Szechuan lovage rhizome and poria cocos seed production system based on machine vision

Abstract

The invention discloses a ligusticum wallichii and poria cocos seed production system based on machine vision, and relates to the field of poria cocos seed intelligent seed production. The system comprises a feeding unit, a first transmission unit, a cutting unit, an image acquisition unit, a first recognition unit, an appearance detection unit, a first screening unit, a first collection unit, a smell detection unit, a second screening unit, an internal detection unit and a third screening unit. Through cooperation of the units, the quality of the poria cocos seeds can be automatically and accurately judged, accurate screening of poria cocos seed insect pests or diseases is achieved, and the system can accurately obtain healthy poria cocos seeds.

Description

Technical Field

[0001] The present invention relates to the field of intelligent seed production of Ling seeds, and in particular to a machine vision-based system for producing seeds of Chuanxiong Ling seeds. Background Art

[0002] Chuanxiong is a traditional Chinese medicine. Its propagation usually uses the Lingzhong, which is the section of the stem of Chuanxiong. Figure 1 As shown, Figure 1 The middle 1 is the ling stem, on which there are multiple ling seeds 2. The ling seeds have roots and tooth germs. The ling seeds need to be cut from the ling stem to obtain the corresponding ling seeds.

[0003] The current method of producing Chuanxiong Ling seeds is to obtain the Ling stems, place the Ling stems on a cutting machine for cutting to obtain Ling seeds, then pack the Ling seeds, and then plant the packaged Ling seeds.

[0004] In the existing method, all the Ling seeds on the Ling stem are cut off and packaged as Ling seeds for subsequent sowing during seed production. The current method for producing Chuanxiong Ling seeds has the following problems: 1. The Ling seeds on the Ling stem may include those with insect pests or diseases. These Ling seeds are likely to fail to grow normally in the future, which will affect the subsequent yield; 2. Cut all the ling seeds on the ling stalk and pack them together so that different types of ling seeds are mixed together. In this way, different types of ling seeds will be planted in the same area during sowing, resulting in uneven growth of the ling stalks in the same planting area, different nutritional requirements, and different management. Planting management is difficult, and if it is not managed separately, the yield will be affected; 3. Currently, some methods will check whether the Ling seeds have pests or diseases after cutting, but the inspection method is manual inspection, which is inefficient and prone to errors. The inspection is only an appearance inspection, and it is not possible to conduct a comprehensive and accurate inspection of the pests and diseases of the Ling seeds. Summary of the invention

[0005] The purpose of the present invention is to automatically distinguish different types of Ling seeds through intelligent means, and to accurately obtain healthy Ling seeds.

[0006] In order to achieve the above-mentioned invention object, the present invention provides a Chuanxiong Ling seed production system based on machine vision, the system comprising: A feeding unit, used for feeding the ling rods to the first transmission unit; A first transmission unit, used for transmitting the ling stem to the cutting unit; A cutting unit, used for cutting the ling stems to obtain ling seeds; An image acquisition unit, used for acquiring images of the species; The first recognition unit is used to classify and identify the types of Ling on the Ling stem before cutting, and obtain the type detection result of the Ling; An appearance inspection unit is used to perform appearance defect inspection on the ling seeds based on the ling seeds image to obtain appearance defect inspection results; A first screening unit is used to transmit the Ling species with qualified appearance defect detection results to the odor detection unit, and transmit the Ling species with unqualified appearance defect detection results to the first collection unit; An odor detection unit is used to detect the odor of the Ling seeds and obtain the odor detection result; A second screening unit is used to transmit the Ling species with qualified smell detection results to the internal detection unit, and transmit the Ling species with unqualified smell detection results to the first collection unit; An internal detection unit, used for performing internal corruption detection on the Ling seeds and obtaining internal corruption detection results; The third screening unit is used to sort the Ling seeds that are qualified in the internal corruption detection result into Ling seed storage boxes corresponding to different categories according to the Ling seed category detection result, and transmit the Ling seeds that are unqualified in the internal corruption detection result to the first collection unit.

