A negative pressure insect detection and imaging system and method for a sex pheromone detector

Abstract

This invention relates to the field of insect pest detection and identification technology, specifically a negative pressure insect fixation and imaging system and method for a pheromone trapping instrument. Based on the joint swaying force that the target pest can exert when in a predicted morphological state, the feasibility of eliminating negative pressure resistance is determined by the main and sub-lines, and the cutting plane is statistically analyzed. Negative pressure resistance parameters are output, and a high-power fan is controlled to draw air from the insect-collecting channel to fix the target pest. The binary image information of the supplementary lighting captured when the target pest falls into the insect-collecting channel and occupies the grid area is inverted to construct a positive-negative supplementary lighting voxel pattern. This is combined with the target pest's ideal negative pressure fixation posture and the positive-negative supplementary lighting voxel pattern occupancy analysis to instantaneously control negative pressure insect fixation and supplementary lighting imaging by a high-definition industrial camera. This invention can stably fix the target pest under negative pressure and accurately photograph it, effectively improving the imaging quality and information recognition accuracy of individual pests.

Description

Technical Field

[0001] This invention relates to the field of insect infestation detection and identification technology, and in particular to a negative pressure insect identification and photography system and method for a pheromone detection and reporting instrument. Background Technology

[0002] With the development of agricultural pest monitoring and green control technologies, pheromone-based pest monitoring instruments are widely used for precise early warning and visual data collection of agricultural pests and diseases. These instruments primarily rely on releasing specific pheromones to lure target pests into the device, where they are then captured and counted through physical means. This helps agricultural technicians understand pest dynamics and rationally schedule control measures. Currently, most pheromone-based pest monitoring devices rely on manual, periodic inspections, recording, and cleaning, which suffers from high labor intensity, untimely data updates, and significant subjective errors. While existing pheromone trapping devices incorporate automatic image capture technology, the insects' positions are not fixed after falling into the trapping area, easily leading to blurry images or duplicate identification, affecting data accuracy. Furthermore, most devices rely on natural descent or adhesive plate fixation, making it difficult to effectively control the insects' posture and position, increasing the processing difficulty of image recognition algorithms. Simultaneously, the lack of an effective and stable insect-fixing mechanism makes it easy for external factors such as wind or vibration to interfere with insect positioning and imaging. In addition, although negative pressure attraction methods are used in other sampling devices, existing negative pressure insect-fixing control has unreasonable deviations, making it difficult to apply appropriate negative pressure based on the insect's body size, easily causing insect injury or insecure fixation. Summary of the Invention

[0003] This invention overcomes the shortcomings of the prior art and provides a negative pressure insect identification and imaging system and method for a sex pheromone detection instrument.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] The first aspect of this invention provides a negative pressure insect detection and imaging method for a pheromone trap, comprising the following steps:

[0006] S102: Obtain the historical physical morphology parameters of the target pests captured by the sex pheromone trapping instrument. Based on the transition distribution hash bucket of the standard physical morphology of the target pests under different seasonal conditions, perform time-series growth lag calculation on the historical physical morphology parameters to obtain a relatively standardized growth lag matrix of the historical physical morphology of the target pests.

[0007] S104: Based on the dynamic change tensor structure expansion decomposition of the growth lag matrix, obtain the dynamic growth factor load matrix of the target pest's directional growth at the current seasonal time. Construct an autoregressive model of body morphology by estimating and accumulating the contribution of the dynamic growth factor load matrix. Use the regression operation of the autoregressive model of body morphology to obtain the estimated value of the body morphology of the target pest trapped by the sex pheromone trap at the current seasonal time.

[0008] S106: Based on the joint swinging force that the target pest can exert when it is in the predicted body shape, the feasibility of eliminating the main line and sub-line of negative pressure resistance is determined and the cutting plane is statistically analyzed. The negative pressure resistance parameters are output, and the high-power fan is controlled to draw air from the insect receiving channel to fix the target pest based on the negative pressure resistance parameters.

[0009] S108: By reversing the information of the supplementary lighting binary image captured when the target pest falls into the grid area of ​​the insect receiving channel flip plate, a positive-negative supplementary lighting voxel pattern is constructed. Combined with the ideal negative pressure fixation posture of the target pest and the occupation analysis of the positive-negative supplementary lighting voxel pattern, the negative pressure insect fixation and the supplementary lighting of the high-definition industrial camera are controlled in real time.

[0010] More specifically, step S102 includes the following steps:

[0011] The research needs of the target pests and the trapping logs of the sex pheromone detector are obtained, and the preset sex pheromone release strategy of the sex pheromone detector for the research needs is obtained by identification.

[0012] By extracting historical morphological parameters of the target pests when the sex pheromone release strategy was executed on a specified time chain by the sex pheromone trapping instrument, a knowledge graph of insect mating was obtained based on big data.

[0013] Based on the insect mating knowledge graph, several historical physical morphology parameters are indexed and analyzed to output the baseline mating behavior characteristics of target pests with different physical morphologies at different seasonal change periods. Based on the baseline mating behavior characteristics, a family of transition hash functions is pre-defined to show that physical morphology produces significant differences under the dominance of seasonal change.

[0014] Construct a transition hash map of the standard changes in the physical morphology of the target pest under different seasonal conditions, and obtain reference values ​​of the standard physical morphology of the target pest under different seasonal conditions through an insect mating knowledge graph.

[0015] Based on the family of transition hash functions, each of the standard physical morphology reference values ​​is hashed into a signature symbol to obtain several standard transition hash signatures and the transition hash value of each standard transition hash signature. The standard physical morphology reference values ​​corresponding to the standard transition hash signatures with the same transition hash value are put into the same bucket to obtain multiple transition distribution hash buckets.

[0016] Calculate the combined hash value of each historical body shape parameter in the transition hash map, find the transition hash bucket with the combined hash value attached, mark it as the target transition distribution hash bucket, and define the standard body shape reference value in the target transition distribution hash bucket as the candidate reference value.

[0017] Calculate the adjacency distance between each candidate reference value and each historical morphological parameter, and interpolate based on the adjacency distance to fit the growth lag matrix of the historical morphological form of the target pest on the specified time series, which generates the seasonal transition compared to the standard morphological form.

[0018] More specifically, step S104 includes the following steps:

[0019] The dynamic change tensor chain of the target pest's growth accompanied by seasonal fluctuations is obtained based on the growth lag matrix. The dynamic change tensor chain is expanded and the influencing factors are singularly decomposed based on the directional growth criteria that meet the research needs at the current seasonal time, so as to obtain the dynamic growth factor loading matrix of the target pest's directional growth at each hierarchical dimension.

[0020] The core tensor of the target pest that tends to stabilize under the premise of being disturbed by different dynamic growth factors is defined as the first dynamic growth change core tensor.

[0021] Multiply each of the dynamic growth factor loading matrices with the dynamic transition tensor chain n-mode to obtain the dynamic growth transition core tensor, which is defined as the second dynamic growth transition core tensor. Calculate the difference between the first dynamic growth transition core tensor and the first dynamic growth transition core tensor to obtain the core tensor difference value.

