Multi-task visual perception tomato maturity detection method

Abstract

The invention discloses a multi-task visual perception tomato maturity detection method which comprises the following steps: constructing an observation model according to position distribution information of tomato fruits; processing actual position information and scanning position information of the tomatoes by using a volume Kalman filter, and calculating a covariance matrix of tomato position coordinates and a covariance matrix of obstacle position coordinates; calculating a volume point according to the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates, and obtaining coordinate information of the tomato position after visual perception; performing state volume point propagation to obtain an estimated value of the tomato position; and fuzzy control is utilized to take the matching degree of the actual variance and the theoretical variance of the residual error of the lycopene content as an input quantity and a noise regulation factor in the process of measuring the tomato hardness as an output quantity, and the input quantity is taken as a final positioning result after visual fusion filtering. The tomato maturity can be effectively, intelligently and nondestructively detected, and the method has good robustness and superiority.

Description

Technical Field

[0001] The present invention relates to the technical field of food detection and visual detection, and in particular to a tomato maturity detection method based on multi-task visual perception. Background Art

[0002] Lycopene is widely found in tomatoes, tomato products, watermelons, grapefruits and other fruits. It is the main pigment in ripe tomatoes, a carotenoid found in plant foods, and one of the most common carotenoids. Lycopene has deep red needle-shaped crystals, which not only have strong antioxidant activity, anti-cancer and anti-cancer effects, but also have the advantages of being soluble in chloroform, benzene and oils but not in water. The quality of tomato fruit is directly related to the antioxidant activity of lycopene. Therefore, a fast, accurate and effective method for detecting tomato maturity can ensure that lycopene reaches excellent quality and meets the vitamins and inorganic salts required by the human body.

[0003] After searching relevant patents and literature at home and abroad, for example, the paper "Tomato maturity detection method in natural environment based on improved YOLOv5s" combines the distribution characteristics of the tomato growth environment, introduces the MobileNetv3 network and the ECANet attention mechanism, and improves the YOLOv5s target detection algorithm. This method has better recognition effect, good robustness and real-time performance, and can meet the needs of accurate real-time recognition of tomatoes of different maturity. In addition, the paper "Tomato maturity detection based on convolutional neural network and transfer learning" optimizes the hyperparameters of three convolutional neural network models, MobilenetV2, InceptionV3 and VGG19, based on transfer learning, and establishes a detection model based on tomato maturity, solving the problems of complex calculation and low precision of existing tomato maturity detection models. For example, the invention patent "Cherry tomato maturity recognition method and device based on quantitative nuclear magnetic resonance images" with patent application number: CN202410518621.2 is based on the chaotic particle swarm inversion algorithm, and performs single index and triple index inversion on the echo data of each structure, presenting the overall and three-component moisture status of each structure, thereby obtaining a very accurate maturity judgment result.

[0004] Although the above method can identify the maturity of tomatoes, it has high requirements for model accuracy. Only calculating a single mathematical equation for the maturity of tomatoes can easily lead to poor control effects. In addition, in large-scale planting and industrial applications, tomato robots are needed to identify and automatically pick ripe tomatoes. To achieve automated picking of tomato fruits, it is necessary to quickly extract a large amount of coordinate position information of tomatoes at different maturity stages. However, when the tomato lidar visual information faces complex and changeable environments and other interference factors, the traditional algorithm that performs multiple single tasks in the order of visual recognition is limited by the comprehensive modeling of sensor information and computing power, and it is difficult to meet the recognition needs of tomato robots. Summary of the invention

[0005] The present invention aims to solve one of the technical problems existing in the related art at least to a certain extent.

[0006] One object of the present invention is to provide a tomato maturity detection method based on multi-task visual perception, which adopts fuzzy controlled volumetric Kalman for tomato maturity detection. The fusion positioning error has lower requirements compared with laser and vision, and can track the maturity target value with lower error, and has good robustness and superiority.

[0007] Another object of the present invention is to provide a tomato maturity detection system with multi-task visual perception.

[0008] In order to achieve the above-mentioned object, the present invention provides a tomato maturity detection method based on multi-task visual perception, comprising:

[0009] An observation model is constructed based on the location distribution information of mature and intact tomato fruits obtained by the detection instrument;

[0010] Using the volumetric Kalman filter to process the actual position information and the scanned position information of the tomato, the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates are calculated;

[0011] According to the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates, the volume point is calculated to obtain the coordinate information of the tomato position after visual perception; then the state volume point propagation is performed to obtain the estimated value of the tomato position;

[0012] Fuzzy control is used to take the matching degree between the residual actual variance and theoretical variance of lycopene content as input, and the noise adjustment factor in the process of measuring tomato hardness as output, which is the final positioning result after visual fusion filtering.