[0007] Among them, the production method of Chuanxiong Ling seeds in the present system is: Chuanxiong is loaded to the first transmission unit through the loading unit; then the Ling stem is transmitted to the cutting unit by the first transmission unit; in the process of transmission, the Ling seeds on the Ling stem are classified and identified by the first identification unit before cutting to obtain the category detection result of the Ling seeds. It should be emphasized that in the present invention, the Ling seeds are classified before cutting, and the purpose is to accurately classify the Ling seeds. The classification of the Ling seeds is related to the size of the Ling seeds and the position on the Ling stem. If the size of the Ling seeds does not change after cutting, its position on the Ling stem cannot be obtained, and therefore the Ling seeds cannot be accurately classified. The present invention can accurately classify the Ling seeds through the above method, and the purpose of classification is to sort the Ling seeds into the Ling seed storage boxes corresponding to different categories in the future, so that the same type of Ling seeds are planted in the same area, and different Ling seeds are planted in different areas. The growth conditions of the same type are not much different, and there is no need to distinguish and manage them, which reduces the difficulty of planting management, facilitates unified management, and improves efficiency.

[0008] After the cutting unit cuts the ling stem to obtain the ling seeds, the key link in this system can be carried out later, that is, automatically and accurately judging the quality of the ling seeds, realizing accurate screening of ling seeds pests or diseases, so that this system can accurately obtain healthy ling seeds. The specific method is as follows: The first inspection is the appearance inspection, in which the appearance defect inspection of the ling seeds is performed on the ling seeds based on the ling seeds image by using the appearance inspection unit to obtain the appearance defect inspection result; the unhealthy ling seeds with abnormal appearance can be detected by the appearance inspection, but some ling seeds have normal appearance, but mold or rot has occurred inside, which cannot be detected by the appearance inspection, and such abnormalities usually volatilize gases with corresponding odors, so it is possible to further judge whether the ling seeds are healthy through odor detection, that is, the ling seeds are subjected to odor detection by the odor detection unit to obtain the odor detection result, and the health of the ling seeds is further judged based on the odor detection result. The applicant further studied and found If the epidermis of the Ling seeds is relatively intact, or the location of the mold or rot is deeper, the odor emitted by the mold or rot of the Ling seeds will be less, and some internal abnormalities such as empty bags or hollow drums will not produce gases with corresponding odors. These cases are not easy to be detected by the odor detection unit. Therefore, in order to further improve the accuracy of whether the Ling seeds are healthy, after the Ling seeds pass the appearance and odor inspections, the system also needs to perform internal inspections. The internal inspections can be used to accurately detect the internal corruption of the Ling seeds and obtain internal corruption detection results. After the above-mentioned appearance inspections, odor inspections and internal inspections, the Ling seeds that pass the final inspection can have a higher health accuracy rate.

[0009] Furthermore, the odor detection unit includes: A first gu ling seed obtaining unit, used to obtain a number of first gu ling seeds that can volatilize corresponding organic compounds due to the influence of diseases or insect pests, and to obtain a number of normal second gu ling seeds; An infrared spectrum analysis unit, used for performing infrared spectrum analysis on the first and second Ling species to obtain first and second infrared spectrum analysis results respectively; A first training sample obtaining unit, used for annotating the first and second infrared spectrum analysis results to obtain a first training sample; an odor detection model obtaining unit, configured to obtain an odor detection model based on the first training sample training model; Among them, by constructing several first Ling species that can volatilize corresponding organic compounds due to the influence of diseases or pests and several normal second Ling species, and then constructing a first training sample that can reflect normal Ling species and Ling species with peculiar smell, the model is used to extract the features of the first training sample for learning and training, and then obtaining an odor detection model that can detect Ling species with peculiar smell.

[0010] Among them, detecting changes in the surface or internal chemical composition of the Ling seeds through infrared spectroscopy can indirectly reflect whether the smell is abnormal. This detection method is accurate, non-destructive, and easy to operate.

[0011] The odor detection unit detects the odor of the species to be detected in the following manner: Using an infrared spectrum analysis unit to detect the smell of the tested Poria species to obtain an infrared spectrum analysis result of the tested Poria species; The infrared spectrum analysis result of the to-be-detected Ling species is input into the odor detection model, and the odor detection model outputs whether the odor of the to-be-detected Ling species is qualified or unqualified.