[0022] Construct a cumulative contribution domain, estimate each dynamic growth factor in the dynamic growth factor loading matrix based on the core tensor difference value, obtain multiple dynamic growth factor estimates, and simultaneously accumulate the contribution rate of all dynamic growth factors in the cumulative contribution domain;

[0023] During the cumulative contribution process, only dynamic growth factors whose estimated value is greater than the preset estimated value are extracted and marked as key prominent factors. If the cumulative contribution rate reaches the preset cumulative contribution rate, the extraction process is stopped, and one or K key prominent factors are obtained.

[0024] A dynamic growth transition mapping space is constructed based on one or K key prominent factors. The lag matrix is ​​then projected onto this dynamic growth transition mapping space to establish an autoregressive model of body morphology.

[0025] The dynamic growth of the target pest at the current seasonal time is calculated by using an autoregressive model of physical morphology, and a physical morphology regression score matrix is ​​generated. Based on the physical morphology regression score matrix, the estimated value of the physical morphology of the target pest trapped by the sex pheromone trap at the current seasonal time is determined.

[0026] More specifically, the process of obtaining the dynamic transition tensor chain of the target pest's growth accompanied by seasonal fluctuations based on the growth lag matrix, unfolding the dynamic transition tensor chain based on the directional growth criteria that meet the research requirements at the current seasonal time, and singularly decomposing the influencing factors to obtain the dynamic growth factor loading matrix of the target pest's directional growth at each hierarchical dimension, specifically includes the following steps:

[0027] Based on the growth lag matrix, the lag tensor hierarchy structure of the seasonal transition dynamic characteristics of the body morphology is extracted, and the dynamic transition tensor chain of the target pest's growth accompanied by seasonal fluctuations is determined according to the lag tensor hierarchy structure.

[0028] The current season and time of the sex pheromone detection instrument are obtained. Based on the big data network, the directional growth criteria of the target pests in the current season and the timing of growth mutations that follow the directional growth criteria are obtained.

[0029] Based on the directional growth criterion, the target pest is located in different hierarchical dimensions at the current season. Simultaneously, based on the actual nodes of growth mutation, the target growth rank necessary for the target pest in each hierarchical dimension is set. The covariance expansion of the dynamic transition tensor chain is performed on each hierarchical dimension to obtain multiple tensor covariance expansion matrices.

[0030] A singular decomposition algorithm is introduced. By taking the factor singular value composed of the left singular vector of each tensor covariance expansion matrix, the dynamic growth factor loading matrix corresponding to the output of each tensor covariance expansion matrix when the target pest is growing in each hierarchical dimension is obtained.

[0031] More specifically, step S106 includes the following steps:

[0032] Based on the predicted body morphology, the joint swinging force of each joint part under the influence of sex pheromones during the normal activity of the target pest is retrieved from the big data network. The joint swinging force is defined as the target elimination variable. At the same time, the predetermined negative pressure adjustment decision of the high-power fan is obtained.

[0033] A relaxed mainline model for resisting joint swing force under negative pressure is constructed. Only one elimination estimation variable is preset, considering only the target elimination variable. The predetermined negative pressure inhibition index for the joint swing force of the target pest generated by the high-power fan for the body shape prediction is extracted through the predetermined negative pressure parameter adjustment decision. The predetermined negative pressure inhibition index is defined as the original constraint limit.

[0034] The relaxation principal model is solved by eliminating the estimated variables to obtain the current target elimination variable solution of the relaxation principal. The current target elimination variable solution is substituted into the original constraint limit to construct a sub-line model for eliminating the target elimination variable with a predetermined negative pressure resistance index. The allowable safe inhibition range of the target pest is preset under negative pressure.

[0035] If the sub-line model can completely eliminate the target elimination variable, then the sub-line dual variable is introduced into the sub-line model to generate the optimal cut for eliminating the target elimination variable; if the sub-line model cannot completely eliminate the target elimination variable, then the dual infeasible variable is used to eliminate the current target elimination variable solution one by one to generate a feasible cut.

[0036] Repeat the above steps of solving the relaxation principal model and eliminating the sub-model, add the generated optimal cut and feasible cut to the relaxation principal model, and obtain the current negative pressure resistance value of the relaxation principal model and the current optimal elimination value of the sub-model in real time.

[0037] If the relative error between the current negative pressure resistance value and the current optimal elimination value does not exceed the allowable safe suppression range, then stop adding optimal cuts and feasible cuts, and finally output the negative pressure resistance parameter. Based on the negative pressure resistance parameter, control the high-power fan to draw air from the insect receiving channel to resist the joint swinging force of the fixed target pest.

[0038] More specifically, step S108 includes the following steps:

[0039] Obtain the preset resolution of the high-definition camera and the spatial model of the insect-catching channel flap. Based on the preset resolution, discretize and plan several voxel grids on the spatial model to construct the occupied grid area of ​​the insect-catching channel flap.

[0040] The supplementary lighting binary image information of the target pest falling into the neighborhood of the occupied grid is obtained by a high-definition camera. The 1 supplementary lighting voxel grid that is not occupied by the target pest in the supplementary lighting binary image information is defined as a free space point, and the 0 supplementary lighting voxel grid that is occupied by the target pest is defined as an obstacle point.

[0041] The Euclidean distance transformation algorithm is introduced to calculate and store the shortest Euclidean distance from each free space point to the nearest adjacent obstacle point, and the distance calculation of the obstacle point itself is automatically ignored and marked as 0, generating a positive unsigned distance field.

[0042] During computation and storage, the obstacle points and free space points in the supplementary lighting binary image information are reversed and swapped, and the distance field is recalculated to generate an inverse unsigned distance field. Based on the inverse unsigned distance field, it is determined whether each supplementary lighting voxel grid is located inside the obstacle point.

[0043] If the fill light voxel grid is located inside the obstacle point, then the fill light voxel grid is assigned a positive sign; if the fill light voxel grid is located outside the obstacle point, then the fill light voxel grid is assigned a negative sign, resulting in a positive-negative fill light voxel pattern.

[0044] Based on research needs, obtain the ideal negative pressure fixation posture that can capture the expected identification information of the target pest, and obtain the range of degrees of freedom of the target pest's corresponding joint parts when making the ideal negative pressure fixation posture based on big data.

[0045] A Bayesian voxel inference network was constructed using the Bayesian voxel update rule. Based on the activity degree of freedom interval, the occupancy of joint voxels of the target pest in the negative pressure fixed posture was calculated in the Bayesian voxel inference network to obtain the occupancy probability of each supplementary light voxel grid.

[0046] If the positive-negative fill light voxel pattern shows a negative sign in the fill light voxel grid where the occupancy probability is greater than the preset occupancy probability, then the high-power fan is instantly controlled to fix the target pest on the insect channel flap with negative pressure and to take a picture with the fill light from the high-definition industrial camera.

[0047] The second aspect of the present invention provides a negative pressure insect identification and imaging system for a sex pheromone detector. The negative pressure insect identification and imaging system includes a memory and a processor. The memory stores a negative pressure insect identification and imaging method program for a sex pheromone detector. When the negative pressure insect identification and imaging method program is executed by the processor, the steps of the negative pressure insect identification and imaging method described in any one of the present invention are implemented.

