics; and conventional image recognition technology is sensitive to motion artifacts, resulting in a 35-40% decrease in recognition accuracy.

[0003] With the development of precision animal husbandry, sensor technology and artificial intelligence, intelligent monitoring methods based on multimodal data fusion have gradually become a research hotspot. Infrared thermal imaging technology, as a non-invasive, real-time monitoring method, can measure the body temperature of ewes without causing stress or tension in the sheep. It can reflect their internal physiological state by detecting the temperature distribution on the surface of the object.

[0004] However, in estrus monitoring of ewes, the ear is a temperature-sensitive area of ​​the body, and its temperature changes are closely related to the estrus status. Multimodal ear thermal imaging technology combines multiple imaging modes to more comprehensively capture and analyze the temperature information of the ewe's ear, thereby improving the accuracy and reliability of estrus monitoring.

[0005] Therefore, there is an urgent need for an intelligent monitoring system and method for the estrus status of ewes based on multimodal ear thermal imaging. Summary of the Invention

[0006] The present invention aims to provide an intelligent monitoring system and method for ewe estrus status based on multimodal ear thermal imaging to solve the problems raised in the above background technology.

[0007] In order to achieve the above object, the present invention provides the following technical solutions:

[0008] An intelligent monitoring system for ewe estrus status based on multimodal ear thermal imaging, including a hardware perception module, a data analysis module, and an output application module;

[0009] The hardware perception module is composed of an intelligent ear tag device, a motion compensation system, and a communication module, and is used to perceive and generate data and transmit it to the data analysis module;

[0010] The data analysis module receives front-end data transmitted from the hardware perception module, processes the front-end data using thermal image enhancement and key point detection, performs cloud model analysis using spatiotemporal feature fusion, and transmits the analyzed data to the output application module;

[0011] The output application module performs dynamic decision-making based on the received analysis data according to the weighted calculation of the estrus index and the three-level early warning mechanism, and renders the heat map in real time with WebGL and pushes the visualized decision results in a multi-segment collaborative manner based on hierarchical push.

[0012] Preferably, the smart ear tag device has a built-in sensor array, and the sensor array is a dual-spectrum infrared imaging module. The parameters of the dual-spectrum infrared imaging module are:

[0013] Working band: thermal imaging, 8-14μm, vascular imaging, 3-5μm;

[0014] Resolution: Thermal imaging, 320×240@30fps, vascular imaging, 160×120@15fps;

[0015] Temperature measurement range: 35-42℃, error is ±0.2℃.

[0016] Preferably, the motion compensation system is composed of a six-axis MEMS inertial sensor and a dynamic focusing mechanism, and has a built-in anti-shake algorithm;

[0017] The dynamic focusing mechanism is a piezoelectric ceramic driven mirror group with a displacement accuracy of ±5μm;

[0018] The anti-shake algorithm is based on real-time compensation of Kalman filtering, with a maximum compensation angle of ±25°.

[0019] Preferably, the communication module is configured as follows:

[0020] LoRaWAN transmission protocol, specifically the 868MHz frequency band;

[0021] Data encapsulation format: custom binary protocol, specifically header checksum + timestamp + sensor ID.

[0022] A method for an intelligent monitoring system for ewe estrus status based on multimodal ear thermal imaging, characterized in that the method steps include:

[0023] S1. Build a front-end pre-processing process to provide clear image data and refined feature data for subsequent processes through thermal image enhancement and feature extraction;

[0024] S2. Build a cloud-based analysis platform to analyze and process image data and feature data from front-end preprocessing by building a hybrid model architecture and model training strategy, and transmit it to the subsequent decision-making analysis process;

[0025] S3. Design status judgment rules, conduct final decision analysis on the data from the cloud analysis platform by designing an estrus index calculation method and establishing a graded early warning mechanism, and transmit the judgment results to the user interface;

[0026] S4. Design a user interface to display the final results to staff through data visualization, and establish a multi-channel alarm push mechanism, including SMS text messages, WeChat mini-program push, and pasture management platform pop-up windows, to show staff the specific status of the sheep in real time.

[0027] Preferably, in step S1, the thermal image is enhanced by applying the CLAHE algorithm in combination with motion artifact elimination, wherein:

[0028] The settings of the CLAHE algorithm are a grid size of 32 × 32 and a contrast limit of 2.0;

[0029] Motion artifact removal: Use optical flow to calculate the displacement between adjacent frames and apply affine transformation compensation.

