Golden chrysanthemum picking point positioning method, device and equipment and medium

Abstract

The invention relates to a golden emperor chrysanthemum picking point positioning method, device and equipment and a medium, and relates to the field of image processing, and the method comprises the steps: determining the outer contour center and the recess center of a golden emperor chrysanthemum to be picked; calculating and determining offset vector coordinates of the center of the outer contour pointing to the center of the recess, and calculating and determining a growth orientation angle of the golden chrysanthemum to be picked in the plane of the golden chrysanthemum image by adopting an arc tangent function according to the offset vector coordinates; calling a preset calibration model to determine the growth inclination angle of the golden chrysanthemum according to the ratio of the model length of the offset vector to the outer contour radius of the golden chrysanthemum to be picked; and according to the growth orientation angle and the growth inclination angle, constructing a growth attitude space vector of the to-be-picked golden chrysanthemum, and according to the growth attitude space vector and the three-dimensional space coordinate of the outer contour center, determining a target picking point of the to-be-picked golden chrysanthemum. The method prevents the stamen from being torn and falling off, and ensures that the cutting position is far away from the stamen and close to the base of the pedicel.

Description

Technical Field

[0001] This application relates to the field of image processing, and in particular to a method for locating the picking point of Chrysanthemum morifolium, a corresponding device, electronic equipment, and a computer-readable storage medium. Background Technology

[0002] As a high-value crop that is both medicinal and edible, the integrity of the golden chrysanthemum flowers is directly linked to the quality of the product, and the demand for automated harvesting technology is becoming increasingly urgent in large-scale planting scenarios.

[0003] Traditional manual harvesting is inefficient, labor-intensive, and its labor costs are rising year by year, making it difficult to meet the needs of industrial upgrading. Chrysanthemum harvesting mechanisms (such as cable-driven parallel robots) have become the core direction of industry development. However, existing chrysanthemum harvesting mechanisms still face many technical bottlenecks in practical applications, resulting in a high rate of harvesting damage, which seriously restricts product quality and the efficiency of large-scale production.

[0004] Cable-driven parallel robots (CDPR) have become the preferred equipment for automated harvesting of golden chrysanthemums due to their unique advantages such as easy construction, large workspace, fast movement speed, low inertia, and high mass-to-weight ratio. However, when operating in outdoor chrysanthemum gardens, they are easily affected by factors such as wind, contact forces, and changes in their own center of gravity, which can cause vibrations and place higher demands on the accuracy of harvesting positioning.

[0005] Currently, existing cable-driven parallel robots for harvesting Chrysanthemum morifolium have the following main technical shortcomings:

[0006] Firstly, the robotic arm in the cable-driven parallel robot directly grips the fragile flower stamen, causing the stamen to tear or fall off, or the cutting position deviates from the optimal area of ​​the flower stem. If it is too close to the stamen, it will damage the stamen; if it is too far away, it will leave too long a flower stem, affecting the quality.

[0007] Secondly, when multiple adjacent flowers or leaves are obscured in an image, rough positioning may misjudge the "stamens and leaves" or "outlines of adjacent flowers" as a single target, causing the robotic arm to touch multiple flowers at the same time when grasping, resulting in damage to the stamens due to compression.

[0008] Third, the growth posture of Chrysanthemum in the field is random, with a complex state of planar orientation deflection and three-dimensional spatial tilt. Existing flexible cable-driven parallel robots have not quantified the flower growth posture parameters. The robotic arm can only perform clamping or cutting actions at a fixed angle. This rigid operation mode leads to uneven force on the flowers. Especially when the stamens are deviated or the flowers are tilted, it is easy to cause problems such as stamen breakage and petal tearing, which makes it difficult to meet the core requirements of high-quality picking.

[0009] Fourth, the current chrysanthemum picking positioning systems of existing institutions can only roughly identify the position of flowers. They do not fully combine the morphological characteristics and growth posture differences of Chrysanthemum morifolium flowers, and cannot provide accurate positioning and posture reference for Chrysanthemum morifolium picking. They are also difficult to adapt to the high-precision control requirements of cable-driven parallel robots for Chrysanthemum morifolium picking, which further aggravates the problem of picking damage.

[0010] In summary, existing flexible cable-driven parallel robots suffer from problems such as the robotic arm directly gripping the fragile flower stamens, causing them to tear or fall off, and the robotic arm being able to perform gripping or cutting actions only at a fixed angle. This rigid operating mode leads to uneven force on the flower, especially when the stamens shift or the flower tilts, which can easily cause problems such as stamen breakage and petal tearing. The applicant has made corresponding explorations to solve this problem. Summary of the Invention

[0011] The purpose of this application is to solve the above-mentioned problems by providing a method for locating the picking point of Chrysanthemum morifolium, a corresponding device, electronic equipment and computer-readable storage medium.

[0012] To achieve the various objectives of this application, the following technical solution is adopted:

[0013] A method for locating the picking point of Chrysanthemum morifolium, proposed to meet one of the purposes of this application, includes:

[0014] Acquire images of Chrysanthemum morifolium containing one or more Chrysanthemum morifolium images to be picked, input the Chrysanthemum morifolium images into an image segmentation model that has been trained to convergence, and extract the first segmentation image corresponding to the outer edge region of the flower of the Chrysanthemum morifolium to be picked and the second segmentation image corresponding to the concave region of the stamen from the Chrysanthemum morifolium images.

[0015] The first center of the first minimum circumscribed circle surrounding all first pixels in the outer edge region of the flower is calculated based on the first segmentation image to determine the outer contour center of the chrysanthemum to be picked; the second center of the second minimum circumscribed circle surrounding all second pixels in the stamen concave region is calculated based on the second segmentation image to determine the concave center of the chrysanthemum to be picked.

[0016] The offset vector coordinates pointing from the center of the outer contour to the center of the depression are calculated and determined, and the arctangent function is used to calculate the growth orientation angle of the chrysanthemum to be picked in the chrysanthemum image plane based on the offset vector coordinates.

[0017] The growth tilt angle of the golden chrysanthemum is determined by calling a preset calibration model based on the ratio between the magnitude of the offset vector and the outer contour radius of the golden chrysanthemum to be harvested.

[0018] Based on the growth orientation angle and the growth tilt angle, a growth posture space vector of the chrysanthemum to be harvested is constructed. Based on the growth posture space vector and the three-dimensional spatial coordinates of the outer contour center, the target harvesting point of the chrysanthemum to be harvested is determined to complete the harvesting point positioning of the chrysanthemum.

[0019] Optionally, the step of inputting the image of the golden chrysanthemum into an image segmentation model that has been trained to convergence, and extracting the first segmentation image corresponding to the outer edge region of the flower of the golden chrysanthemum to be picked and the second segmentation image corresponding to the concave region of the stamen from the image of the golden chrysanthemum includes:

[0020] The image segmentation model, which has been trained to convergence, performs multi-level encoding to downsample the original image of the Chrysanthemum morifolium and produces intermediate feature information corresponding to each scale. The intermediate feature information represents the contour features of the outer edge region of the flower and the contour features of the concave region of the stamen in the Chrysanthemum morifolium image.

[0021] The image segmentation model, which has been trained to convergence, performs multi-level decoding. Based on the intermediate feature information at the smallest scale, it performs upsampling operation and successively jumps and fuses the intermediate feature information produced by the same level encoding as a reference to decode and generate higher-scale image feature information. The image feature information is used to represent the contour features of the outer edge region of the flower and the concave region of the stamen in the form of a mask.

[0022] The image segmentation model performs image segmentation on the original image of the Chrysanthemum morifolium based on the mask image data fused from all image feature information, and extracts the first segmentation image corresponding to the outer edge region of the flower and the second segmentation image corresponding to the concave region of the stamen from the Chrysanthemum morifolium image.

[0023] Optionally, the step of calculating the first center of the first minimum circumscribed circle surrounding all first pixels in the outer edge region of the flower based on the first segmentation image to determine the outer contour center of the chrysanthemum to be picked; and calculating the second center of the second minimum circumscribed circle surrounding all second pixels in the stamen depression region based on the second segmentation image to determine the depression center of the chrysanthemum to be picked, includes:

[0024] Extract all first pixels belonging to the outer edge region of the flower from the first segmentation image corresponding to the outer edge region of the flower, so as to obtain the first two-dimensional pixel coordinates of each first pixel in the image of the golden chrysanthemum;

[0025] Extract all second pixels belonging to the flower stamen concave region from the second segmentation image corresponding to the flower stamen concave region, so as to obtain the second two-dimensional pixel coordinates of each second pixel in the golden chrysanthemum image;

[0026] The minimum circumcircle algorithm is used to fit the smallest area and completely enclose all the first pixels based on the coordinates of all the first two-dimensional pixels. The first center of the first minimum circumcircle is determined as the outer contour center of the chrysanthemum to be picked, and the first radius of the first minimum circumcircle is determined as the outer contour radius of the chrysanthemum to be picked.