[0048] This invention addresses the technical deficiencies in the prior art, and its beneficial technical effects are as follows:

[0049] Historical morphological parameters of the target pest when it was captured by the sex pheromone trap are obtained. Based on the transition distribution hash bucket of the target pest's standard morphological changes under different seasonal conditions, the historical morphological parameters are used to calculate the temporal growth lag of the historical morphological parameters, resulting in a relatively standardized growth lag matrix of the target pest's historical morphological form. The dynamic change tensor structure of the growth lag matrix is ​​expanded and decomposed to obtain the dynamic growth factor loading matrix of the target pest's directional growth at the current seasonal moment. An autoregressive model of morphological form is constructed by estimating and accumulating the contributions of the dynamic growth factor loading matrix. The regression operation of the autoregressive model of morphological form is used to obtain the morphological characteristics of the sex pheromone trap at the current seasonal moment. The system estimates the physical morphology of the trapped target pest. Based on the joint swinging force that the target pest can exert when in the estimated physical morphology state, it determines the feasibility of eliminating the main and sub-lines of negative pressure resistance and performs plane segmentation statistics, outputting negative pressure resistance parameters. Based on the negative pressure resistance parameters, it controls a high-power fan to draw air from the insect-catching channel to fix the target pest. It constructs a positive-negative supplementary light voxel pattern by reversing the information of the supplementary light binary image captured when the target pest falls into the insect-catching channel flap and occupies the grid area. Combined with the ideal negative pressure fixation posture of the target pest and the positive-negative supplementary light voxel pattern occupancy analysis, it enables instantaneous control of negative pressure insect fixation and supplementary light shooting by a high-definition industrial camera. This invention can predict the morphological data that the pheromone trapping device may capture at the current moment based on the dynamic growth fluctuations of the target pest's historical morphological data with different seasonal changes. This makes the negative pressure fixation of the target pest more accurate and stable, avoiding the phenomenon of damage to the target pest or unreliable fixation due to unreasonable negative pressure control. At the same time, it can accurately control the timing of negative pressure fixation and photography, ensuring that the effective identity information of the target pest can be captured, and significantly improving the imaging quality and target pest recognition rate of the equipment. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained from these drawings without creative effort.

[0051] Figure 1 A flowchart of the first method of a negative pressure insect identification and photography method for a sex pheromone detector is shown;

[0052] Figure 2 A flowchart of the second method of a negative pressure insect identification and photography method for a sex pheromone detector is shown;

[0053] Figure 3 A system framework diagram of a negative pressure insect-detecting and photographing system for a sex pheromone detector is shown. Detailed Implementation

[0054] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0055] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0056] The first aspect of this invention provides a negative pressure insect detection and imaging method for a sex pheromone detector, such as... Figure 1 As shown, it includes the following steps:

[0057] S102: Obtain the historical physical morphology parameters of the target pests captured by the sex pheromone trapping instrument. Based on the transition distribution hash bucket of the standard physical morphology of the target pests under different seasonal conditions, perform time-series growth lag calculation on the historical physical morphology parameters to obtain a relatively standardized growth lag matrix of the historical physical morphology of the target pests.

[0058] S104: Based on the dynamic change tensor structure expansion decomposition of the growth lag matrix, obtain the dynamic growth factor load matrix of the target pest's directional growth at the current seasonal time. Construct an autoregressive model of body morphology by estimating and accumulating the contribution of the dynamic growth factor load matrix. Use the regression operation of the autoregressive model of body morphology to obtain the estimated value of the body morphology of the target pest trapped by the sex pheromone trap at the current seasonal time.

[0059] S106: Based on the joint swinging force that the target pest can exert when it is in the predicted body shape, the feasibility of eliminating the main line and sub-line of negative pressure resistance is determined and the cutting plane is statistically analyzed. The negative pressure resistance parameters are output, and the high-power fan is controlled to draw air from the insect receiving channel to fix the target pest based on the negative pressure resistance parameters.

[0060] S108: By reversing the binary image information of supplementary lighting captured when the target pest falls into the grid area of ​​the insect-catching channel flip plate, a positive-negative supplementary lighting voxel pattern is constructed. Combined with the ideal negative pressure fixation posture of the target pest and the occupancy analysis of the positive-negative supplementary lighting voxel pattern, the negative pressure insect fixation and supplementary lighting imaging of the high-definition industrial camera are used for instantaneous control of negative pressure insect fixation.

[0061] More specifically, step S102 includes the following steps:

[0062] The research needs of the target pests and the trapping logs of the sex pheromone detector are obtained, and the preset sex pheromone release strategy of the sex pheromone detector for the research needs is obtained by identification.

[0063] By extracting historical morphological parameters of the target pests when the sex pheromone release strategy was executed on a specified time chain by the sex pheromone trapping instrument, a knowledge graph of insect mating was obtained based on big data.

[0064] Based on the insect mating knowledge graph, several historical physical morphology parameters are indexed and analyzed to output the baseline mating behavior characteristics of target pests with different physical morphologies at different seasonal change periods. Based on the baseline mating behavior characteristics, a family of transition hash functions is pre-defined to show that physical morphology produces significant differences under the dominance of seasonal change.

[0065] Construct a transition hash map of the standard changes in the physical morphology of the target pest under different seasonal conditions, and obtain reference values ​​of the standard physical morphology of the target pest under different seasonal conditions through an insect mating knowledge graph.

[0066] Based on the family of transition hash functions, each of the standard physical morphology reference values ​​is hashed into a signature symbol to obtain several standard transition hash signatures and the transition hash value of each standard transition hash signature. The standard physical morphology reference values ​​corresponding to the standard transition hash signatures with the same transition hash value are put into the same bucket to obtain multiple transition distribution hash buckets.

[0067] Calculate the combined hash value of each historical body shape parameter in the transition hash map, find the transition hash bucket with the combined hash value attached, mark it as the target transition distribution hash bucket, and define the standard body shape reference value in the target transition distribution hash bucket as the candidate reference value.

[0068] Calculate the adjacency distance between each candidate reference value and each historical morphological parameter, and interpolate based on the adjacency distance to fit the growth lag matrix of the historical morphological form of the target pest on the specified time series, which generates the seasonal transition compared to the standard morphological form.