[0030] Preferably, the feature extraction in step S1 is specifically as follows:

[0031] Ear key point detection: Build a U-Net segmentation model with an input size of 256×256; output five anatomical landmarks, including the ear tip, ear base, and external auditory canal entrance;

[0032] Dynamic characteristic calculations, including temperature gradient, fluctuation index, and vascular filling, are calculated using the following formula:

[0033] Temperature gradient: △T = T_tympanic - T_ambient;

[0034] Fluctuation index: FI = σ(T) / μ(T);

[0035] Vascular filling index: VFI = ∑(pixel value > threshold) / ROI area.

[0036] Preferably, the construction of the hybrid model architecture in step S2 includes a temporal processing branch and a spatial feature branch, specifically:

[0037] Time series processing branch: 3-layer BiLSTM, 128 hidden units; temporal attention mechanism, scaled dot product attention;

[0038] Spatial feature branch: ResNet-18 backbone network, ImageNet pre-training; Feature Pyramid Network integrates multi-scale features.

[0039] Preferably, the multi-feature weighted formula for calculating the estrus index in step S3 is:

[0040] E index=0.4*ΔT+0.3*FI+0.2*VFI+0.1*Activity

[0041] Preferably, the hierarchical warning mechanism in step S3 specifically includes three processes: warning level, E index range and response strategy, specifically:

[0042] Warning levels: Level I, Level II and Level III;

[0043] E index range: corresponding to level I is 0.7-1.0; level II is 0.5-0.7; level III is 0.3-0.5;

[0044] Response strategy: Level I is to immediately notify the breeder; Level II is to initiate secondary verification and start video analysis; Level III is to record observation logs.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] This invention uses multimodal ear thermal imaging to comprehensively capture and analyze ewe ear temperature information. It also combines advanced data analysis and artificial intelligence technologies to intelligently process and interpret thermal imaging data, accurately determining the ewe's estrus status. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 The present invention is a flow chart of an intelligent monitoring system and method for ewe estrus status based on multimodal ear thermal imaging. DETAILED DESCRIPTION

[0048] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments:

[0049] like Figure 1 As shown, an intelligent monitoring system for ewe estrus status based on multimodal ear thermal imaging includes a hardware perception module, a data analysis module and an output application module;

[0050] Hardware perception module: It consists of an intelligent ear tag device, a motion compensation system, and a communication module. It is used to perceive and generate data and transmit it to the data analysis module.

[0051] Data analysis module: Receives front-end data transmitted from the hardware perception module, processes the front-end data using thermal image enhancement and key point detection, performs cloud-based model analysis using spatiotemporal feature fusion, and transmits the analyzed data to the output application module;

[0052] Output application module: The received analysis data is used for dynamic decision-making based on the weighted calculation of the estrus index and the three-level early warning mechanism, and the decision results are visualized through the real-time rendering of the heat map and the multi-segment collaborative push of the hierarchical push based on WebGL.

[0053] The method of the intelligent monitoring system for ewe estrus status based on multimodal ear thermal imaging is specifically as follows:

[0054] S1. Build a front-end pre-processing process to provide clear image data and refined feature data for subsequent processes through thermal image enhancement and feature extraction;

[0055] S2. Build a cloud-based analysis platform to analyze and process image data and feature data from front-end preprocessing by building a hybrid model architecture and model training strategy, and transmit it to the subsequent decision-making analysis process;

[0056] S3. Design status judgment rules, conduct final decision analysis on the data from the cloud analysis platform by designing an estrus index calculation method and establishing a graded early warning mechanism, and transmit the judgment results to the user interface;

[0057] S4. Design a user interface to display the final results to staff through data visualization, and establish a multi-channel alarm push mechanism, including SMS text messages, WeChat mini-program push, and pasture management platform pop-up windows, to show staff the specific status of the sheep in real time.