[0013] A further preferred technical solution of the present invention is that the observation model is constructed based on the position distribution information of the ripe and intact tomato fruits obtained by laser radar scanning; specifically:

[0014] Corresponding to the position information of the target detection of a ripe and intact tomato fruit detected by the detection instrument at a certain moment, the constructed observation model is expressed as:

[0015]

[0016] In formula (1), s k ,q k represents the coordinates of the kth tomato. represents the scanning position of the detection instrument at time t; α k ,β kIndicates the transformation relationship between the α-axis and the β-axis of the detection instrument relative to the tomato position; l k , represents the distance and direction angle between the detection instrument and the tomato; θ k ,Q k Indicates the tomato distance and azimuth interference information;

[0017] By obtaining the tomato distance and direction angle of the detection instrument, the position information of the tomato within time t is obtained.

[0018] Preferably, the method uses a volumetric Kalman filter to process the actual position information and the scanned position information of the tomato, and calculates the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates; specifically:

[0019] Through the cubature Kalman filter, the position information of the tomato and the scanning position of the detection instrument are used as the state equation and the measurement equation respectively, and the tomato and obstacle information are crossed and expressed in the form of Gaussian probability density function product integral:

[0020]

[0021]

[0022] In formula (2) and formula (3), is the kth tomato position coordinate measured by the detection instrument, is the kth tomato position coordinate collected on site, P k is the kth tomato position coordinate covariance matrix measured by the detection instrument, R k is the covariance matrix of the tomato position coordinates collected on site, E[·] is the state estimate, is the kth measurement data, x A is the coordinate origin of the kth tomato position measured by the detection instrument, y A It is the origin of the kth tomato position coordinates collected on site.

[0023] Preferably, the volume points are calculated according to the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates to obtain the coordinate information of the tomato position after visual perception; specifically:

[0024] The covariance matrix of the kth tomato position coordinates and the covariance matrix of the obstacle position coordinates measured by the detection instrument are used to calculate the volume points respectively to obtain the X-axis coordinate system information after visual perception. and Y-axis measurement coordinate system information It is expressed as:

[0025]

[0026]

[0027] In formula (4) and formula (5), ξ K is the position measurement error, Yes K The transposed matrix, S k represents the coordinate matrix of the kth tomato’s location, YesS k The transposed matrix of is the correction term of the covariance matrix of the kth tomato position coordinates measured by the detection instrument, is the obstacle position coordinate covariance matrix correction term, YesS k The inverse matrix of .

[0028] Preferably, the state volume point propagation is performed to obtain an estimated value of the tomato position; the specific method is:

[0029] The visually perceived X-axis coordinate system information and the Y-axis measurement coordinate system information are propagated into state volume points, respectively, and are expressed as:

[0030]

[0031]

[0032] In equations (6) and (7), λ(·) represents the volume point propagation model of the visual volume point X-axis coordinate system, Represents the information volume point propagation model of the visual volume point Y-axis coordinate system; and They represent the estimated value of the X-axis coordinate system information and the estimated value of the Y-axis measurement coordinate system information respectively; H k is the Kalman gain matrix, for The previous moment estimate, r Y It is the jth measurement data of the detection instrument, N and j are measurement constants, and k is the total number of measurements.

[0033] Preferably, the fuzzy control uses the matching degree of the residual actual variance and theoretical variance of the lycopene content as input, and the noise adjustment factor in the process of measuring tomato hardness as output, as the final positioning result after visual fusion filtering, and the specific method is:

[0034] The fuzzy control takes the matching degree c between the residual actual variance and theoretical variance of lycopene content as input and the noise adjustment factor a in the process of measuring tomato hardness as output.

[0035] The fuzzy control matching degree of the actual variance and theoretical variance of the volume point propagation model of the X-axis coordinate system and the Y-axis measurement coordinate system is expressed as:

[0036]

[0037] In formula (8), when χ deviates from the estimated value of the X-axis coordinate system information and the estimated value of the Y-axis measurement coordinate system information, the noise covariance matrix Q is adjusted by adjusting the interference factor a. k , change the residual matrix P g The size of is taken as the final positioning result after visual fusion filtering.