[0069] It is important to note that the key to negative pressure insect control lies in the appropriate level of pressure applied to the pests. Different pests vary in appearance, size, and morphology, thus requiring different levels of negative pressure. For example, larger pests require slightly higher negative pressure to prevent their movement. However, seasonal fluctuations mean that the size parameters of pests captured by pheromone traps are not uniform each time. Traditional negative pressure insect control systems lack the ability to estimate pest size, making it difficult to precisely control the negative pressure based on the approximate size of the pests to be controlled. This can easily lead to deviations such as excessive or insufficient negative pressure, damaging the pests. Therefore, a rough estimate of the pest's appearance and morphology parameters is crucial. The growth and changes in the pest's appearance and morphology are dynamically influenced by seasonal fluctuations at different stages. Traditional insect size prediction models fail to fully consider this key dynamic factor of seasonal fluctuations, making it difficult to accurately capture the dynamic characteristics of size transitions at different seasonal growth stages, significantly reducing the accuracy of pest size prediction. To address this, this method uses a transition hash map representing the standard changes in the physical morphology of target pests under different seasonal conditions as a temporal hotspot framework. Within this transition hash map, corresponding hash buckets are used to locate the reference values ​​of the standard physical morphology of target pests under different seasonal conditions, thus displaying a temporal hotspot distribution blueprint of the stable standard changes in the physical morphology of target pests under different seasonal fluctuations. Then, using this defined hotspot blueprint, the seasonal transitions of several historical physical morphological parameters of target pests trapped by the sex pheromone trap are tracked and the distance between these transitions is calculated. This clarifies the growth lag matrix of the seasonal transitions in the historical physical morphology of target pests compared to the standard physical morphology, further revealing the dynamic characteristics of the transitions in the physical morphology of pests over time as the sex pheromone trapping season changes. This ensures the dynamic capture accuracy of subsequent target pest size prediction, improves the reliability of physical morphology estimation, and makes negative pressure insect trapping more stable and reasonable.

[0070] It should be noted that the target pests include, but are not limited to, the rice stem borer and the rice leaf roller. The transition hash function family is used to map similar data points to the same hash value with a high probability, ensuring that subsequent hash operations can effectively capture the temporal correlation between standard body morphologies based on specific seasonal variation measures. Furthermore, since the target pests are induced by sex pheromones, they exhibit certain mating behavior patterns. Therefore, a transition hash function family is pre-defined based on baseline mating behavior characteristics to identify significant differences in body morphology under seasonal variation. Then, the specified transition hash function family is used to hash each standard body morphology reference value into a signature, thereby mapping the high-dimensional temporal distribution vector of the standard body morphology to a low-dimensional signature form. This reduces accidental collisions in body size transitions caused by temporal conflicts, improving the accuracy of depicting the body size growth of target pests following standardized temporal dynamic changes. Next, based on the transition hash value and the standard transition hash signature, points with the same value are placed in the same bucket. The location pattern of the hash buckets constitutes a temporal transition distribution map of the standardized changes in the body morphology of the target pest under different seasonal fluctuations, thereby determining the temporal seasonal node of each standard body morphology. This can be understood as a distribution pattern similar to a heat map, providing a reliable index benchmark for the dynamic positioning of subsequent historical data. Subsequently, the target transition distribution hash buckets with combined hash values ​​of historical body morphology parameters are searched. The standard body morphology reference values ​​in the target transition distribution hash buckets are candidate reference values ​​that are in the same seasonal time series as the historical body morphology parameters. By calculating the adjacency distance between each candidate reference value and the historical body morphology parameters, the lag in the growth of the target pest's body size based on the historical body morphology compared to the standard body morphology at the same time can be revealed. This allows for the accurate capture of the dynamic growth characteristics of the differences in body size transitions of target pests trapped by the sex pheromone trap in different seasons, effectively improving the accuracy and reliability of target pest body size estimation.

[0071] More specifically, step S104 includes the following steps:

[0072] The dynamic change tensor chain of the target pest's growth accompanied by seasonal fluctuations is obtained based on the growth lag matrix. The dynamic change tensor chain is expanded and the influencing factors are singularly decomposed based on the directional growth criteria that meet the research needs at the current seasonal time, so as to obtain the dynamic growth factor loading matrix of the target pest's directional growth at each hierarchical dimension.

[0073] The core tensor of the target pest that tends to stabilize under the premise of being disturbed by different dynamic growth factors is defined as the first dynamic growth change core tensor.

[0074] Multiply each of the dynamic growth factor loading matrices with the dynamic transition tensor chain n-mode to obtain the dynamic growth transition core tensor, which is defined as the second dynamic growth transition core tensor. Calculate the difference between the first dynamic growth transition core tensor and the first dynamic growth transition core tensor to obtain the core tensor difference value.

[0075] Construct a cumulative contribution domain, estimate each dynamic growth factor in the dynamic growth factor loading matrix based on the core tensor difference value, obtain multiple dynamic growth factor estimates, and simultaneously accumulate the contribution rate of all dynamic growth factors in the cumulative contribution domain;

[0076] During the cumulative contribution process, only dynamic growth factors whose estimated value is greater than the preset estimated value are extracted and marked as key prominent factors. If the cumulative contribution rate reaches the preset cumulative contribution rate, the extraction process is stopped, and one or K key prominent factors are obtained.

[0077] A dynamic growth transition mapping space is constructed based on one or K key prominent factors. The lag matrix is ​​then projected onto this dynamic growth transition mapping space to establish an autoregressive model of body morphology.

[0078] The dynamic growth of the target pest at the current seasonal time is calculated by using an autoregressive model of physical morphology, and a physical morphology regression score matrix is ​​generated. Based on the physical morphology regression score matrix, the estimated value of the physical morphology of the target pest trapped by the sex pheromone trap at the current seasonal time is determined.

[0079] It should be noted that traditional body size prediction models are obtained by direct training and validation using historical data. Therefore, the numerical prediction of the body size of target pests can usually only be estimated probabilistically by referring to the apparent body size change patterns in historical morphological data. However, under the constraint of seasonal changes, the exploration performance based on body size change patterns is poor and may have a large prediction bias. This makes it impossible to explore the subtle body size change trends under the dynamic influence of seasonal time, thus making it difficult to achieve accurate estimation of the body size of target pests at the current trapping season nodes of the maintenance trapping and monitoring instrument. This leads to unreasonable control of the negative pressure application for fixed pest photography. To address this, this method first performs tensor structure expansion calculations based on the growth lag matrix to capture the dynamic characteristics of seasonal fluctuations in body morphology. This allows for a detailed analysis of the factors causing the lag in body morphology compared to standardized growth, and the exploration and extraction of potential factors that may perturb the seasonal growth of pest body size. This ensures that the interference and constraints of different seasonal growth factors on the body size of the currently trapped target pests are fully considered, thus making the body size estimation of target pests more accurate while adhering to the seasonal temporal change trend. Next, by calculating the core tensor difference between the first and second dynamic growth change core tensors, the interaction relationship between the known historical body morphology growth pattern and the stable growth pattern when the target pest's body morphology is perturbed by different dynamic growth factors can be captured. Based on this interaction relationship, the influence magnitude of each potential dynamic growth factor can be further quantified, i.e., the dynamic growth factor estimate. This allows the subsequent prediction of the body morphology of the trapped target pests to better mimic the virtual body shape contour of seasonal growth based on the perturbation weights of potential dynamic growth factors.