[0027] The minimum circumcircle algorithm is used to fit the second minimum circumcircle with the smallest area that completely surrounds all the second pixels based on the coordinates of all the second two-dimensional pixels. The center of the second circle corresponding to the second minimum circumcircle is determined as the concave center of the golden chrysanthemum to be picked.

[0028] Optionally, the step of calculating and determining the offset vector coordinates from the center of the outer contour to the center of the depression, and using the arctangent function to calculate and determine the growth orientation angle of the chrysanthemum to be harvested in the chrysanthemum image plane based on the offset vector coordinates, includes:

[0029] Obtain the third two-dimensional pixel coordinates corresponding to the outer contour center of the chrysanthemum to be picked and the fourth two-dimensional pixel coordinates corresponding to the concave center;

[0030] The offset vector coordinates pointing from the outer contour center to the concave center are determined based on the difference between the third two-dimensional pixel coordinates corresponding to the outer contour center and the fourth two-dimensional pixel coordinates corresponding to the concave center.

[0031] Using a preset arctangent function, the angle value of the offset vector in the image plane of the golden chrysanthemum is calculated based on the coordinates of the offset vector, and the angle value is determined as the growth orientation angle of the golden chrysanthemum to be harvested in the image plane of the golden chrysanthemum.

[0032] Optionally, the step of calling a preset calibration model to determine the growth tilt angle of the golden chrysanthemum based on the ratio between the magnitude of the offset vector and the outer contour radius of the golden chrysanthemum to be harvested includes:

[0033] The magnitude of the offset vector from the center of the outer contour to the center of the recess is calculated based on the coordinates of the offset vector pointing from the center of the outer contour to the center of the recess.

[0034] The ratio between the magnitude of the offset vector and the outer contour radius of the chrysanthemum to be harvested is calculated and input into a preset calibration model to output the growth tilt angle of the chrysanthemum to be harvested. The calibration model is a linear fitting model, which is based on multiple sets of chrysanthemum samples with known growth tilt angles. It is obtained by fitting the correspondence between the ratio of the magnitude of the offset vector pointing from the center of the outer contour to the center of the depression and the outer contour radius and the growth tilt angle.

[0035] Optionally, the step of constructing a growth posture space vector of the chrysanthemum to be harvested based on the growth orientation angle and the growth tilt angle, and determining the target harvesting point of the chrysanthemum to be harvested based on the growth posture space vector and the three-dimensional spatial coordinates of the outer contour center, includes:

[0036] Based on the coordinate system of the robotic arm, and based on the mapping relationship between spherical coordinates and Cartesian coordinates, a growth posture space vector representing the spatial growth posture of the chrysanthemum to be harvested is constructed according to the growth orientation angle and growth tilt angle of the chrysanthemum to be harvested.

[0037] By using preset camera-robotic hand-eye calibration parameters, the two-dimensional pixel coordinates of the outer contour center of the chrysanthemum to be picked are converted into three-dimensional spatial coordinates of the outer contour center under the robotic arm's working coordinate system.

[0038] Determine the picking offset distance from the chrysanthemum to be picked, and calculate the first product between the picking offset distance and the growth posture space vector. The picking offset distance represents the distance from the center of the outer contour of the chrysanthemum to be picked to the optimal cutting position of the flower stem below the flower stamen along the direction from the center of the outer contour to the center of the flower stamen depression.

[0039] Calculate and determine the first sum between the first product and the three-dimensional spatial coordinates of the outer contour center, and use the first sum as the target picking point of the golden chrysanthemum to be picked, so as to complete the picking point positioning of the golden chrysanthemum.

[0040] Optionally, the underlying network architecture of the image segmentation model includes the U2net model;

[0041] The growth orientation angle is the horizontal deflection angle of the stamen relative to the center of the outer contour of the golden chrysanthemum to be harvested in the two-dimensional plane of the image, which represents the orientation of the flower in the plane.

[0042] The growth tilt angle is the tilt angle of the golden chrysanthemum to be harvested relative to the vertical direction in three-dimensional space, which represents the degree of tilt of the flower's posture in space.

[0043] A chrysanthemum picking point positioning device provided for another purpose of this application includes:

[0044] The image segmentation module is configured to acquire images of Chrysanthemum morifolium containing one or more Chrysanthemum morifolium images to be picked, input the Chrysanthemum morifolium images into an image segmentation model that has been trained to convergence, and extract a first segmentation image corresponding to the outer edge region of the flower of the Chrysanthemum morifolium to be picked and a second segmentation image corresponding to the concave region of the stamen from the Chrysanthemum morifolium images.

[0045] The center determination module is configured to calculate the center of the first minimum circumscribed circle of the first minimum circumscribed circle of the flower's outer edge region based on the first segmentation image, so as to determine the outer contour center of the chrysanthemum to be picked; and to calculate the center of the second minimum circumscribed circle of the second minimum circumscribed circle of the second minimum circumscribed circle of the flower's stamen depression region based on the second segmentation image, so as to determine the depression center of the chrysanthemum to be picked.

[0046] The growth orientation angle determination module is configured to calculate and determine the offset vector coordinates of the outer contour center pointing to the center of the depression, and to use the arctangent function to calculate and determine the growth orientation angle of the chrysanthemum to be picked in the chrysanthemum image plane based on the offset vector coordinates;

[0047] The growth tilt angle determination module is configured to call a preset calibration model to determine the growth tilt angle of the golden chrysanthemum based on the ratio between the magnitude of the offset vector and the outer contour radius of the golden chrysanthemum to be harvested.

[0048] The picking point positioning module is configured to construct a growth posture space vector of the chrysanthemum to be picked based on the growth orientation angle and the growth tilt angle, and determine the target picking point of the chrysanthemum to be picked based on the growth posture space vector and the three-dimensional spatial coordinates of the outer contour center, so as to complete the picking point positioning of the chrysanthemum.

[0049] An electronic device provided for another purpose of this application includes a central processing unit and a memory, the central processing unit being configured to invoke and run a computer program stored in the memory to perform the steps of the method for locating the picking point of the golden chrysanthemum described in this application.

[0050] A computer-readable storage medium is provided for another purpose of this application, which stores, in the form of computer-readable instructions, a computer program implemented according to the method for locating the picking point of the golden chrysanthemum, which, when called by a computer, executes the steps included in the corresponding method.

[0051] Compared to existing technologies, this application addresses the shortcomings of conventional cable-driven parallel robots, which suffer from issues such as the robotic arm directly gripping fragile flower stamens, leading to tearing and detachment of the stamens, and the robotic arm being limited to gripping or cutting actions at fixed angles. This rigid operating mode results in uneven stress on the flower, especially when the stamens shift or the flower tilts, easily causing problems such as stamen breakage and petal tearing. This application offers advantages including, but not limited to, the following:

[0052] Firstly, addressing the technical shortcomings of existing cable-driven parallel robots in accurately locating the picking point of Chrysanthemum morifolium, which easily damages the stamens or leaves excessively long stems, this application uses a U2Net model to precisely extract the outer edge of the flower and the concave area of ​​the stamen. By fitting the minimum circumcircle to determine the center of the outer contour and the center of the concave area, the offset vector coordinates are determined. The growth orientation angle is calculated based on the offset vector coordinates. Then, based on the growth orientation angle and the growth tilt angle, the growth posture space vector of the Chrysanthemum morifolium to be picked is constructed. Finally, the optimal cutting position of the stem is locked. This application can avoid the robot arm directly gripping the fragile stamens of Chrysanthemum morifolium, preventing the stamens from tearing and falling off, and ensuring that the cutting position is far away from the stamens and close to the base of the stem, effectively controlling the residual length of the stem and ensuring the integrity of the flower and the quality of the product.