[0012] Wherein, the appearance detection unit comprises: A second ling species obtaining unit, used for obtaining a plurality of third ling species with diseases, and used for obtaining a plurality of fourth ling species with insect pests; A feature extraction unit, used for extracting disease features from the image of the third type of ling, and for extracting insect pest features from the image of the fourth type of ling; A second training sample obtaining unit is used to mark all disease characteristics and insect pest characteristics to obtain a second training sample; an appearance detection model obtaining unit, configured to obtain an appearance detection model based on the second training sample training model; Among them, there are differences in color between healthy Ling seeds and abnormal Ling seeds. For example, the surface color of healthy Ling seeds is uniform, while signs of Ling seeds disease will appear black spots, brown spots, mildew (green, gray, black mold), yellowing or large-area discoloration. Healthy Ling seeds have no holes, while signs of Ling seeds insect pests are holes, moth marks or insect eggs attached to the surface. Therefore, by capturing the characteristic manifestations of these pests and diseases, these features are used to construct a second training sample. The second training sample training model can obtain an appearance detection model with the ability to detect the appearance characteristics of Ling seeds pests and diseases. The appearance detection model can detect Ling seeds pests and diseases from the appearance.

[0013] The appearance inspection unit performs appearance defect inspection on the inspected species in the following manner: Acquire and obtain images of the species to be detected; The image of the ling species to be detected is input into the appearance detection model, and the appearance detection model outputs the appearance pest and disease detection results of the ling species. If the appearance pest and disease detection results include pests or diseases, the appearance defect detection result is unqualified; if the appearance pest and disease detection results do not include pests and diseases, the appearance defect detection result is qualified.

[0014] Furthermore, the internal detection unit includes: A third ling species obtaining unit, used to obtain a plurality of fifth ling species having a corruption region therein; Spectral camera, used to collect the spectral characteristics of the fifth species of Ling; A third training sample obtaining unit is used to annotate the spectral features of the fifth Ling species to obtain a third training sample; An internal corruption detection model obtaining unit, configured to obtain an internal corruption detection model based on the third training sample training model; Among them, the internal detection unit conducts internal corruption detection on Ling seeds in the following ways: Use a spectral camera to obtain the spectral characteristics of the species to be detected; The spectral characteristics of the to-be-detected corm species are input into the internal corruption detection model to obtain the internal corruption detection results of the to-be-detected corm species.

[0015] Among them, the principle of the internal detection unit is to use light of different bands (visible light, near infrared, short-wave infrared) to penetrate the surface of the Ling seeds and capture the spectral characteristics of the internal chemical components (such as water, starch, and corruption metabolites), so as to determine whether there is any abnormality inside the Ling seeds. In order to realize automatic and accurate detection of whether the internal part of the Ling seeds is abnormal, this system first finds the Ling seeds with internal corruption areas, and then uses a spectral camera to collect the spectral characteristics of these Ling seeds. These spectral characteristics reflect the internal corruption situation, and then use these characteristics to construct training samples to train an internal corruption detection model with the ability to identify internal corruption. The internal corruption detection model can be used to detect internal corruption of the Ling seeds.

[0016] Furthermore, the first identification unit classifies and identifies the types of soya on the soya stem in the following manner: Taking pictures to obtain the image of the ling rod; Extract from the ling stem to obtain the size of each ling seed, the size of the ling stem connected to the ling seed, and the position information of the ling seed on the ling stem; Based on the size of each type of Ling, the size of the Ling stem connected to the type of Ling and the position information of the type of Ling on the Ling stem, the classification information of each type of Ling is obtained.

[0017] Among them, the above method can automatically and accurately realize the classification of Ling species, and whether the classification of Chuanxiong Ling species is accurate will directly affect the planting effect. Different types of Ling species have different planting strategies, and the yield of Chuanxiong can be indirectly guaranteed by accurately classifying Ling species.