[0058] The specific implementation process is as follows:

[0059] 1. Hardware Architecture Design

[0060] 1. Smart ear tag device

[0061] 1.1 Sensor Array

[0062] Dual-spectrum infrared imaging module

[0063] Working band: 8-14μm (thermal imaging) + 3-5μm (vascular imaging)

[0064] Resolution: 320×240@30fps (thermal imaging) / 160×120@15fps (vascular imaging)

[0065] Temperature measurement range: 35-42℃ (accuracy ±0.2℃)

[0066] 1.2 Motion Compensation System

[0067] Six-axis MEMS inertial sensor (MPU-6050)

[0068] Dynamic focusing mechanism: piezoelectric ceramic driven mirror group (displacement accuracy ±5μm)

[0069] Anti-shake algorithm: Real-time compensation based on Kalman filter (maximum compensation angle ±25°)

[0070] 1.3 Communication Module

[0071] LoRaWAN transmission protocol (868MHz frequency band)

[0072] Data encapsulation format: custom binary protocol (header checksum + timestamp + sensor ID)

[0073] 2. Data Processing Flow

[0074] 2.1 Front-end Preprocessing (Edge Computing Unit)

[0075] 2.1.1 Thermal Image Enhancement

[0076] Apply the CLAHE algorithm (grid size 32×32, contrast limit 2.0)

[0077] Motion artifact removal: Use optical flow to calculate the displacement between adjacent frames and apply affine transformation compensation

[0078] 2.1.2 Feature Extraction

[0079] Ear key point detection:

[0080] Build a U-Net segmentation model (input size 256×256)

[0081] Output 5 anatomical landmarks (ear tip, ear root, external auditory canal entrance, etc.)

[0082] Dynamic feature calculation is shown in the following table:

[0083] Feature Type Calculation formula unit Temperature gradient △T=T_tympanic-T_ambient ℃ Volatility Index FI=σ(T) / μ(T) ％ Vascular turgor VFI = ∑(pixel value > threshold) / ROI area ％

[0084] 2.2 Cloud Analysis Platform

[0085] 2.2.1 Hybrid Model Architecture

[0086] Temporal processing branch: 3-layer BiLSTM (128 hidden units); temporal attention mechanism (scaled dot product attention).

[0087] Spatial feature branch: ResNet-18 backbone network (ImageNet pre-training); Feature Pyramid Network (FPN) integrates multi-scale features.

[0088] 2.2.2 Model Training Strategy

[0089] The dataset composition is shown in the following table:

[0090] variety Number of samples Data duration Positive-negative sample ratio Dorper sheep 120 90 days 1:3 Lake Sheep 80 60 days 1:2 Suffolk sheep 50 45 days 1:1.5

[0091] 3. Decision Output System

[0092] 3.1 Status determination rules

[0093] 3.1.1 Calculation of estrus index

[0094] Multi-feature weighted formula:

[0095] E_index=0.4*ΔT+0.3*FI+0.2*VFI+0.1*Activity

[0096] 3.1.2 The hierarchical early warning mechanism is shown in the following table:

[0097] Warning level E_index range Response Strategy Level I 0.7-1.0 Immediate notification of breeding Level II 0.5-0.7 Start secondary verification (video analysis) Level III 0.3-0.5 Record observation log

[0098] 3.2 User Interface Design

[0099] 3.2.1 Data Visualization

[0100] Heatmap dynamic rendering (WebGL acceleration)

[0101] Time series analysis panel (supports 7 / 30-day trend viewing)

[0102] 3.2.2 Alarm Push

[0103] Multi-channel notification:

[0104] SMS text messages (emergency)

[0105] WeChat applet push (regular reminder)

[0106] Pasture management platform pop-up window (real-time monitoring).

[0107] The above is only an embodiment of the present invention, and the common knowledge such as the specific technical solutions and / or characteristics in the solution are not described in detail here. It should be pointed out that for those skilled in the art, without departing from the technical solution of the present invention, several variations and improvements can be made, which should also be regarded as the scope of protection of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the description can be used to interpret the content of the claims.

Claims

1. An intelligent monitoring system for ewe estrus status based on multimodal ear thermal imaging, characterized by: The system includes a hardware perception module, a data analysis module and an output application module; The hardware perception module is composed of an intelligent ear tag device, a motion compensation system, and a communication module, and is used to perceive and generate data and transmit it to the data analysis module; The data analysis module receives front-end data transmitted from the hardware perception module, processes the front-end data using thermal image enhancement and key point detection, performs cloud model analysis using spatiotemporal feature fusion, and transmits the analyzed data to the output application module; The output application module performs dynamic decision-making based on the received analysis data according to the weighted calculation of the estrus index and the three-level early warning mechanism, and renders the heat map in real time with WebGL and pushes the visualized decision results in a multi-segment collaborative manner based on hierarchical push.