[0053] Secondly, addressing the issue of existing cable-driven parallel robots easily misjudging targets in scenarios with multiple adjacent flowers and occluded leaves, this application employs the U2Net model to achieve pixel-level precise segmentation of the outer edge of the flower and the concave area of ​​the stamen. This effectively separates a single flower from the background and adjacent flowers, avoiding the situation where multiple targets are misjudged as a single target. It ensures that the robotic arm focuses only on a single flower to be picked and will not touch adjacent flowers or mixed leaves at the same time. From the positioning perspective, it eliminates damage such as stamen compression deformation and petal breakage, improving the stability of batch picking.

[0054] Thirdly, addressing the shortcomings of existing cable-driven parallel robots, such as lack of posture adaptation and uneven force on flowers due to fixed-angle operation, this application calculates the growth orientation angle using an offset vector and quantifies the growth tilt angle by combining the ratio of the offset vector's magnitude to the outer contour radius. This constructs a growth posture space vector that characterizes the three-dimensional spatial posture of the flower. This allows the robot arm to adapt to the flower's planar orientation deflection and three-dimensional tilt state, performing clamping and cutting actions along the flower's natural growth posture. This replaces the traditional rigid fixed-angle operation mode, avoiding problems such as flower stamen breakage and petal tearing caused by uneven force on the flower, and adapting to the complex scenario of random flower growth in the field.

[0055] Fourth, this application eliminates the need for manual intervention in image segmentation and harvesting point localization, avoiding the inefficiency and high cost of manual harvesting while improving harvesting point localization accuracy to the millimeter level, significantly reducing the harvesting damage rate of Chrysanthemum morifolium. Simultaneously, its anti-interference segmentation capability and posture adaptive design can adapt to complex field scenarios such as dense growth of multiple flowers and leaf occlusion, meeting the high-efficiency harvesting needs of large-scale planting, contributing to the automation upgrade of the Chrysanthemum morifolium industry, and ensuring the commercial quality and planting returns of this high-value crop. Attached Figure Description

[0056] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0057] Figure 1 This is a flowchart illustrating the method for locating the picking point of Chrysanthemum morifolium in the embodiments of this application;

[0058] Figure 2 This is a schematic diagram of a cable-driven parallel robot harvesting golden chrysanthemums in an embodiment of this application.

[0059] Figure 3 This is a schematic diagram of the golden chrysanthemum to be harvested in an embodiment of this application;

[0060] Figure 4 This is a flowchart illustrating the process of extracting the first segmentation image corresponding to the outer edge region of the flower and the second segmentation image corresponding to the concave region of the stamen in this embodiment of the application.

[0061] Figure 5 This is a schematic diagram of the process for determining the concave center of the golden chrysanthemum to be harvested in an embodiment of this application;

[0062] Figure 6 This is a schematic diagram illustrating the process of determining the growth orientation angle of the golden chrysanthemum to be harvested within the image plane of the golden chrysanthemum in this application embodiment;

[0063] Figure 7 This is a schematic diagram illustrating the process of determining the growth tilt angle of the golden chrysanthemum in this embodiment of the application;

[0064] Figure 8 This is a flowchart illustrating the process of determining the target picking point for the golden chrysanthemum in this embodiment of the application.

[0065] Figure 9 This is a schematic diagram of the golden chrysanthemum picking point positioning device in the embodiments of this application;

[0066] Figure 10 This is a schematic diagram of the structure of the computer device in the embodiments of this application. Detailed Implementation

[0067] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0068] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this application means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or wireless coupling. The term “and / or” as used herein includes all or any units and all combinations of one or more associated listed items.

[0069] Those skilled in the art will understand that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0070] Those skilled in the art will understand that the terms "client," "terminal," and "terminal device" as used herein include both devices that receive wireless signals, devices that only possess wireless signal receiver capabilities without transmission capabilities, and devices with receiving and transmitting hardware, devices that have receiving and transmitting hardware capable of bidirectional communication over a bidirectional communication link. Such devices may include: cellular or other communication devices such as personal computers or tablets, having single-line displays, multi-line displays, or cellular or other communication devices without multi-line displays; PCS (Personal Communications Service) that can combine voice, data processing, fax, and / or data communication capabilities; PDAs (Personal Digital Assistants) that may include radio frequency receivers, pagers, internet / intranet access, web browsers, notebooks, calendars, and / or GPS (Global Positioning System) receivers; and conventional laptops and / or handheld computers or other devices that have and / or include radio frequency receivers. As used herein, "client," "terminal," and "terminal device" can be portable, transportable, installed in a means of transportation (air, sea, and / or land), or suitable and / or configured to operate locally and / or in a distributed manner, operating in any other location on Earth and / or in space. "Client," "terminal," and "terminal device" as used herein can also be a communication terminal, an internet access terminal, or a music / video playback terminal, such as a PDA, a MID (Mobile Internet Device), and / or a mobile phone with music / video playback capabilities, or a smart TV, set-top box, etc.

[0071] The hardware referred to by the names "server," "client," and "service node" in this application is essentially an electronic device with the equivalent capabilities of a personal computer. It is a hardware device with the necessary components revealed by the von Neumann architecture, such as a central processing unit (including an arithmetic logic unit and a control unit), memory, input devices, and output devices. The computer program is stored in its memory, and the central processing unit loads the program stored in the secondary storage into the main memory to run it, execute the instructions in the program, and interact with the input and output devices to complete specific functions.

[0072] It should be noted that the concept of "server" used in this application can also be extended to the case of server clusters. Based on the network deployment principles understood by those skilled in the art, the servers should be logically divided. Physically, these servers can be independent of each other but accessible through interfaces, or they can be integrated into a single physical computer or a computer cluster. Those skilled in the art should understand this flexibility and should not use it to constrain the implementation of the network deployment method in this application.

[0073] One or more of the technical features of this application, unless explicitly specified herein, can be deployed on a server and accessed by a client remotely calling the online service interface provided by the server, or can be directly deployed and run on a client for access.

[0074] Unless otherwise specified, the neural network models referenced or potentially referenced in this application may be deployed on a remote server and invoked remotely on the client, or deployed on a client with the capability to invoke directly. In some embodiments, when running on the client, the corresponding intelligence may be acquired through transfer learning in order to reduce the requirements on the client's hardware resources and avoid excessive consumption of the client's hardware resources.

[0075] Unless otherwise specified, all data involved in this application may be stored remotely on a server or on a local terminal device, as long as it is suitable for use by the technical solution of this application.

[0076] Those skilled in the art will understand that although the various methods in this application are described based on the same concept and thus present commonality among them, they can be performed independently unless otherwise specified. Similarly, the various embodiments disclosed in this application are all based on the same inventive concept; therefore, concepts expressed in the same way, as well as concepts that are appropriately changed for convenience but are expressed differently, should be understood equivalently.

[0077] Unless otherwise expressly stated, the various embodiments disclosed in this application can be combined in a cross-cutting manner to flexibly construct new embodiments, as long as such combination does not depart from the inventive spirit of this application and can meet the needs of the prior art or solve a certain deficiency in the prior art. Those skilled in the art should be aware of such modifications.

[0078] Please see Figure 1 , Figure 2 as well as Figure 3 In one embodiment of the method for locating the picking point of Chrysanthemum morifolium of this application, the method includes:

[0079] Step S10: Obtain images of Chrysanthemum morifolium containing one or more Chrysanthemum morifolium images to be picked, input the Chrysanthemum morifolium images into an image segmentation model that has been trained to convergence, and extract the first segmentation image corresponding to the outer edge region of the flower of the Chrysanthemum morifolium to be picked and the second segmentation image corresponding to the concave region of the stamen from the Chrysanthemum morifolium images.

[0080] The Chrysanthemum morifolium picking point localization system in the cable-driven parallel robot can respond to the instruction to locate the picking point of Chrysanthemum morifolium, acquire Chrysanthemum morifolium images containing one or more Chrysanthemum morifoliums to be picked, input the Chrysanthemum morifolium images into an image segmentation model that has been trained to convergence, and extract a first segmentation image corresponding to the outer edge region of the flower of the Chrysanthemum morifolium to be picked and a second segmentation image corresponding to the concave region of the stamen from the Chrysanthemum morifolium images; wherein, the basic network architecture of the image segmentation model includes the U2net model.