[0018] Furthermore, the cutting unit comprises: A second identification unit, a first clamping portion, a second clamping portion, a plurality of third clamping portions, a controller, a processor, and a plurality of cutting modules; The second recognition unit is used to capture a first overall image of the ling stalk to be cut, recognize the ling stalk head, ling stalk tail and ling stalk species in the first overall image, and obtain position information of the ling stalk head, position information of the ling stalk tail and the number M of ling stalk species; The controller is used to control the first clamping part and the second clamping part to clamp and fix the head and tail of the ling stem to be cut respectively based on the position information of the head and tail of the ling stem, so that the ling stem is suspended and fixed above all cutting modules; The processor is used to calculate the number N of third clamping parts involved in clamping and fixing the ling stems to be cut based on the number M of ling species, where M and N are both integers greater than 1; After the ling stem is suspended and fixed above all cutting modules, the second identification unit is used to capture a second overall image of the ling stem to be cut, identify the ling species in the second overall image, and obtain the position information of each ling species; The controller is used to control the third clamping part to clamp the middle part of the shaft between two adjacent types of ling based on the position information of each type of ling, and one third clamping part corresponds to the middle part between two adjacent types of ling; After the N third clamping parts have completed clamping, the controller is used to control the cutting module to cut the shafts on both sides of the Ling seeds along a cutting path from bottom to top to obtain the Ling seeds.

[0019] Among them, through the study of the structure of the ling stem and the ling seeds, it was found that the ling stem is narrower than the ling seeds, and there are multiple ling seeds distributed on the ling stem. According to the traditional flat-push progressive cutting method, one end of the ling stem is placed in the cutting machine, and after cutting the next ling seed, it is pushed forward for a distance and then the next ling seed is cut. Each ling seed needs to be cut twice, once on the ling stem on its left and right sides. This results in a ling stem needing to be cut several times and moved forward several times, and the cutting efficiency is low. In addition, since the ling seed is larger than the rod body of the ling stem, the ling seed will be squeezed or rubbed against the cutting table during the cutting and moving process, so that some of the ling seeds will be damaged during the cutting process. In addition, the root seedlings and tooth germs of the ling seeds themselves are easily damaged, so the above cutting method will also have the problem of easily damaging the ling seeds.

[0020] Therefore, in order to solve the problems of low cutting efficiency and easy damage of ling seeds in traditional ling seed cutting methods, this system has improved the cutting unit. The core is that multiple cutting modules can perform cutting separately, and the entire ling rod can obtain all ling seeds in one cutting time. The cutting efficiency is high, and the ling seeds are suspended in the air during cutting and have no contact with the cutting table or other surfaces. The ling seeds can be well protected and the problem of damage to ling seeds during traditional cutting can be solved.

[0021] Through the applicant's research on the structure of the ling stem and the ling seeds, it was found that the target of cutting is the ling seeds. Therefore, the cutting strategy is to use the first clamping part and the second clamping part to clamp and fix the head and tail of the ling stem to be cut, so that the ling seeds are suspended above the cutting module. Then, the number of ling seeds is identified and the corresponding number of the third clamping parts is calculated. The third clamping part is used to clamp the middle part of the stem between two adjacent ling seeds. The purpose of this is that there will be a cutting component on each side of the ling seed to cut during cutting, so that the ling seeds are separated from the stem. Then, the separated ling seeds are collected and the unnecessary stems are still in a clamping state, so as to avoid the mixing of the ling seeds and the stem after cutting and the need for secondary screening, thereby saving the process. During cutting, the cutting module cuts the stems on both sides of the suspended ling seeds along the cutting path from bottom to top, thereby achieving low-damage cutting of the ling seeds and protecting the ling seeds. During the cutting process, the ling seeds have no contact with the table top and other surfaces, thereby ensuring the integrity of the root system and tooth germ on the ling seeds.

[0022] Furthermore, after all the Ling seeds are cut, the controller is also used to control the first clamping part, the second clamping part and the third clamping part to move the clamped object to above the first collection unit, and then control the first clamping part, the second clamping part and the third clamping part to end clamping, release the clamped object into the first collection unit, and then control the first clamping part, the second clamping part and the third clamping part to return to their respective initial positions.

[0023] After cutting is completed, each clamping part moves the object it clamps to the first collection unit, achieving a one-time recovery of unnecessary parts during cutting, avoiding mixing with the ling seeds, and avoiding multiple cleaning and sorting of unnecessary ling stem parts produced during cutting.

[0024] And the third clamping part is controlled to be clamped in the middle of the shaft between two adjacent ling seeds. The purpose of this design is that the part of the ling seeds that is usually cut is the shaft part close to the ling seeds. Clamping the third clamping part in the middle of the shaft can ensure a corresponding safety distance between the third clamping part and the cutter.