[0038] Another aspect of the present invention provides a multi-task visual perception tomato maturity detection system, comprising:

[0039] A data acquisition module is used to construct an observation model based on the position distribution information of mature and intact tomato fruits obtained by the detection instrument;

[0040] A state information extraction module, used for processing the actual position information and the scanned position information of the tomato by using a volumetric Kalman filter, and calculating the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates;

[0041] The visual perception and state propagation module is used to calculate the volume points according to the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates to obtain the coordinate information of the tomato position after visual perception; and is used to perform state volume point propagation to obtain the estimated value of the tomato position;

[0042] The fuzzy control module is used to use fuzzy control to take the matching degree of the residual actual variance and theoretical variance of the lycopene content as input and the noise adjustment factor in the process of measuring tomato hardness as output as the final positioning result after visual fusion filtering.

[0043] Another aspect of the present invention provides a non-transitory computer-readable storage medium having computer instructions stored thereon, the computer instructions enabling a computer to execute the above-mentioned multi-task visual perception method for detecting tomato maturity.

[0044] Yet another aspect of the present invention provides an electronic device, comprising: a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus, and the processor calls logic instructions in the memory to execute the above-mentioned multi-task visual perception tomato maturity detection method.

[0045] Yet another aspect of the present invention provides a computer program product, comprising a computer program stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer executes the above-mentioned multi-task visual perception tomato maturity detection method.

[0046] Beneficial effects: (1) The present invention adopts fuzzy controlled volumetric Kalman for tomato maturity detection. The fusion positioning error is lower than that of laser and vision, and can track the maturity target value with lower error. It has good robustness and superiority. Combined with the example simulation results, it is shown that the present invention can effectively, intelligently and non-destructively detect the maturity of tomatoes.

[0047] (2) Compared with the tomato maturity detection method based on convolutional neural network and transfer learning, the present invention takes into account limiting conditions such as the matching degree of noise adjustment factor and variance, and solves the problems that the prediction process is affected by many factors, parameter sensitivities interfere with each other, and it is difficult to establish an accurate optimization model.

[0048] (3) Compared with the tomato maturity detection method based on support vector machine, the present invention can obtain higher detection accuracy at a low loss value, reduce the speed of parameter optimization, reduce the coupling of serious mutual correlation between variables, effectively improve the search efficiency, and enable large-scale sample parameters to converge within a limited time, which can provide a reference for tomato maturity detection research. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a flow chart of the tomato maturity detection method based on multi-task visual perception of the present invention;

[0050] Figure 2 This is a result diagram of tomato maturity prediction using the method of the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments, and they should not be understood as limitations on the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In the description of the present invention, it should be understood that the terms used are only for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0052] Combine the following Figure 1-Figure 2 The present invention describes a tomato maturity detection method and system based on multi-task visual perception.

[0053] Example 1: This example provides a tomato maturity detection method based on multi-task visual perception. In view of the existing technology, when tomato lidar visual information faces complex and changeable environment and other interference factors, in order to realize the automatic picking of tomato fruits, it is necessary to quickly extract the coordinate position information of a large number of tomatoes at different maturity stages. However, the traditional visual recognition algorithm that sequentially performs multiple single tasks is limited by the comprehensive modeling of sensor information and computing power, and it is difficult to meet the tomato robot recognition needs.

[0054] The method of this embodiment, such as Figure 1 As shown, including:

[0055] An observation model is constructed based on the location distribution information of mature and intact tomato fruits obtained by the detection instrument;

[0056] Using the volumetric Kalman filter to process the actual position information and the scanned position information of the tomato, the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates are calculated;

[0057] According to the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates, the volume point is calculated to obtain the coordinate information of the tomato position after visual perception; then the state volume point propagation is performed to obtain the estimated value of the tomato position;

[0058] Fuzzy control is used to take the matching degree between the residual actual variance and theoretical variance of lycopene content as input, and the noise adjustment factor in the process of measuring tomato hardness as output, which is the final positioning result after visual fusion filtering.

[0059] Specifically, the multi-task visual perception tomato maturity detection method comprises the following steps:

[0060] S1. Construct an observation model based on the location distribution information of ripe and intact tomato fruits obtained by LiDAR scanning;

[0061] Corresponding to the position information of the target detection of a ripe and intact tomato fruit detected by the detection instrument at a certain moment, the constructed observation model is expressed as:

[0062]

[0063] In formula (1), s k ,q k represents the coordinates of the kth tomato. represents the scanning position of the detection instrument at time t; α k ,β k Indicates the transformation relationship between the α-axis and the β-axis of the detection instrument relative to the tomato position; l k , represents the distance and direction angle between the detection instrument and the tomato; θ k,Q k Indicates the tomato distance and azimuth interference information;

[0064] By obtaining the tomato distance and direction angle of the detection instrument, the position information of the tomato within time t is obtained.