[0081] Specifically, the image of the golden chrysanthemum is input into an image segmentation model that has been trained to convergence. The model extracts a first segmentation image corresponding to the outer edge region of the flower to be picked and a second segmentation image corresponding to the concave region of the stamen from the image. In the prior art, the robotic arm can only blindly wrap the overall outline of the flower and cannot distinguish between the petals and the background (including leaves, flower stems, etc.), which can easily lead to tearing and falling off of the fragile petals. The first segmentation image clearly separates the petal area of ​​the golden chrysanthemum to be picked from the background through a mask. Even if there are slight defects or wrinkles on the edge of the petal, the outer outline of the petal can be accurately delineated, and the area of ​​the petal that cannot be touched can be clearly identified. The robotic arm can plan a safe area for gripping or cutting based on this outline and operate only in the flower stem area outside or below the outer edge region of the flower, completely avoiding direct contact with the fragile petals.

[0082] The concave area of ​​the stamen is the core connecting the stem and petals of the golden chrysanthemum. The stem directly below the stamen is thick, highly lignified, and far from the petals, making it the optimal cutting or clamping position. This position avoids damaging the petals and ensures the stem remains flat after harvesting. Existing techniques typically cut roughly along the center of the flower's overall outline, often resulting in cutting to the base of the petals or leaving excessively long stems. The second segmentation image directly and precisely locates the core position of the concave area of ​​the stamen, providing a crucial basis for subsequent calculations of the growth orientation angle and determination of the harvesting point. This ensures that the robotic arm's cutting position accurately falls on the stem below the stamen, resolving the cutting deviation problem of existing techniques.

[0083] In some embodiments, a Cable-Driven Parallel Robot (CDPR), also known as a cable-driven parallel robot or a rope-traction parallel robot, is a parallel robot that uses multiple flexible ropes to replace traditional rigid links and drives an end effector to achieve spatial movement through rope tension.

[0084] In some embodiments, the basic network architecture of the image segmentation model includes the U2net model; the growth orientation angle is the horizontal deflection angle of the stamen relative to the center of the outer contour in the two-dimensional plane of the image of the chrysanthemum to be picked, which represents the orientation of the flower in the plane; the growth tilt angle is the tilt angle of the flower body relative to the vertical direction in the three-dimensional space of the chrysanthemum to be picked, which represents the degree of tilt of the flower's posture in space.

[0085] In some embodiments, the concave areas of the stamens and the wrinkles at the edges of the petals of the Chrysanthemum morifolium are considered detailed features. U2Net's small-scale features can accurately capture these details, avoiding blurred segmentation boundaries due to feature loss, such as confusion between stamens and petals, or misjudgment of pixels at wrinkles. The overall outline of the flower and the global association between the target and the background (leaves, soil, etc.) are considered large-scale semantic features. U2Net's deep encoding can stably lock the target region, avoiding missed or incorrect segmentation due to cluttered backgrounds, and ensuring the integrity of the segmentation map. The core architecture of U2Net is a nested U-shaped structure and multi-stage downsampling. It achieves full-scale feature extraction from pixel-level details to global semantics through four encoder-decoder modules, and each stage retains an independent feature map. Finally, all scale information is integrated through a feature fusion module.

[0086] Furthermore, U2Net employs depthwise separable convolutions and a reduced channel count design, resulting in a small number of model parameters, fast inference speed, and stable operation without the need for a high-performance GPU. Field harvesting equipment typically uses embedded processors; U2Net's lightweight nature reduces hardware deployment costs and avoids stuttering and latency caused by overly heavy models.

[0087] Furthermore, field-grown Chrysanthemum morifolium often faces problems such as partial leaf occlusion, morning dew reflection, incomplete petals, and cluttered soil background. U2Net can filter these interferences and avoid situations such as missing outlines or misidentification of the background as the target in the segmentation image. U2Net enhances the discriminative power of target features through Local Contrast Normalization (LCN) and attention mechanisms, and has a strong ability to suppress interferences such as background noise, partial occlusion, and changes in lighting. Even if there is partial occlusion or feature distortion in the target area, the model can still lock the target through global feature association.

[0088] U2Net possesses multiple advantages, including precise segmentation of details, lightweight real-time deployment, resistance to interference in complex environments, and efficient separation of multiple targets. It can adapt to the segmentation requirements for precise harvesting of Chrysanthemum morifolium in fields, ensuring the segmentation accuracy of the flower's outer edge avoidance boundary and the concave picking point of the stamen, while also meeting the real-time and low-cost requirements of embedded deployment. It provides highly reliable core data support for subsequent center positioning, orientation angle calculation, and robot path planning. Therefore, the U2Net model is used as the image segmentation model in this application. Once the U2Net model is trained to convergence, it can be put into production use.

[0089] In a specific embodiment, please refer to Figure 4 The steps of inputting the image of the golden chrysanthemum into an image segmentation model that has been trained to convergence, and extracting the first segmentation image corresponding to the outer edge region of the flower of the golden chrysanthemum to be picked and the second segmentation image corresponding to the concave region of the stamen from the image of the golden chrysanthemum include:

[0090] Step S101: The image segmentation model, which has been trained to convergence, performs multi-level encoding to downsample the original image of the Chrysanthemum morifolium and produces intermediate feature information corresponding to each scale. The intermediate feature information represents the contour features of the outer edge region of the flower and the contour features of the concave region of the stamen in the Chrysanthemum morifolium image.

[0091] To accurately extract the contour features of the outer edge region of the flower and the contour features of the concave region of the stamen of the golden chrysanthemum to be harvested, and to filter out background interference such as leaves and soil, a dual feature support of detailed and global features is provided for the subsequent decoding stage. Images of golden chrysanthemums containing the flowers to be harvested are input into a U2net model that has been trained to convergence. The encoder module of the U2net model performs downsampling operations, simultaneously producing intermediate feature information corresponding to each scale. During downsampling, the field of view of the U2net model gradually expands as the scale decreases. Small-scale features focus on local detailed features such as petal edge wrinkles and tiny concave regions of the stamen; large-scale features focus on global semantic features such as the overall outline of the flower and the spatial relationship between the stamen and petals, achieving multi-dimensional feature complementarity between local detailed features and global semantic features.

[0092] Step S102: The image segmentation model, which has been trained to convergence, performs multi-level decoding. Based on the intermediate feature information at the smallest scale, it performs upsampling operation and successively jumps and fuses the intermediate feature information produced by the same level encoding as a reference to decode and generate higher-scale image feature information. The image feature information is used to represent the contour features of the outer edge region of the flower and the concave region of the stamen in the form of a mask.

[0093] The image feature information is used to represent the contour features of the outer edge region of the flower and the concave region of the stamen in the form of a mask. The mask is essentially a binary feature map. "1" marks the pixel position of the outer edge of the flower and the concave region of the stamen, and "0" marks the background. This provides pixel-level target region markings for subsequent segmentation and clarifies the category of each pixel.

[0094] Starting with the smallest-scale intermediate feature information output by the encoder, the decoder of the U2net model performs upsampling operations. During each upsampling stage, the intermediate feature information produced by the encoder at the same level is simultaneously fused to generate higher-scale image feature information that matches the scale of the original image.

[0095] Upsampling can restore the image size, but it can easily lead to blurred contour boundaries; while skip-connection fusion can supplement the decoder with local detail features of the corresponding scale of the encoder, making up for the loss of details in upsampling, so that the generated features retain both global positioning accuracy and clear contour boundaries.

[0096] Step S103: The image segmentation model performs image segmentation on the original image of the Chrysanthemum morifolium image based on the mask image data fused from all image feature information, and extracts the first segmentation image corresponding to the outer edge region of the flower and the second segmentation image corresponding to the concave region of the stamen from the Chrysanthemum morifolium image.

[0097] The U2net model fuses all the image feature information generated by decoding to form the final mask image data. Based on this mask data, the original image is filtered at the pixel level, and pixels marked as the outer edge of the flower in the mask are retained to generate the first segmentation image. Pixels marked as the stamen recessed area in the mask are retained to generate the second segmentation image.

[0098] Step S20: Calculate the first center of the first minimum circumscribed circle surrounding all first pixels in the outer edge region of the flower according to the first segmentation image, so as to determine the outer contour center of the golden chrysanthemum to be picked; calculate the second center of the second minimum circumscribed circle surrounding all second pixels in the stamen concave region according to the second segmentation image, so as to determine the concave center of the golden chrysanthemum to be picked.