[0080] It should be noted that n-mode multiplication multiplies a matrix with a tensor along a certain dimension (mode), altering the mode representation in that dimension and thus reconstructing the original tensor structure to accurately extract the core tensor. Based on the magnitude of the dynamic growth factor estimates, the most critical and heavily weighted k potential perturbation factors are selected to characterize the seasonal temporal changes in body morphology. Therefore, this method only extracts key prominent factors whose dynamic growth factor estimates are greater than preset estimates as inputs for the autoregressive mapping of dynamic growth changes, significantly improving the accuracy of body morphology prediction. The number of k potential perturbation factors is determined by the cumulative contribution of these factors; therefore, the contribution rate of all dynamic growth factors is accumulated in the cumulative contribution domain, effectively reducing noise and redundant information in the data and improving the stability and generalization ability of the prediction model. By constructing a dynamic growth transition mapping space using these key prominent factors and projecting the lag matrix into this space, an autoregressive model of body morphology is established. Finally, a body morphology regression score matrix is ​​calculated using this autoregressive model. This matrix reflects the position and distribution of the body shape of the trapped target pests at the current seasonal time, and is further transformed into a predicted value of the body morphology of the target pests trapped by the sex pheromone trapping instrument at the current seasonal time. This method can accurately characterize and estimate the body morphology of target pests trapped at the current seasonal time under the premise of potential dynamic growth factor perturbations with seasonal changes. Compared with traditional prediction models, it pays more attention to exploring the trend of subtle body shape changes under the dynamic influence of seasonal time, providing a reliable pressure standard for accurate negative pressure fixation of target pests.

[0081] More specifically, the process of obtaining the dynamic transition tensor chain of the target pest's growth accompanied by seasonal fluctuations based on the growth lag matrix, unfolding the dynamic transition tensor chain based on the directional growth criteria that meet the research requirements at the current seasonal time, and singularly decomposing the influencing factors to obtain the dynamic growth factor loading matrix of the target pest's directional growth at each hierarchical dimension, specifically includes the following steps:

[0082] Based on the growth lag matrix, the lag tensor hierarchy structure of the seasonal transition dynamic characteristics of the body morphology is extracted, and the dynamic transition tensor chain of the target pest's growth accompanied by seasonal fluctuations is determined according to the lag tensor hierarchy structure.

[0083] The current season and time of the sex pheromone detection instrument are obtained. Based on the big data network, the directional growth criteria of the target pests in the current season and the timing of growth mutations that follow the directional growth criteria are obtained.

[0084] Based on the directional growth criterion, the target pest is located in different hierarchical dimensions at the current season. Simultaneously, based on the actual nodes of growth mutation, the target growth rank necessary for the target pest in each hierarchical dimension is set. The covariance expansion of the dynamic transition tensor chain is performed on each hierarchical dimension to obtain multiple tensor covariance expansion matrices.

[0085] A singular decomposition algorithm is introduced. By taking the factor singular value composed of the left singular vector of each tensor covariance expansion matrix, the dynamic growth factor loading matrix corresponding to the output of each tensor covariance expansion matrix when the target pest is growing in each hierarchical dimension is obtained.

[0086] It should be noted that, in the calculation of the tensor structure of seasonal fluctuations in body morphology and the search for and extraction of potential factors that may perturb the seasonal growth of pest body size, this method first extracts the lag tensor hierarchy structure of the dynamic characteristics of seasonal leaps in body morphology. The lag tensor hierarchy structure reflects the higher-order chain structure of lag vectors, matrices, etc., in which historical changes in body morphology lag behind the standardized body size as seasonal leaps occur. It is the overall linear framework of the lag trend formed by the influence of different historical potential factors. Since the prediction of the body morphology of the captured target pests is real-time, the dynamic change tensor chain needs to be expanded using directional growth criteria that meet the research requirements under the current seasonal conditions. These directional growth criteria guide the target pests towards standardized growth along certain regular paths while meeting research requirements. For example, if the research requirements specify certain identifying characteristics of the target pests, they need to develop according to the prescribed growth stages from the larval feeding stage to the adult degeneration stage. Therefore, there are hierarchical dimensions for different growth stages. These hierarchical dimensions serve as the basis for disassembling the dynamic transition tensor chain in matrix form. This allows us to uncover the internal structure of the dynamic transition tensor chain that causes the historical morphological growth stages to lag, thereby improving the accuracy of tracing the source of subsequent potential perturbation factors. Development following this directional growth principle involves a point in time when a growth stage transitions to the next, namely, the growth mutation timing node. This growth mutation timing node gives each growth stage a defined indicator, namely the necessary target growth rank for each hierarchical dimension. By using the singular decomposition algorithm to extract the factor singular values ​​composed of the forward target growth order sequence of the left singular vector, we can extract the most critical and important orthogonal components and directions under each growth stage's hierarchical dimension, thus obtaining the factor matrix for each hierarchical dimension. This method can analyze the linear trend of the time lag between historical body shape and standardized body shape by different historical potential factors, and extract potential dynamic growth factors that may have key perturbations. This allows for a full consideration of the interference and constraints of different seasonal growth factors on the current body shape of the target pests being trapped, thereby making the body shape estimation of the target pests more accurate while conforming to the seasonal time-series change trend.

[0087] More specifically, step S106, as follows: Figure 2 As shown, the specific steps include:

[0088] S202: Based on the predicted body morphology, retrieve the joint swinging force of each joint part under the influence of sex pheromones during the normal activity of the target pest in the big data network, define the joint swinging force as the target elimination variable, and at the same time obtain the predetermined negative pressure adjustment decision of the high-power fan.

[0089] S204: Construct a relaxed mainline model for negative pressure resistance to joint swinging force, considering only the target elimination variable and pre-setting an elimination estimation variable. Extract the predetermined negative pressure inhibition index of the joint swinging force of the target pest generated by the high-power fan for the body shape prediction through the predetermined negative pressure parameter adjustment decision. Define the predetermined negative pressure inhibition index as the original constraint limit.

[0090] S206: Solve the relaxation principal model based on the elimination of estimated variables to obtain the current target elimination variable solution of the relaxation principal. Substitute the current target elimination variable solution into the original constraint limit to construct a sub-line model for eliminating the target elimination variable with a predetermined negative pressure resistance index. Preset the allowable safe inhibition range of the target pest under negative pressure.

[0091] S208: If the sub-line model can completely eliminate the target elimination variable, then introduce the sub-line dual variable into the sub-line model to generate the optimal cut for eliminating the target elimination variable; if the sub-line model cannot completely eliminate the target elimination variable, then use the dual infeasible variable to eliminate the current target elimination variable solution one by one to generate a feasible cut.

[0092] S210: Repeat the above steps of solving the relaxation principal model and eliminating the sub-model, add the generated optimal cut and feasible cut to the relaxation principal model, and obtain the current negative pressure resistance value of the relaxation principal model and the current optimal elimination value of the sub-model in real time.

[0093] S212: If the relative error between the current negative pressure resistance value and the current optimal elimination value does not exceed the allowable safe suppression range, then stop adding optimal cuts and feasible cuts, and finally output the negative pressure resistance parameter. Based on the negative pressure resistance parameter, control the high-power fan to draw air from the insect receiving channel to resist the joint swinging force of the fixed target pest.