[0065] S2, using a volumetric Kalman filter to process the actual position information and the scanned position information of the tomato, and calculating the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates;

[0066] Through the cubature Kalman filter, the position information of the tomato and the scanning position of the detection instrument are used as the state equation and the measurement equation respectively, and the tomato and obstacle information are crossed and expressed in the form of Gaussian probability density function product integral:

[0067]

[0068]

[0069] In formula (2) and formula (3), is the kth tomato position coordinate measured by the detection instrument, is the kth tomato position coordinate collected on site, P k is the kth tomato position coordinate covariance matrix measured by the detection instrument, R k is the covariance matrix of the tomato position coordinates collected on site, E[·] is the state estimate, is the kth measurement data, x A is the coordinate origin of the kth tomato position measured by the detection instrument, y A It is the origin of the kth tomato position coordinates collected on site.

[0070] S3, calculating the volume points according to the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates, and obtaining the coordinate information of the tomato position after visual perception;

[0071] The covariance matrix of the kth tomato position coordinates and the covariance matrix of the obstacle position coordinates measured by the detection instrument are used to calculate the volume points respectively to obtain the X-axis coordinate system information after visual perception. and Y-axis measurement coordinate system information It is expressed as:

[0072]

[0073]

[0074] In formula (4) and formula (5), ξ K is the position measurement error, Yes KThe transposed matrix, S k represents the coordinate matrix of the kth tomato’s location, YesS k The transposed matrix of is the correction term of the covariance matrix of the kth tomato position coordinates measured by the detection instrument, is the obstacle position coordinate covariance matrix correction term, YesS k The inverse matrix of .

[0075] S4, perform state volume point propagation to obtain an estimated value of the tomato position;

[0076] The visually perceived X-axis coordinate system information and the Y-axis measurement coordinate system information are propagated into state volume points, respectively, and are expressed as:

[0077]

[0078]

[0079] In equations (6) and (7), λ(·) represents the volume point propagation model of the visual volume point X-axis coordinate system, Represents the information volume point propagation model of the visual volume point Y-axis coordinate system; and They represent the estimated value of the X-axis coordinate system information and the estimated value of the Y-axis measurement coordinate system information respectively; H k is the Kalman gain matrix, for The previous moment estimate, r Y It is the jth measurement data of the detection instrument, N and j are measurement constants, and k is the total number of measurements.

[0080] S5, using fuzzy control to take the matching degree of the residual actual variance and theoretical variance of the lycopene content as input, and the noise adjustment factor in the process of measuring tomato hardness as output, as the final positioning result after visual fusion filtering;

[0081] The fuzzy control takes the matching degree c between the residual actual variance and theoretical variance of lycopene content as input and the noise adjustment factor a in the process of measuring tomato hardness as output.

[0082] The fuzzy control matching degree of the actual variance and theoretical variance of the volume point propagation model of the X-axis coordinate system and the Y-axis measurement coordinate system is expressed as:

[0083]

[0084] In formula (8), when χ deviates from the estimated value of the X-axis coordinate system information and the estimated value of the Y-axis measurement coordinate system information, the noise covariance matrix Q is adjusted by adjusting the interference factor a.k , change the residual matrix P g The size of is taken as the final positioning result after visual fusion filtering.

[0085] At this time, the laser instrument's observation of tomato maturity is convergent, and the prediction results are shown in Figure 2 , the error between the measured value and the predicted value is small.

[0086] Embodiment 2: This embodiment provides a multi-task visual perception tomato maturity detection system, including:

[0087] A data acquisition module is used to construct an observation model based on the position distribution information of mature and intact tomato fruits obtained by the detection instrument;

[0088] A state information extraction module, used for processing the actual position information and the scanned position information of the tomato by using a volumetric Kalman filter, and calculating the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates;

[0089] The visual perception and state propagation module is used to calculate the volume points according to the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates to obtain the coordinate information of the tomato position after visual perception; and is used to perform state volume point propagation to obtain the estimated value of the tomato position;

[0090] The fuzzy control module is used to use fuzzy control to take the matching degree of the residual actual variance and theoretical variance of the lycopene content as input and the noise adjustment factor in the process of measuring tomato hardness as output as the final positioning result after visual fusion filtering.