[0099] A method is used to obtain images of one or more Chrysanthemum morifolium plants to be picked. These images are then input into an image segmentation model that has been trained to convergence. The model extracts a first segmentation image corresponding to the outer edge region of the flower and a second segmentation image corresponding to the stamen recess region from the images. Based on the first segmentation image, the model calculates the first center of the first minimum circumscribed circle surrounding all first pixels in the outer edge region of the flower to determine the outer contour center of the Chrysanthemum morifolium. Based on the second segmentation image, the model calculates the second center of the second minimum circumscribed circle surrounding all second pixels in the stamen recess region to determine the recess center of the Chrysanthemum morifolium.

[0100] In some embodiments, please refer to Figure 5 The steps of calculating the first center of the first minimum circumscribed circle surrounding all first pixels in the outer edge region of the flower based on the first segmentation image to determine the outer contour center of the chrysanthemum to be picked; and calculating the second center of the second minimum circumscribed circle surrounding all second pixels in the stamen depression region based on the second segmentation image to determine the depression center of the chrysanthemum to be picked, include:

[0101] Step S201: Extract all first pixels belonging to the outer edge region of the flower from the first segmentation image corresponding to the outer edge region of the flower, so as to obtain the first two-dimensional pixel coordinates of each first pixel in the image of the golden chrysanthemum.

[0102] Based on the first segmentation map output in step S103 above, that is, the mask map of the outer edge region of the flower, all first pixels marked as target pixels are selected, and the two-dimensional pixel coordinates of these pixels in the original Chrysanthemum indicum image are read synchronously to form the first two-dimensional pixel coordinate set.

[0103] Step S202: Extract all second pixels belonging to the flower stamen concave region from the second segmentation image corresponding to the flower stamen concave region, so as to obtain the second two-dimensional pixel coordinates of each second pixel in the golden chrysanthemum image;

[0104] Based on the second segmentation map output in step S103 above, that is, the mask map of the flower stamen concave region, all second pixels marked as target pixels are filtered out, and the two-dimensional pixel coordinates of these pixels in the original Chrysanthemum morifolium image are read to form a second two-dimensional pixel coordinate set.

[0105] Step S203: Using the minimum circumcircle algorithm, fit the smallest area of ​​the first minimum circumcircle that completely surrounds all the first two-dimensional pixel points based on all the first two-dimensional pixel coordinates. Determine the first center of the first minimum circumcircle as the outer contour center of the chrysanthemum to be picked, and determine the first radius of the first minimum circumcircle as the outer contour radius of the chrysanthemum to be picked.

[0106] The minimum circumcircle algorithm includes Welzl's fast algorithm, etc. This algorithm finds the smallest circle that covers all target pixels through geometric iteration. Its core advantage is that it does not rely on local pixel distribution, but only on the overall contour boundary. Even if the outer edge of the flower has slight defects, petal wrinkles, or other irregularities, it can still fit a stable circle that reflects the true outer boundary of the flower. The center of the circle is the geometric center of the contour, and the radius is the equivalent size of the contour. Using the minimum circumcircle algorithm, based on the first two-dimensional pixel coordinate set, a first minimum circumcircle with the smallest area that completely surrounds all first pixels is fitted and generated. The center coordinates of the first circle corresponding to this first minimum circumcircle, which are also the third two-dimensional pixel coordinates, are extracted and determined as the outer contour center of the chrysanthemum to be picked. The radius of the first minimum circumcircle is extracted and determined as the outer contour radius of the chrysanthemum to be picked.

[0107] Step S204: Using the minimum circumcircle algorithm, fit the second minimum circumcircle with the smallest area that completely surrounds all the second pixels based on the coordinates of all the second two-dimensional pixels, and determine the center of the second circle corresponding to the second minimum circumcircle as the concave center of the golden chrysanthemum to be picked.

[0108] The concave area of ​​the flower stamen is distributed in a concentrated, almost circular pattern. The pixel density is high and the boundaries are relatively regular. The minimum circumcircle algorithm can accurately capture its core position and avoid center offset caused by uneven distribution of pollen and texture inside the flower stamen. Compared with the calculation of the average pixel coordinates, the center of the circumcircle can better reflect the physical center of the connection between the flower stamen and the flower stem, and has stronger stability.

[0109] The minimum circumcircle algorithm is used to fit and generate the second minimum circumcircle with the smallest area that completely surrounds all the second pixels based on the second two-dimensional pixel coordinate set. The center coordinates of the second circle corresponding to the second minimum circumcircle, which are also the fourth two-dimensional pixel coordinates, are extracted and determined as the concave center of the golden chrysanthemum to be picked.

[0110] Step S30: Calculate and determine the offset vector coordinates of the outer contour center pointing to the center of the depression, and use the arctangent function to calculate and determine the growth orientation angle of the chrysanthemum to be picked in the chrysanthemum image plane based on the offset vector coordinates;

[0111] Based on the first segmentation image, the center of the first minimum circumscribed circle surrounding all first pixels in the outer edge region of the flower is calculated to determine the outer contour center of the chrysanthemum to be picked. Then, based on the second segmentation image, the center of the second minimum circumscribed circle surrounding all second pixels in the stamen depression region is calculated to determine the depression center of the chrysanthemum to be picked. Next, the offset vector coordinates pointing from the outer contour center to the depression center are calculated, and the arctangent function is used to calculate the growth orientation angle of the chrysanthemum to be picked within the chrysanthemum image plane based on these offset vector coordinates. The growth orientation angle is the horizontal deflection angle of the stamen relative to the outer contour center within the two-dimensional plane of the image, representing the flower's orientation within the plane.

[0112] In some embodiments, please refer to Figure 6 The steps of calculating and determining the offset vector coordinates from the center of the outer contour to the center of the depression, and using the arctangent function to calculate and determine the growth orientation angle of the chrysanthemum to be harvested in the chrysanthemum image plane based on the offset vector coordinates, include:

[0113] Step S301: Obtain the third two-dimensional pixel coordinates corresponding to the outer contour center of the chrysanthemum to be picked and the fourth two-dimensional pixel coordinates corresponding to the concave center.

[0114] From step S203 above, the third two-dimensional pixel coordinates corresponding to the center of the outer contour of the chrysanthemum to be picked are extracted; from step S204 above, the fourth two-dimensional pixel coordinates corresponding to the center of the depression are extracted, so as to ensure that the third two-dimensional pixel coordinates and the fourth two-dimensional pixel coordinates are based on the original pixel coordinate system of the chrysanthemum image.

[0115] Step S302: Determine the offset vector coordinates of the outer contour center pointing to the concave center based on the difference between the third two-dimensional pixel coordinates corresponding to the outer contour center and the fourth two-dimensional pixel coordinates corresponding to the concave center.

[0116] In a two-dimensional plane, the offset vector coordinates between two points are essentially the set of differences between the endpoint coordinates and the starting point coordinates. Their direction directly reflects the spatial orientation from the starting point to the endpoint, and their magnitude reflects the straight-line distance between the two points. In this step, the direction of the offset vector from the outer contour center to the concave center corresponds to the natural orientation of the flower body center towards the stamen core, which highly aligns with the growth characteristics of the golden chrysanthemum's stamen within the flower, where its orientation is determined by its position. Based on the difference between the third two-dimensional pixel coordinates corresponding to the outer contour center and the fourth two-dimensional pixel coordinates corresponding to the concave center, using the third two-dimensional pixel coordinates of the outer contour center as the starting point and the fourth two-dimensional pixel coordinates of the concave center as the endpoint, the offset vector coordinates from the outer contour center to the concave center are calculated using the coordinate difference. The offset vector coordinates can be expressed as... .

[0117] Step S303: Using a preset arctangent function, calculate and determine the angle value of the offset vector in the image plane of the golden chrysanthemum based on the coordinates of the offset vector, and determine the angle value as the growth orientation angle of the golden chrysanthemum to be harvested in the image plane of the golden chrysanthemum.

[0118] The arctangent function is expressed as: arctangent function It can directly identify the quadrant where the vector is located and point the offset vector coordinates from the center of the outer contour to the center of the depression. Input to the arctangent function The angle value of the offset vector pointing from the center of the outer contour to the center of the depression in the image plane of Chrysanthemum morifolium is calculated. Taking the positive x-axis direction of the image plane (such as horizontal to the right) as the reference, the angle is rotated counterclockwise to the direction of the offset vector. This angle directly quantifies the deflection direction of the stamen relative to the center of the flower body, that is, the growth orientation of the flower. Therefore, the angle value of the offset vector in the image plane of Chrysanthemum morifolium is taken as the growth orientation angle of the Chrysanthemum morifolium to be harvested in the image plane of Chrysanthemum morifolium.