[0094] It should be noted that after determining the reliable morphological estimate of the target pest trapped by the pheromone trap, air can be further extracted from the insect-collecting channel to immobilize the target pest under negative pressure. However, existing negative pressure immobilization systems have poor control and computational performance, making it difficult to create a reasonable negative pressure environment based on the known pest size. This results in excessive negative pressure in some cases, causing varying degrees of damage and stress to the target pest, while insufficient negative pressure in others leads to unreliable immobilization, causing frequent movement of the target pest during the imaging process. To address this, this method constructs a relaxation principal model by using negative pressure to resist joint swaying force. This model includes a main variable (target elimination variable) and an estimated variable representing the optimal value of a sub-branch (elimination estimated variable), providing a positive iterative principal framework for the negative pressure environment to completely eliminate the joint swaying force of the target pest. In this process, the actual control of the negative pressure environment mainly relies on the predetermined negative pressure parameter adjustment decision of the high-power fan. Whether the target pest can be reliably fixed and with appropriate force depends on the predetermined negative pressure inhibition index generated by the high-power fan for the joint swinging force of the target pest based on the predicted body shape. This is a reasonable negative pressure fixing constraint premise for eliminating joint swinging force. Therefore, this method first solves the relaxation principal model based on eliminating estimated variables to obtain the current target elimination variable solution. This current target elimination variable solution is used to fix the variables in the sub-branch. Then, the current target elimination variable solution is substituted into the original constraint boundary to generate the sub-branch model. This further transforms the high-dimensional problem of negative pressure resistance to joint swinging force from a high-dimensional mixed space into a sub-problem of eliminating the target elimination variable with a continuous predetermined negative pressure resistance index that is easier to analyze. This allows the required negative pressure resistance amplitude to be solved while eliminating joint swinging force, effectively improving the accuracy of negative pressure resistance control and linkage control performance.

[0095] It should be noted that if the sub-line model can completely eliminate the target elimination variable, it indicates that the negative pressure resistance solution for eliminating joint swing force in the sub-line model is feasible and reasonable, satisfying the implicit constraints of the main line model. Therefore, it is necessary to introduce sub-line duality theory to generate the optimal cutting plane for eliminating the target elimination variable. If the sub-line model cannot completely eliminate the target elimination variable, it indicates that the negative pressure resistance solution for eliminating joint swing force in the sub-line model is extremely unreasonable and cannot reliably fix the target pest. Therefore, it is necessary to use dual infeasible variables to eliminate the current target elimination variable solution one by one to generate a feasible cutting plane, because this will lead to the infeasibility of the sub-line model, thereby effectively and progressively approximating the feasible region of the high-dimensional main line problem, avoiding repeated exploration of negative pressure control solutions in known infeasible or inefficient regions, and improving the rationality and accuracy of negative pressure control. Among them, if the relative error between the current negative pressure resistance value and the current optimal elimination value does not exceed the allowable safe suppression range, it indicates that the negative pressure resistance solution for eliminating the target elimination variable with the given negative pressure resistance index can keep the fixation of the target pest within a safe range, thus not causing harm to the target pest. This method enables the rational calculation and planning of negative pressure resistance control parameters for high-power fans, with the goal of eliminating the joint swaying force of target pests. This allows for the application of appropriate negative pressure based on the size of the target pests, avoiding insecure fixation, reducing the harm coefficient to the target pests, and improving the sample protection performance of negative pressure insect fixation photography.

[0096] More specifically, step S108 includes the following steps:

[0097] Obtain the preset resolution of the high-definition camera and the spatial model of the insect-catching channel flap. Based on the preset resolution, discretize and plan several voxel grids on the spatial model to construct the occupied grid area of ​​the insect-catching channel flap.

[0098] The supplementary lighting binary image information of the target pest falling into the neighborhood of the occupied grid is obtained by a high-definition camera. The 1 supplementary lighting voxel grid that is not occupied by the target pest in the supplementary lighting binary image information is defined as a free space point, and the 0 supplementary lighting voxel grid that is occupied by the target pest is defined as an obstacle point.

[0099] The Euclidean distance transformation algorithm is introduced to calculate and store the shortest Euclidean distance from each free space point to the nearest adjacent obstacle point, and the distance calculation of the obstacle point itself is automatically ignored and marked as 0, generating a positive unsigned distance field.

[0100] During computation and storage, the obstacle points and free space points in the supplementary lighting binary image information are reversed and swapped, and the distance field is recalculated to generate an inverse unsigned distance field. Based on the inverse unsigned distance field, it is determined whether each supplementary lighting voxel grid is located inside the obstacle point.

[0101] If the fill light voxel grid is located inside the obstacle point, then the fill light voxel grid is assigned a positive sign; if the fill light voxel grid is located outside the obstacle point, then the fill light voxel grid is assigned a negative sign, resulting in a positive-negative fill light voxel pattern.

[0102] Based on research needs, obtain the ideal negative pressure fixation posture that can capture the expected identification information of the target pest, and obtain the range of degrees of freedom of the target pest's corresponding joint parts when making the ideal negative pressure fixation posture based on big data.

[0103] A Bayesian voxel inference network was constructed using the Bayesian voxel update rule. Based on the activity degree of freedom interval, the occupancy of joint voxels of the target pest in the negative pressure fixed posture was calculated in the Bayesian voxel inference network to obtain the occupancy probability of each supplementary light voxel grid.

[0104] If the positive-negative fill light voxel pattern shows a negative sign in the fill light voxel grid where the occupancy probability is greater than the preset occupancy probability, then the high-power fan is instantly controlled to fix the target pest on the insect channel flap and to provide fill light for the high-definition industrial camera.

[0105] It should be noted that the purpose of negative pressure insect fixation photography is to obtain characteristic image data of pests for accurate pest species identification. However, because the target pests possess self-awareness, the characteristic information they require according to research needs may be obscured. For example, when photographing the legs of a target pest, the unfolding of its wing joints may cover or obscure the local contour of the leg joints, making it impossible for the high-definition industrial camera to capture a clear image of the legs. It is necessary to remove the obstruction when the wings are folded to capture an image of the leg joints. Therefore, the timing of negative pressure fixation and the reasonable capture of the high-definition industrial camera are crucial. To address this, this method constructs an occupation grid field of the insect-receiving channel flap and obtains supplementary lighting binary image information of the target pest when it falls into the neighborhood of the occupation grid field. This supplementary lighting binary image information is a preliminary real-time image before negative pressure fixation, which can reflect whether various parts of the target pest obscure the occupation grid field. Specifically, by introducing an Euclidean distance transformation algorithm to calculate and store the shortest Euclidean distance from each free space point to its nearest adjacent obstacle point, and automatically ignoring the distance calculation of the obstacle point itself and marking it as 0, a distance map containing only positive feedback is obtained, namely a positive unsigned distance field. This positive unsigned distance field indicates how far a free space point is from the target pest obstacle point; the larger the distance, the farther away from the obstacle point. During the calculation and storage, the obstacle points and free space points in the supplementary lighting binary map information are reversed and interchanged, and the distance field is recalculated to generate a reverse unsigned distance field. This is mainly to calculate the distance from points inside the obstacle to the nearest free space boundary, used to obtain the internal distance of the obstacle point. This reverse unsigned distance field is the internal distance seen from inside the obstacle.