[0091] Embodiment 3: This embodiment provides a non-transitory computer-readable storage medium having computer instructions stored thereon, the computer instructions causing a computer to execute a multi-task visual perception method for detecting tomato maturity, the method comprising the following steps:

[0092] An observation model is constructed based on the location distribution information of mature and intact tomato fruits obtained by the detection instrument;

[0093] Using the volumetric Kalman filter to process the actual position information and the scanned position information of the tomato, the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates are calculated;

[0094] According to the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates, the volume point is calculated to obtain the coordinate information of the tomato position after visual perception; then the state volume point propagation is performed to obtain the estimated value of the tomato position;

[0095] Fuzzy control is used to take the matching degree between the residual actual variance and theoretical variance of lycopene content as input, and the noise adjustment factor in the process of measuring tomato hardness as output, which is the final positioning result after visual fusion filtering.

[0096] Embodiment 4: This embodiment provides an electronic device, which may include: a processor, a communications interface, a memory, and a communications bus, wherein the processor, the communications interface, and the memory communicate with each other via the communications bus. The processor may call logic instructions in the memory to execute a multi-task visual perception method for detecting tomato maturity, which includes the following steps:

[0097] An observation model is constructed based on the location distribution information of mature and intact tomato fruits obtained by the detection instrument;

[0098] Using the volumetric Kalman filter to process the actual position information and the scanned position information of the tomato, the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates are calculated;

[0099] According to the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates, the volume point is calculated to obtain the coordinate information of the tomato position after visual perception; then the state volume point propagation is performed to obtain the estimated value of the tomato position;

[0100] Fuzzy control is used to take the matching degree between the residual actual variance and theoretical variance of lycopene content as input, and the noise adjustment factor in the process of measuring tomato hardness as output, which is the final positioning result after visual fusion filtering.

[0101] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0102] Embodiment 5: This embodiment provides a computer program product, which includes a computer program. The computer program can be stored in a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can perform a multi-task visual perception tomato maturity detection method, which includes the following steps:

[0103] An observation model is constructed based on the location distribution information of mature and intact tomato fruits obtained by the detection instrument;

[0104] Using the volumetric Kalman filter to process the actual position information and the scanned position information of the tomato, the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates are calculated;

[0105] According to the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates, the volume point is calculated to obtain the coordinate information of the tomato position after visual perception; then the state volume point propagation is performed to obtain the estimated value of the tomato position;

[0106] Fuzzy control is used to take the matching degree between the residual actual variance and theoretical variance of lycopene content as input, and the noise adjustment factor in the process of measuring tomato hardness as output, which is the final positioning result after visual fusion filtering.

[0107] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0108] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-task visual perception method for detecting tomato maturity, characterized in that: include: An observation model is constructed based on the location distribution information of mature and intact tomato fruits obtained by the detection instrument; Using the volumetric Kalman filter to process the actual position information and the scanned position information of the tomato, the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates are calculated; According to the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates, the volume points are calculated to obtain the coordinate information of the tomato position after visual perception; Then perform state volume point propagation to obtain the estimated value of the tomato position; Fuzzy control is used to take the matching degree between the residual actual variance and theoretical variance of lycopene content as input, and the noise adjustment factor in the process of measuring tomato hardness as output, which is the final positioning result after visual fusion filtering.

2. The multi-task visual perception tomato maturity detection method according to claim 1, characterized in that: The observation model is constructed based on the position distribution information of the ripe and intact tomato fruits obtained by the laser radar scanning; specifically: Corresponding to the position information of the target detection of a ripe and intact tomato fruit detected by the detection instrument at a certain moment, the constructed observation model is expressed as: In formula (1), s k ,q k represents the coordinates of the kth tomato. represents the scanning position of the detection instrument at time t; α k ,β k Indicates the transformation relationship between the α-axis and the β-axis of the detection instrument relative to the tomato position; l k , represents the distance and direction angle between the detection instrument and the tomato; θ k ,Q k Indicates the tomato distance and azimuth interference information; By obtaining the tomato distance and direction angle of the detection instrument, the position information of the tomato within time t is obtained.