2. The intelligent monitoring system for ewe estrus status based on multimodal ear thermal imaging according to claim 1, characterized in that: The smart ear tag device has a built-in sensor array, which is a dual-spectrum infrared imaging module. The parameters of the dual-spectrum infrared imaging module are: Working band: thermal imaging, 8-14μm, vascular imaging, 3-5μm; Resolution: Thermal imaging, 320×240@30fps, vascular imaging, 160×120@15fps; Temperature measurement range: 35-42℃, error is ±0.2℃.

3. The intelligent monitoring system for ewe estrus status based on multimodal ear thermal imaging according to claim 1, characterized in that: The motion compensation system consists of a six-axis MEMS inertial sensor and a dynamic focusing mechanism, and has a built-in anti-shake algorithm; The dynamic focusing mechanism is a piezoelectric ceramic driven mirror group with a displacement accuracy of ±5μm; The anti-shake algorithm is based on real-time compensation of Kalman filtering, with a maximum compensation angle of ±25°.

4. The intelligent monitoring system for ewe estrus status based on multimodal ear thermal imaging according to claim 1, characterized in that: The communication module is configured as follows: LoRaWAN transmission protocol, specifically the 868MHz frequency band; Data encapsulation format: custom binary protocol, specifically header checksum + timestamp + sensor ID.

5. A method for an intelligent monitoring system for ewe estrus status based on multimodal ear thermal imaging, characterized in that: The method steps include: S1. Build a front-end pre-processing process to provide clear image data and refined feature data for subsequent processes through thermal image enhancement and feature extraction; S2. Build a cloud-based analysis platform to analyze and process image data and feature data from front-end preprocessing by building a hybrid model architecture and model training strategy, and transmit it to the subsequent decision-making analysis process; S3. Design status judgment rules, conduct final decision analysis on the data from the cloud analysis platform by designing an estrus index calculation method and establishing a graded early warning mechanism, and transmit the judgment results to the user interface; S4. Design a user interface to display the final results to staff through data visualization, and establish a multi-channel alarm push mechanism, including SMS text messages, WeChat mini-program push, and pasture management platform pop-up windows, to show staff the specific status of the sheep in real time.

6. The method of the intelligent monitoring system for ewe estrus status based on multimodal ear thermal imaging according to claim 5, characterized in that: In step S1, the thermal image is enhanced by applying the CLAHE algorithm in combination with motion artifact elimination, wherein: The settings of the CLAHE algorithm are a grid size of 32 × 32 and a contrast limit of 2.0; Motion artifact removal: Use optical flow to calculate the displacement between adjacent frames and apply affine transformation compensation.

7. The method of the intelligent monitoring system for ewe estrus status based on multimodal ear thermal imaging according to claim 5, characterized in that: The feature extraction in step S1 is specifically as follows: Ear key point detection: Build a U-Net segmentation model with an input size of 256×256; output five anatomical landmarks, including the ear tip, ear base, and external auditory canal entrance; Dynamic characteristic calculations, including temperature gradient, fluctuation index, and vascular filling, are calculated using the following formula: Temperature gradient: △T = T_tympanic - T_ambient; Fluctuation index: FI = σ(T) / μ(T); Vascular filling index: VFI = ∑(pixel value > threshold) / ROI area.

8. The method of the intelligent monitoring system for ewe estrus status based on multimodal ear thermal imaging according to claim 5, characterized in that: The construction of the hybrid model architecture in step S2 includes a temporal processing branch and a spatial feature branch, specifically: Time series processing branch: 3-layer BiLSTM, 128 hidden units; temporal attention mechanism, scaled dot product attention; Spatial feature branch: ResNet-18 backbone network, ImageNet pre-training; Feature Pyramid Network fuses multi-scale features.

9. The method of the intelligent monitoring system for ewe estrus status based on multimodal ear thermal imaging according to claim 5, characterized in that: The multi-feature weighted formula for calculating the estrus index in step S3 is: E index=0.4*ΔT+0.3*FI+0.2*VFI+0.1*Activity 10. The method of the intelligent monitoring system for ewe estrus status based on multimodal ear thermal imaging according to claim 5, characterized in that: The hierarchical warning mechanism in step S3 specifically includes three processes: warning level, E index range and response strategy, specifically: Warning levels: Level I, Level II and Level III; E index range: 0.7-1.0 corresponding to level I; Grade II: 0.5-0.7; Grade III: 0.3-0.5; Response strategy: corresponding to level I is immediate notification of breeding; Level II initiates secondary verification and video analysis; Level III records observation logs.

Images