[0106] It should be noted that the positive-negative supplementary lighting voxel pattern represents the Euclidean distance from each occupied grid to the nearest obstacle point. The sign indicates whether the occupied grid is "inside" or "outside" the obstacle point, which can express the current posture of the target pest on the insect-catching channel flap. Next, by constructing a Bayesian voxel inference network, the probability that each supplementary lighting voxel grid may be occupied when the target pest adopts a negative pressure fixation posture is calculated in the Bayesian voxel inference network based on the activity degree of freedom interval, thereby obtaining the grid occupancy pattern that can capture the expected identification information of the target pest. If the positive-negative supplementary lighting voxel pattern shows that the supplementary lighting voxel grid with an occupancy probability greater than the preset occupancy probability is negative, it means that the local part that can capture the expected identification information of the target pest will not be blocked. This is the best time to fix the target pest's posture with negative pressure and take pictures. Therefore, the high-power fan is instantly controlled to fix the target pest on the insect-catching channel flap with negative pressure and to supplement the lighting for high-definition industrial camera shooting. This method enables the analysis of instantaneous occlusion in the best posture for instantaneous negative pressure fixation of high-power fans and supplementary lighting shooting of high-definition industrial cameras, thereby improving the timing performance of negative pressure fixation and shooting, and improving the accuracy of negative pressure fixation and shooting quality.

[0107] The second aspect of this invention provides a negative pressure insect-detecting and photographing system for a sex pheromone detector, such as... Figure 3 As shown, the negative pressure insect-taking and imaging system includes a memory 31 and a processor 32. The memory 31 stores a negative pressure insect-taking and imaging method program for a pheromone detector. When the negative pressure insect-taking and imaging method program is executed by the processor 32, any of the steps of the negative pressure insect-taking and imaging method described above are implemented.

[0108] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A negative pressure insect identification and imaging method for a sex pheromone detector, characterized in that, Includes the following steps: S102: Obtain the historical physical morphology parameters of the target pests captured by the sex pheromone trapping instrument. Based on the transition distribution hash bucket of the standard physical morphology of the target pests under different seasonal conditions, perform time-series growth lag calculation on the historical physical morphology parameters to obtain a relatively standardized growth lag matrix of the historical physical morphology of the target pests. The time-series growth stagger calculation in step S102 specifically includes the following steps: Calculate the combined hash value of each historical body shape parameter in the transition hash map, find the transition hash bucket with the combined hash value attached, mark it as the target transition distribution hash bucket, and define the standard body shape reference value in the target transition distribution hash bucket as the candidate reference value. Calculate the adjacency distance between each candidate reference value and each historical morphological parameter, and interpolate the historical morphological morphology of the target pest on the specified time series chain to generate the growth lag matrix of seasonal transitions compared with the standard morphological morphology. S104: Based on the dynamic change tensor structure expansion decomposition of the growth lag matrix, obtain the dynamic growth factor load matrix of the target pest's directional growth at the current seasonal time. Construct an autoregressive model of body morphology by estimating and accumulating the contribution of the dynamic growth factor load matrix. Use the regression operation of the autoregressive model of body morphology to obtain the estimated value of the body morphology of the target pest trapped by the sex pheromone trap at the current seasonal time. S106: Construct a relaxed mainline model for the negative pressure resistance joint swing force, including a target elimination variable and an elimination estimation variable representing the optimal value of the sub-line branch; Based on the joint swing force that the target pest can exert when it is in the predicted body shape, determine the feasibility of eliminating the target through the mainline and sub-line negative pressure resistance and perform cutting plane statistics, output negative pressure resistance parameters, and control the fan to draw air from the insect receiving channel to fix the target pest based on the negative pressure resistance parameters; S108: By reversing the information of the supplementary light binary image captured when the target pest falls into the grid area of ​​the insect receiving channel flip plate, a positive-negative supplementary light voxel pattern is constructed. Combined with the ideal negative pressure fixation posture of the target pest and the occupation analysis of the positive-negative supplementary light voxel pattern, the negative pressure insect fixation and the supplementary light photography of the high-definition industrial camera are controlled in real time. Specifically, step S108 involves reversing the binary image information captured when the target pest falls into the grid area of ​​the insect-collecting channel flap to construct a positive-negative supplementary light voxel pattern, which includes the following steps: Obtain the preset resolution of the high-definition camera and the spatial model of the insect-catching channel flap. Based on the preset resolution, discretize and plan several voxel grids on the spatial model to construct the occupied grid area of ​​the insect-catching channel flap. The supplementary lighting binary image information of the target pest falling into the occupied grid area is obtained by a high-definition camera. The 1 supplementary lighting voxel grid that is not occupied by the target pest in the supplementary lighting binary image information is defined as a free space point, and the 0 supplementary lighting voxel grid that is occupied by the target pest is defined as an obstacle point. The Euclidean distance transformation algorithm is introduced to calculate and store the shortest Euclidean distance from each free space point to the nearest adjacent obstacle point, and the distance calculation of the obstacle point itself is automatically ignored and marked as 0, generating a positive unsigned distance field. During computation and storage, the obstacle points and free space points in the supplementary lighting binary image information are reversed and swapped, and the distance field is recalculated to generate an inverse unsigned distance field. Based on the inverse unsigned distance field, it is determined whether each supplementary lighting voxel grid is located inside the obstacle point. If the fill light voxel grid is located inside the obstacle point, a positive sign is assigned to the fill light voxel grid; if the fill light voxel grid is located outside the obstacle point, a negative sign is assigned to the fill light voxel grid, resulting in a positive-negative fill light voxel pattern.

2. The negative pressure insect-taking and photographing method of a sex pheromone detector according to claim 1, characterized in that, In step S102, the historical physical morphological parameters of the target pests captured by the sex pheromone trapping instrument are obtained. The calculation of the transition distribution hash bucket based on the standard physical morphology of the target pests under different seasonal conditions specifically includes the following steps: The research needs of the target pests and the trapping logs of the sex pheromone detector are obtained, and the preset sex pheromone release strategy of the sex pheromone detector for the research needs is obtained by identification. By extracting historical morphological parameters of the target pests when the sex pheromone release strategy was executed on a specified time chain by the sex pheromone trapping instrument, a knowledge graph of insect mating was obtained based on big data. Based on the insect mating knowledge graph, several historical physical morphology parameters are indexed and analyzed to output the baseline mating behavior characteristics of target pests with different physical morphologies at different seasonal change periods. Based on the baseline mating behavior characteristics, a family of transition hash functions is pre-defined to show that physical morphology produces significant differences under the dominance of seasonal change. Construct a transition hash map of the standard changes in the physical morphology of the target pest under different seasonal conditions, and obtain reference values ​​of the standard physical morphology of the target pest under different seasonal conditions through an insect mating knowledge graph. Based on the family of transition hash functions, each of the standard physical morphology reference values ​​is hashed into a signature symbol, resulting in several standard transition hash signatures and the transition hash value of each standard transition hash signature. The standard physical morphology reference values ​​corresponding to the standard transition hash signatures with the same transition hash value are placed in the same bucket, resulting in multiple transition distribution hash buckets.