[0119] Step S40: Call the preset calibration model to determine the growth tilt angle of the golden chrysanthemum based on the ratio between the magnitude of the offset vector and the outer contour radius of the golden chrysanthemum to be harvested.

[0120] After calculating and determining the offset vector coordinates pointing from the outer contour center to the concave center, and using the arctangent function to calculate and determine the growth orientation angle of the chrysanthemum to be picked in the chrysanthemum image plane based on the offset vector coordinates, a preset calibration model is called to determine the growth tilt angle of the chrysanthemum based on the ratio between the magnitude of the offset vector and the outer contour radius of the chrysanthemum to be picked; wherein, the growth tilt angle is the tilt angle of the chrysanthemum to be picked relative to the vertical direction in three-dimensional space, which characterizes the degree of tilt of the flower posture in space.

[0121] In some embodiments, please refer to Figure 7 The step of determining the growth tilt angle of the golden chrysanthemum by calling a preset calibration model and using the ratio between the magnitude of the offset vector and the outer contour radius of the golden chrysanthemum to be harvested includes:

[0122] Step S401: Calculate and determine the magnitude of the offset vector from the outer contour center to the recess center based on the coordinates of the offset vector from the outer contour center to the recess center.

[0123] Based on the coordinates of the offset vector pointing from the center of the outer contour to the center of the depression, the magnitude of the offset vector pointing from the center of the outer contour to the center of the depression is calculated and determined, which is the straight-line distance between the center of the outer contour and the center of the depression of the golden chrysanthemum. Specifically, based on the coordinates of the offset vector pointing from the center of the outer contour to the center of the depression, the straight-line distance corresponding to the offset vector pointing from the center of the outer contour to the center of the depression is obtained through the two-dimensional vector magnitude formula. This straight-line distance is the magnitude of the offset vector, and its function is to accurately quantify the actual distance between the two key points of the outer contour center and the center of the depression.

[0124] Step S402: Calculate and determine the ratio between the magnitude of the offset vector and the outer contour radius of the chrysanthemum to be harvested, and input the ratio into a preset calibration model to output the growth tilt angle of the chrysanthemum to be harvested. The calibration model is a linear fitting model, which is based on multiple sets of chrysanthemum samples with known growth tilt angles. It is obtained by fitting the correspondence between the ratio of the magnitude of the offset vector pointing from the center of the outer contour to the center of the depression and the outer contour radius and the growth tilt angle.

[0125] The ratio between the magnitude of the offset vector and the outer radius of the chrysanthemum to be harvested is calculated. This ratio eliminates the influence of different flower sizes on the determination of the degree of offset, making the offset of flowers of different sizes comparable. Then, the ratio between the magnitude of the offset vector and the outer radius of the chrysanthemum to be harvested is input into a pre-set calibration model. This calibration model can output the growth tilt angle of the chrysanthemum to be harvested. This calibration model is a linear fitting model, and its establishment process includes: first, collecting multiple sets of chrysanthemum samples with known actual growth tilt angles; for each set of samples, calculating the ratio of the magnitude of the offset vector pointing from the center of the outer contour to the center of the depression to the outer contour radius; then, finding a fixed correspondence between this ratio and the actual growth tilt angle of the sample through data fitting, ultimately forming a calibration model that can be directly used to calculate the growth tilt angle.

[0126] Step S50: Construct the growth posture space vector of the chrysanthemum to be harvested based on the growth orientation angle and the growth tilt angle, and determine the target harvesting point of the chrysanthemum to be harvested based on the growth posture space vector and the three-dimensional spatial coordinates of the outer contour center, so as to complete the harvesting point positioning of the chrysanthemum.

[0127] After determining the growth tilt angle of the chrysanthemum by calling a preset calibration model based on the ratio between the magnitude of the offset vector and the outer contour radius of the chrysanthemum to be picked, a growth posture space vector of the chrysanthemum to be picked is constructed based on the growth orientation angle and the growth tilt angle. The target picking point of the chrysanthemum to be picked is determined based on the growth posture space vector and the three-dimensional spatial coordinates of the outer contour center, so as to complete the picking point positioning of the chrysanthemum.

[0128] In some embodiments, please refer to Figure 8 The steps of constructing a growth posture space vector of the chrysanthemum to be harvested based on the growth orientation angle and the growth tilt angle, and determining the target harvesting point of the chrysanthemum to be harvested based on the growth posture space vector and the three-dimensional spatial coordinates of the outer contour center, include:

[0129] Step S501: Based on the robot's work coordinate system, and based on the mapping relationship between spherical coordinates and Cartesian coordinates, construct a growth posture space vector that represents the spatial growth posture of the chrysanthemum to be harvested, according to the growth orientation angle and growth tilt angle of the chrysanthemum to be harvested.

[0130] Using the coordinate system followed by the robotic arm as a reference, the core principle is to transform the growth orientation angle and growth tilt angle, which reflect the growth status of the golden chrysanthemum, into a growth posture space vector that comprehensively represents the three-dimensional growth posture of the flower, through a fixed mapping rule between spherical coordinates and Cartesian coordinates. This vector not only contains the flower's orientation information in the plane but also integrates its tilt state in three-dimensional space, providing a precise posture reference for the robotic arm to locate the picking point and ensuring that the robotic arm can accurately adapt to the flower's spatial growth angle.

[0131] Step S502: Using preset camera-robotic hand-eye calibration parameters, convert the two-dimensional pixel coordinates of the outer contour center of the chrysanthemum to be picked into three-dimensional spatial coordinates of the outer contour center under the robotic arm's working coordinate system.

[0132] Based on pre-set hand-eye calibration parameters between the camera and the robotic arm, a coordinate system transformation is completed. Specifically, the previously acquired two-dimensional pixel coordinates of the outer contour center, based on the pixel coordinate system of the Chrysanthemum morifolium image, are converted into three-dimensional spatial coordinates in the robotic arm's operating coordinate system. The core purpose of this transformation is to eliminate the deviation between the image coordinate system and the robotic arm's operating coordinate system, allowing the position information of the outer contour center to be directly recognized and used by the robotic arm, thus establishing a unified coordinate reference for subsequent calculations of the three-dimensional position of the target picking point.

[0133] Step S503: Determine the picking offset distance from the chrysanthemum to be picked, and calculate the first product between the picking offset distance and the growth posture space vector. The picking offset distance represents the distance from the center of the outer contour of the chrysanthemum to be picked to the optimal cutting position of the flower stem below the flower stamen along the direction from the center of the outer contour to the center of the flower stamen depression.

[0134] The picking offset distance is a fixed value preset according to the growth characteristics of Chrysanthemum morifolium and the requirements of the robotic arm operation. It represents the distance between the center of the outer contour of Chrysanthemum morifolium and the optimal cutting position of the flower stem below the stamen, along the direction from the center of the outer contour to the center of the flower stamen indentation (the direction of growth posture). Subsequently, this picking offset distance is multiplied with the growth posture space vector constructed in step S501. The first product is essentially a space vector along the flower growth posture direction with a length equal to the picking offset distance. Its function is to clarify the direction and distance scale of movement from the center of the outer contour to the optimal picking point.

[0135] Step S504: Calculate and determine the first sum between the first product and the three-dimensional spatial coordinates of the outer contour center, and use the first sum as the target picking point of the golden chrysanthemum to be picked, so as to complete the picking point positioning of the golden chrysanthemum.

[0136] The first product obtained in step S503 is summed with the three-dimensional spatial coordinates of the outer contour center obtained in step S502. The resulting first sum is the target picking point of the golden chrysanthemum to be picked. Specifically, starting from the outer contour center in the robot's working coordinate system, the robot moves a preset picking offset distance along the direction of the flower's growth posture. The final position reached is the best picking point that is both far away from the fragile petals and can accurately cut the thick flower stem, thus completing the three-dimensional spatial positioning of the picking point and providing precise position instructions for the robot to perform non-destructive picking actions.