3. The multi-task visual perception tomato maturity detection method according to claim 2, characterized in that: The method of using a volumetric Kalman filter to process the actual position information and the scanned position information of the tomato to calculate the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates is as follows: Through the cubature Kalman filter, the position information of the tomato and the scanning position of the detection instrument are used as the state equation and the measurement equation respectively, and the tomato and obstacle information are crossed and expressed in the form of Gaussian probability density function product integral: In formula (2) and formula (3), is the kth tomato position coordinate measured by the detection instrument, is the kth tomato position coordinate collected on site, P k is the kth tomato position coordinate covariance matrix measured by the detection instrument, R k is the covariance matrix of the tomato position coordinates collected on site, E[·] is the state estimate, is the kth measurement data, x A is the coordinate origin of the kth tomato position measured by the detection instrument, y A It is the origin of the kth tomato position coordinates collected on site.

4. The method for detecting tomato maturity based on multi-task visual perception according to claim 3, characterized in that: The volume points are calculated according to the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates to obtain the coordinate information of the tomato position after visual perception; specifically: The covariance matrix of the kth tomato position coordinates and the covariance matrix of the obstacle position coordinates measured by the detection instrument are used to calculate the volume points respectively to obtain the X-axis coordinate system information after visual perception. and Y-axis measurement coordinate system information It is expressed as: In formula (4) and formula (5), ξ K is the position measurement error, Yes K The transposed matrix, S k represents the coordinate matrix of the kth tomato’s location, YesS k The transposed matrix of is the correction term of the covariance matrix of the kth tomato position coordinates measured by the detection instrument, is the obstacle position coordinate covariance matrix correction term, YesS k The inverse matrix of .

5. The multi-task visual perception tomato maturity detection method according to claim 4, characterized in that: The state volume point propagation is performed to obtain the estimated value of the tomato position; the specific method is: The visually perceived X-axis coordinate system information and the Y-axis measurement coordinate system information are propagated into state volume points, respectively, and are expressed as: In equations (6) and (7), λ(·) represents the volume point propagation model of the visual volume point X-axis coordinate system, Represents the information volume point propagation model of the visual volume point Y-axis coordinate system; and They represent the estimated value of the X-axis coordinate system information and the estimated value of the Y-axis measurement coordinate system information respectively; H k is the Kalman gain matrix, for The previous moment estimate, r Y It is the jth measurement data of the detection instrument, N and j are measurement constants, and k is the total number of measurements.

6. The method for detecting tomato maturity based on multi-task visual perception according to claim 5, characterized in that: The fuzzy control method uses the matching degree of the residual actual variance and theoretical variance of the lycopene content as input, and the noise adjustment factor in the process of measuring tomato hardness as output, as the final positioning result after visual fusion filtering. The specific method is as follows: The fuzzy control takes the matching degree c between the residual actual variance and theoretical variance of lycopene content as input and the noise adjustment factor a in the process of measuring tomato hardness as output. The fuzzy control matching degree of the actual variance and theoretical variance of the volume point propagation model of the X-axis coordinate system and the Y-axis measurement coordinate system is expressed as: In formula (8), when χ deviates from the estimated value of the X-axis coordinate system information and the estimated value of the Y-axis measurement coordinate system information, the noise covariance matrix Q is adjusted by adjusting the interference factor a. k , change the residual matrix P g The size of is taken as the final positioning result after visual fusion filtering.

7. A multi-task visual perception tomato maturity detection system, characterized in that: include: A data acquisition module is used to construct an observation model based on the position distribution information of mature and intact tomato fruits obtained by the detection instrument; A state information extraction module, used to process the actual position information and the scanned position information of the tomato using a volumetric Kalman filter, and calculate the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates; The visual perception and state propagation module is used to calculate the volume points according to the covariance matrix of the tomato position coordinates and the covariance matrix of the obstacle position coordinates, and obtain the coordinate information of the tomato position after visual perception; And used to perform state volume point propagation to obtain the estimated value of the tomato position; The fuzzy control module is used to use fuzzy control to take the matching degree of the residual actual variance and theoretical variance of the lycopene content as input and the noise adjustment factor in the process of measuring tomato hardness as output as the final positioning result after visual fusion filtering.

8. A non-transitory computer-readable storage medium having computer instructions stored thereon, the computer instructions causing a computer to execute the multi-task visual perception tomato maturity detection method according to any one of claims 1 to 6.

9. An electronic device, comprising: A processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus, and the processor calls the logic instructions in the memory to execute the multi-task visual perception tomato maturity detection method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program stored on a non-transitory computer-readable storage medium, wherein when the computer program is executed by a processor, the computer executes the multi-task visual perception tomato maturity detection method according to any one of claims 1 to 6.

Images