3. The negative pressure insect-taking and photographing method of a sex pheromone detector according to claim 1, characterized in that, Step S104 specifically includes the following steps: The dynamic change tensor chain of the target pest's growth accompanied by seasonal fluctuations is obtained based on the growth lag matrix. The dynamic change tensor chain is expanded and the influencing factors are singularly decomposed based on the directional growth criteria that meet the research needs at the current seasonal time, so as to obtain the dynamic growth factor loading matrix of the target pest's directional growth at each hierarchical dimension. The core tensor of the target pest that tends to stabilize under the premise of being disturbed by different dynamic growth factors is defined as the first dynamic growth change core tensor. Multiply each of the dynamic growth factor loading matrices with the dynamic transition tensor chain n-mode to obtain the dynamic growth transition core tensor, which is defined as the second dynamic growth transition core tensor. Calculate the difference between the first dynamic growth transition core tensor and the first dynamic growth transition core tensor to obtain the core tensor difference value. Construct a cumulative contribution domain, estimate each dynamic growth factor in the dynamic growth factor loading matrix based on the core tensor difference value, obtain multiple dynamic growth factor estimates, and simultaneously accumulate the contribution rate of all dynamic growth factors in the cumulative contribution domain; During the cumulative contribution process, only dynamic growth factors whose estimated value is greater than the preset estimated value are extracted and marked as key prominent factors. If the cumulative contribution rate reaches the preset cumulative contribution rate, the extraction process is stopped, and one or K key prominent factors are obtained. A dynamic growth transition mapping space is constructed based on one or K key prominent factors. The growth lag matrix is ​​then projected onto this dynamic growth transition mapping space to establish an autoregressive model of body shape. The dynamic growth of the target pest at the current seasonal time is calculated by using an autoregressive model of physical morphology, and a physical morphology regression score matrix is ​​generated. Based on the physical morphology regression score matrix, the estimated value of the physical morphology of the target pest trapped by the sex pheromone trap at the current seasonal time is determined.

4. The negative pressure insect-detecting and photographing method of a sex pheromone detector according to claim 3, characterized in that, The process involves obtaining the dynamic transition tensor chain of the target pest's growth accompanied by seasonal fluctuations based on the growth lag matrix, unfolding the dynamic transition tensor chain based on the directional growth criteria that meet the research requirements at the current seasonal time, and singularly decomposing the influencing factors to obtain the dynamic growth factor loading matrix of the target pest's directional growth at each hierarchical dimension. Specifically, this includes the following steps: Based on the growth lag matrix, the lag tensor hierarchy structure of the seasonal transition dynamic characteristics of the body morphology is extracted, and the dynamic transition tensor chain of the target pest's growth accompanied by seasonal fluctuations is determined according to the lag tensor hierarchy structure. The current season and time of the sex pheromone detection instrument are obtained. Based on the big data network, the directional growth criteria of the target pests in the current season and the timing of growth mutations that follow the directional growth criteria are obtained. Based on the directional growth criterion, the target pest is located in different hierarchical dimensions at the current season. Simultaneously, based on the growth mutation timing node, the target growth rank necessary for the target pest in each hierarchical dimension is set. The covariance expansion of the dynamic transition tensor chain is performed on each hierarchical dimension to obtain multiple tensor covariance expansion matrices. A singular decomposition algorithm is introduced. By taking the factor singular value composed of the left singular vector of each tensor covariance expansion matrix, the dynamic growth factor loading matrix corresponding to the output of each tensor covariance expansion matrix when the target pest is growing in each hierarchical dimension is obtained.

5. The negative pressure insect-detecting and photographing method of a sex pheromone detector according to claim 1, characterized in that, Step S106 specifically includes the following steps: Based on the predicted body morphology, the joint swinging force of each joint part under the influence of sex pheromones during the normal activity of the target pest is retrieved from the big data network. The joint swinging force is defined as the target elimination variable, and the predetermined negative pressure adjustment decision of the fan is obtained at the same time. A relaxed mainline model for resisting joint swing force under negative pressure is constructed. Only one elimination estimation variable is preset, considering only the target elimination variable. The predetermined negative pressure inhibition index for the joint swing force generated by the target pest with the predicted body shape is extracted through the predetermined negative pressure parameter adjustment decision. The predetermined negative pressure inhibition index is defined as the original constraint limit. The relaxation principal model is solved by eliminating the estimated variables to obtain the current target elimination variable solution of the relaxation principal. The current target elimination variable solution is substituted into the original constraint limit to construct a sub-line model for eliminating the target elimination variable with a predetermined negative pressure resistance index. The allowable safe inhibition range of the target pest is preset under negative pressure. If the sub-line model can completely eliminate the target elimination variable, then the sub-line dual variable is introduced into the sub-line model to generate the optimal cut for eliminating the target elimination variable; if the sub-line model cannot completely eliminate the target elimination variable, then the dual infeasible variable is used to eliminate the current target elimination variable solution one by one to generate a feasible cut. Repeat the above steps of solving the relaxation principal model and eliminating the sub-model, add the generated optimal cut and feasible cut to the relaxation principal model, and obtain the current negative pressure resistance value of the relaxation principal model and the current optimal elimination value of the sub-model in real time. If the relative error between the current negative pressure resistance value and the current optimal elimination value does not exceed the allowable safe suppression range, then stop adding optimal cuts and feasible cuts, and finally output the negative pressure resistance parameter. Based on the negative pressure resistance parameter, control the fan to draw air from the insect receiving channel to resist the joint swinging force of the fixed target pest.

6. The negative pressure insect-detecting and photographing method of a sex pheromone detector according to claim 1, characterized in that, In step S108, the step of combining the ideal negative pressure fixation posture of the target pest with the positive-negative supplementary light voxel pattern occupancy analysis to instantaneously control the negative pressure pest fixation and the supplementary light imaging of the high-definition industrial camera specifically includes the following steps: Based on research needs, obtain the ideal negative pressure fixation posture that can capture the expected identification information of the target pest, and obtain the range of degrees of freedom of the target pest's corresponding joint parts when making the ideal negative pressure fixation posture based on big data. A Bayesian voxel inference network was constructed using the Bayesian voxel update rule. Based on the activity degree of freedom interval, the occupancy of joint voxels of the target pest in the negative pressure fixed posture was calculated in the Bayesian voxel inference network to obtain the occupancy probability of each supplementary light voxel grid. If the positive-negative fill light voxel pattern shows that the fill light voxel grid with an occupation probability greater than the preset occupation probability is negative, then the fan is instantly controlled to fix the target pest on the insect channel flap and to take a fill light picture with the high-definition industrial camera.

7. A negative pressure insect-detecting and photographing system for a sex pheromone detector, characterized in that, The negative pressure insect-taking and imaging system includes a memory and a processor. The memory stores a negative pressure insect-taking and imaging method program for a sex pheromone detector. When the negative pressure insect-taking and imaging method program is executed by the processor, the steps of the negative pressure insect-taking and imaging method as described in any one of claims 1-6 are implemented.

Images