[0137] As can be seen from the above embodiments, compared with the prior art, the present application addresses the problems of existing flexible cable-driven parallel robots where the robotic arm directly grips the fragile flower stamen, causing the stamen to tear and fall off, and the robotic arm can only perform gripping or cutting actions at a fixed angle. This rigid working mode leads to uneven force on the flower, especially when the stamen is offset or the flower is tilted, which can easily cause problems such as stamen breakage and petal tearing. The present application has, but is not limited to, the following beneficial effects:

[0138] Firstly, addressing the technical shortcomings of existing cable-driven parallel robots in accurately locating the picking point of Chrysanthemum morifolium, which easily damages the stamens or leaves excessively long stems, this application uses a U2Net model to precisely extract the outer edge of the flower and the concave area of ​​the stamen. By fitting the minimum circumcircle to determine the center of the outer contour and the center of the concave area, the offset vector coordinates are determined. The growth orientation angle is calculated based on the offset vector coordinates. Then, based on the growth orientation angle and the growth tilt angle, the growth posture space vector of the Chrysanthemum morifolium to be picked is constructed. Finally, the optimal cutting position of the stem is locked. This application can avoid the robot arm directly gripping the fragile stamens of Chrysanthemum morifolium, preventing the stamens from tearing and falling off, and ensuring that the cutting position is far away from the stamens and close to the base of the stem, effectively controlling the residual length of the stem and ensuring the integrity of the flower and the quality of the product.

[0139] Secondly, addressing the issue of existing cable-driven parallel robots easily misjudging targets in scenarios with multiple adjacent flowers and occluded leaves, this application employs the U2Net model to achieve pixel-level precise segmentation of the outer edge of the flower and the concave area of ​​the stamen. This effectively separates a single flower from the background and adjacent flowers, avoiding the situation where multiple targets are misjudged as a single target. It ensures that the robotic arm focuses only on a single flower to be picked and will not touch adjacent flowers or mixed leaves at the same time. From the positioning perspective, it eliminates damage such as stamen compression deformation and petal breakage, improving the stability of batch picking.

[0140] Thirdly, addressing the shortcomings of existing cable-driven parallel robots, such as lack of posture adaptation and uneven force on flowers due to fixed-angle operation, this application calculates the growth orientation angle using an offset vector and quantifies the growth tilt angle by combining the ratio of the offset vector's magnitude to the outer contour radius. This constructs a growth posture space vector that characterizes the three-dimensional spatial posture of the flower. This allows the robot arm to adapt to the flower's planar orientation deflection and three-dimensional tilt state, performing clamping and cutting actions along the flower's natural growth posture. This replaces the traditional rigid fixed-angle operation mode, avoiding problems such as flower stamen breakage and petal tearing caused by uneven force on the flower, and adapting to the complex scenario of random flower growth in the field.

[0141] Fourth, this application eliminates the need for manual intervention in image segmentation and harvesting point localization, avoiding the inefficiency and high cost of manual harvesting while improving harvesting point localization accuracy to the millimeter level, significantly reducing the harvesting damage rate of Chrysanthemum morifolium. Simultaneously, its anti-interference segmentation capability and posture adaptive design can adapt to complex field scenarios such as dense growth of multiple flowers and leaf occlusion, meeting the high-efficiency harvesting needs of large-scale planting, contributing to the automation upgrade of the Chrysanthemum morifolium industry, and ensuring the commercial quality and planting returns of this high-value crop.

[0142] Please see Figure 9A device for locating the picking point of Chrysanthemum morifolium, provided to meet one of the purposes of this application, includes an image segmentation module 1100, a center determination module 1200, a growth orientation angle determination module 1300, a growth tilt angle determination module 1400, and a picking point positioning module 1500. The image segmentation module 1100 is configured to acquire images of Chrysanthemum morifolium containing one or more Chrysanthemum morifolium to be picked, input the images into an image segmentation model trained to convergence, and extract a first segmentation image corresponding to the outer edge region of the flower and a second segmentation image corresponding to the concave region of the stamen from the images. The center determination module 1200 is configured to calculate the first center of the first minimum circumscribed circle surrounding all first pixels of the outer edge region of the flower based on the first segmentation image to determine the outer contour center of the Chrysanthemum morifolium to be picked; and calculate the second center of the second minimum circumscribed circle surrounding all second pixels of the concave region of the stamen based on the second segmentation image to determine the concave center of the Chrysanthemum morifolium to be picked. The growth orientation angle determination module 1300... 0, configured to calculate and determine the offset vector coordinates of the outer contour center pointing to the concave center, and use the arctangent function to calculate and determine the growth orientation angle of the chrysanthemum to be picked in the chrysanthemum image plane based on the offset vector coordinates; growth tilt angle determination module 1400, configured to call a preset calibration model to determine the growth tilt angle of the chrysanthemum based on the ratio between the magnitude of the offset vector and the outer contour radius of the chrysanthemum to be picked; picking point positioning module 1500, configured to construct the growth posture space vector of the chrysanthemum to be picked based on the growth orientation angle and the growth tilt angle, and determine the target picking point of the chrysanthemum to be picked based on the growth posture space vector and the three-dimensional space coordinates of the outer contour center, so as to complete the picking point positioning of the chrysanthemum.

[0143] Based on any embodiment of this application, please refer to Figure 10 Another embodiment of this application also provides an electronic device, which can be implemented by a computer device, such as... Figure 10The diagram shows the internal structure of a computer device. The computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected via a system bus. The computer-readable storage medium stores an operating system, a database, and computer-readable instructions. The database may store control information sequences. When the computer-readable instructions are executed by the processor, the processor can implement a method for locating the picking point of chrysanthemum. The processor of the computer device provides computing and control capabilities to support the operation of the entire computer device. The memory of the computer device may store computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor can execute the method for locating the picking point of chrysanthemum according to this application. The network interface of the computer device is used for communication with a terminal. Those skilled in the art will understand that… Figure 10 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0144] In this embodiment, the processor is used to execute... Figure 9 The memory stores the specific functions of each module, and stores the program code and various data required to execute the above modules. The network interface is used for data transmission between the user terminal and the server. In this embodiment, the memory stores the program code and data required to execute all modules in the Chrysanthemum Picking Point Positioning Device of this application, and the server can call the server's program code and data to execute the functions of all modules.

[0145] This application also provides a storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the method for locating the picking point of Chrysanthemum morifolium as described in any embodiment of this application.

[0146] This application also provides a computer program product, including a computer program / instructions, which, when executed by one or more processors, implement the steps of the method for locating the picking point of Chrysanthemum morifolium as described in any embodiment of this application.

[0147] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. This computer program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. The aforementioned storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0148] The above description is only a partial embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for locating the picking point of Chrysanthemum morifolium, characterized in that, include: Acquire images of Chrysanthemum morifolium containing one or more Chrysanthemum morifolium images to be picked, input the Chrysanthemum morifolium images into an image segmentation model that has been trained to convergence, and extract the first segmentation image corresponding to the outer edge region of the flower of the Chrysanthemum morifolium to be picked and the second segmentation image corresponding to the concave region of the stamen from the Chrysanthemum morifolium images. The first center of the first minimum circumscribed circle surrounding all first pixels in the outer edge region of the flower is calculated based on the first segmentation image to determine the outer contour center of the chrysanthemum to be picked; the second center of the second minimum circumscribed circle surrounding all second pixels in the stamen concave region is calculated based on the second segmentation image to determine the concave center of the chrysanthemum to be picked. The offset vector coordinates pointing from the center of the outer contour to the center of the depression are calculated and determined, and the arctangent function is used to calculate the growth orientation angle of the chrysanthemum to be picked in the chrysanthemum image plane based on the offset vector coordinates. The growth tilt angle of the golden chrysanthemum is determined by calling a preset calibration model based on the ratio between the magnitude of the offset vector and the outer contour radius of the golden chrysanthemum to be harvested. Based on the growth orientation angle and the growth tilt angle, a growth posture space vector of the chrysanthemum to be harvested is constructed. Based on the growth posture space vector and the three-dimensional spatial coordinates of the outer contour center, the target harvesting point of the chrysanthemum to be harvested is determined to complete the harvesting point positioning of the chrysanthemum.

2. The method for locating the picking point of Chrysanthemum morifolium according to claim 1, characterized in that, The steps of inputting the image of the golden chrysanthemum into an image segmentation model that has been trained to convergence, and extracting the first segmentation image corresponding to the outer edge region of the flower of the golden chrysanthemum to be picked and the second segmentation image corresponding to the concave region of the stamen from the image of the golden chrysanthemum include: The image segmentation model, which has been trained to convergence, performs multi-level encoding to downsample the original image of the Chrysanthemum morifolium and produces intermediate feature information corresponding to each scale. The intermediate feature information represents the contour features of the outer edge region of the flower and the contour features of the concave region of the stamen in the Chrysanthemum morifolium image. The image segmentation model, which has been trained to convergence, performs multi-level decoding. Based on the intermediate feature information at the smallest scale, it performs upsampling operation and successively jumps and fuses the intermediate feature information produced by the same level encoding as a reference to decode and generate higher-scale image feature information. The image feature information is used to represent the contour features of the outer edge region of the flower and the concave region of the stamen in the form of a mask. The image segmentation model performs image segmentation on the original image of the Chrysanthemum morifolium based on the mask image data fused from all image feature information, and extracts the first segmentation image corresponding to the outer edge region of the flower and the second segmentation image corresponding to the concave region of the stamen from the Chrysanthemum morifolium image.

3. The method for locating the picking point of Chrysanthemum morifolium according to claim 1, characterized in that, The steps of calculating the first center of the first minimum circumscribed circle surrounding all first pixels in the outer edge region of the flower according to the first segmentation image to determine the outer contour center of the chrysanthemum to be picked; and calculating the second center of the second minimum circumscribed circle surrounding all second pixels in the stamen depression region according to the second segmentation image to determine the depression center of the chrysanthemum to be picked include: Extract all first pixels belonging to the outer edge region of the flower from the first segmentation image corresponding to the outer edge region of the flower, so as to obtain the first two-dimensional pixel coordinates of each first pixel in the image of the golden chrysanthemum; Extract all second pixels belonging to the flower stamen concave region from the second segmentation image corresponding to the flower stamen concave region, so as to obtain the second two-dimensional pixel coordinates of each second pixel in the golden chrysanthemum image; The minimum circumcircle algorithm is used to fit the smallest area and completely enclose all the first pixels based on the coordinates of all the first two-dimensional pixels. The first center of the first minimum circumcircle is determined as the outer contour center of the chrysanthemum to be picked, and the first radius of the first minimum circumcircle is determined as the outer contour radius of the chrysanthemum to be picked. The minimum circumcircle algorithm is used to fit the second minimum circumcircle with the smallest area that completely surrounds all the second pixels based on the coordinates of all the second two-dimensional pixels. The center of the second circle corresponding to the second minimum circumcircle is determined as the concave center of the golden chrysanthemum to be picked.

4. The method for locating the picking point of Chrysanthemum morifolium according to claim 1, characterized in that, The steps of calculating and determining the offset vector coordinates from the center of the outer contour to the center of the depression, and using the arctangent function to calculate and determine the growth orientation angle of the chrysanthemum to be harvested in the chrysanthemum image plane based on the offset vector coordinates, include: Obtain the third two-dimensional pixel coordinates corresponding to the outer contour center of the chrysanthemum to be picked and the fourth two-dimensional pixel coordinates corresponding to the concave center; The offset vector coordinates pointing from the outer contour center to the concave center are determined based on the difference between the third two-dimensional pixel coordinates corresponding to the outer contour center and the fourth two-dimensional pixel coordinates corresponding to the concave center. Using a preset arctangent function, the angle value of the offset vector in the image plane of the golden chrysanthemum is calculated based on the coordinates of the offset vector, and the angle value is determined as the growth orientation angle of the golden chrysanthemum to be harvested in the image plane of the golden chrysanthemum.

5. The method for locating the picking point of Chrysanthemum morifolium according to claim 3, characterized in that, The step of determining the growth tilt angle of the golden chrysanthemum by calling a preset calibration model and using the ratio between the magnitude of the offset vector and the outer contour radius of the golden chrysanthemum to be harvested includes: The magnitude of the offset vector from the center of the outer contour to the center of the recess is calculated based on the coordinates of the offset vector pointing from the center of the outer contour to the center of the recess. The ratio between the magnitude of the offset vector and the outer contour radius of the chrysanthemum to be harvested is calculated and input into a preset calibration model to output the growth tilt angle of the chrysanthemum to be harvested. The calibration model is a linear fitting model, which is based on multiple sets of chrysanthemum samples with known growth tilt angles. It is obtained by fitting the correspondence between the ratio of the magnitude of the offset vector pointing from the center of the outer contour to the center of the depression and the outer contour radius and the growth tilt angle.

6. The method for locating the picking point of Chrysanthemum morifolium according to claim 1, characterized in that, The steps of constructing a growth posture space vector of the chrysanthemum to be harvested based on the growth orientation angle and the growth tilt angle, and determining the target harvesting point of the chrysanthemum based on the growth posture space vector and the three-dimensional spatial coordinates of the outer contour center, include: Based on the coordinate system of the robotic arm, and based on the mapping relationship between spherical coordinates and Cartesian coordinates, a growth posture space vector representing the spatial growth posture of the chrysanthemum to be harvested is constructed according to the growth orientation angle and growth tilt angle of the chrysanthemum to be harvested. By using preset camera-robotic hand-eye calibration parameters, the two-dimensional pixel coordinates of the outer contour center of the chrysanthemum to be picked are converted into three-dimensional spatial coordinates of the outer contour center under the robotic arm's working coordinate system. Determine the picking offset distance from the chrysanthemum to be picked, and calculate the first product between the picking offset distance and the growth posture space vector. The picking offset distance represents the distance from the center of the outer contour of the chrysanthemum to be picked to the optimal cutting position of the flower stem below the flower stamen along the direction from the center of the outer contour to the center of the flower stamen depression. Calculate and determine the first sum between the first product and the three-dimensional spatial coordinates of the outer contour center, and use the first sum as the target picking point of the golden chrysanthemum to be picked, so as to complete the picking point positioning of the golden chrysanthemum.

7. The method for locating the picking point of Chrysanthemum morifolium according to any one of claims 1 to 6, characterized in that, The basic network architecture of the image segmentation model includes the U2net model; The growth orientation angle is the horizontal deflection angle of the stamen relative to the center of the outer contour of the golden chrysanthemum to be harvested in the two-dimensional plane of the image, which represents the orientation of the flower in the plane. The growth tilt angle is the tilt angle of the golden chrysanthemum to be harvested relative to the vertical direction in three-dimensional space, which represents the degree of tilt of the flower's posture in space.

8. A device for positioning the picking point of Chrysanthemum morifolium, characterized in that, include: The image segmentation module is configured to acquire images of Chrysanthemum morifolium containing one or more Chrysanthemum morifolium images to be picked, input the Chrysanthemum morifolium images into an image segmentation model that has been trained to convergence, and extract a first segmentation image corresponding to the outer edge region of the flower of the Chrysanthemum morifolium to be picked and a second segmentation image corresponding to the concave region of the stamen from the Chrysanthemum morifolium images. The center determination module is configured to calculate the center of the first minimum circumscribed circle of the first minimum circumscribed circle of the flower's outer edge region based on the first segmentation image, so as to determine the outer contour center of the chrysanthemum to be picked; and to calculate the center of the second minimum circumscribed circle of the second minimum circumscribed circle of the second minimum circumscribed circle of the flower's stamen depression region based on the second segmentation image, so as to determine the depression center of the chrysanthemum to be picked. The growth orientation angle determination module is configured to calculate and determine the offset vector coordinates of the outer contour center pointing to the center of the depression, and to use the arctangent function to calculate and determine the growth orientation angle of the chrysanthemum to be picked in the chrysanthemum image plane based on the offset vector coordinates; The growth tilt angle determination module is configured to call a preset calibration model to determine the growth tilt angle of the golden chrysanthemum based on the ratio between the magnitude of the offset vector and the outer contour radius of the golden chrysanthemum to be harvested. The picking point positioning module is configured to construct a growth posture space vector of the chrysanthemum to be picked based on the growth orientation angle and the growth tilt angle, and determine the target picking point of the chrysanthemum to be picked based on the growth posture space vector and the three-dimensional spatial coordinates of the outer contour center, so as to complete the picking point positioning of the chrysanthemum.

9. An electronic device comprising a central processing unit and a memory, characterized in that, The central processing unit is used to invoke and run a computer program stored in the memory to perform the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores, in the form of computer-readable instructions, a computer program implemented according to any one of claims 1 to 7, which, when invoked by a computer, executes the steps included in the corresponding method.

Images