[0025] Preferably, the cutting module comprises: A robotic arm, a first cutter, a second cutter and a collection box; Wherein, the robotic arm, the first cutter and the second cutter are all connected to the controller signal, the front end of the robotic arm is fixedly connected to the mounting end at the bottom of the collection box, the upper part of the collection box is open and used to collect the ling seeds obtained by cutting, the first cutter and the second cutter are fixedly connected to both sides of the collection box, respectively, and the cutting parts of the first cutter and the second cutter extend to the top of the collection box respectively to cut the shafts on both sides of the ling seeds to obtain the ling seeds; wherein the cutting method of the cutting module is as follows: the controller controls the robotic arm to carry the collection box and move from bottom to top toward the target ling seeds, and adjusts the position of the collection box after moving to a preset distance from the target ling seeds, so that the center of the collection box is aligned with the center of the ling seeds, and then the controller turns on the first cutter and the second cutter, and then controls the robotic arm to carry the first cutter and the second cutter to cut the shafts on both sides of the target ling seeds and collect the target ling seeds into the collection box.

[0026] The present invention uses an intelligent mechanical arm to achieve flexible movement, conveniently carry the first cutter and the second cutter to move, realize the alignment and cutting of the cutting position, and use a collection box to collect the Ling seeds obtained after cutting. The specific cutting method is: After obtaining the position of the target ling seeds, the controller controls the mechanical arm to carry the collection box and move upward toward the target ling seeds. The purpose of cutting from bottom to top is to cut from bottom to top so that the target ling seeds and the ling stems fall freely into the collection box for automatic collection after being cut off. The purpose of aligning the center of the collection box with the center of the ling seeds is, on the one hand, to ensure that the ling seeds can be accurately collected by the collection box after cutting, and on the other hand, to align the center of the collection box with the center of the ling seeds so that the total length of each ling seed is equal to the width of the collection box, so that the length of each ling seed after cutting is equal, and then the size of the ling seeds after cutting is unified, which is convenient for subsequent unified packaging and packaging, and for subsequent unified sowing, which is conducive to uniform growth after sowing. Using the first cutter and the second cutter to cut the stems on both sides of the target ling seeds at the same time can make the ling seeds fall off quickly.

[0027] Wherein, a spray unit is provided inside the collection box for spraying a disinfectant solution on the cut Ling seeds. By spraying the disinfectant solution, the cut surface can be quickly sterilized to prevent mildew and improve the survival rate of the Ling seeds.

[0028] The first identification unit is also used to encode each type of ling to obtain the encoding result of each type of ling, and the appearance detection unit performs appearance defect detection on each type of ling in turn based on the encoding result of each type of ling. By identifying and detecting each type of ling one by one, it is easy to distinguish unqualified types of ling separately, and it can be avoided that multiple types of ling are mixed together in a large area, which makes it difficult to distinguish individual unqualified types of ling.

[0029] One or more technical solutions provided by the present invention have at least the following technical effects or advantages: The present invention can automatically distinguish and obtain different types of Ling seeds through intelligent means, and can accurately obtain healthy Ling seeds.

[0030] The invention can ensure that the qualified Ling species finally detected have a higher health accuracy rate through appearance detection, odor detection and internal detection.

[0031] The present invention can obtain all the ling seeds within one cutting time, with high cutting efficiency. During cutting, the ling seeds are suspended in the air and do not contact the cutting table or other surfaces, so the ling seeds can be well protected and the problem of damage to the ling seeds during cutting in the traditional method can be solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of the present invention, and do not constitute a limitation on the embodiments of the present invention; Figure 1 It is a schematic diagram of the structure of the chuanxiong Ling stem of the present invention; Figure 2 This is a schematic diagram of the composition of the Chuanxiong Ling seed production system based on machine vision; Figure 3 is a cutting schematic diagram of a cutting unit; Among them, 1-ling rod, 2-ling seeds, 3-first clamping part, 4-second clamping part, 5-third clamping part, 6-mechanical arm, 7-first cutter, 8-second cutter, 9-collection box. DETAILED DESCRIPTION

[0033] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present invention and the features in the embodiments can be combined with each other without conflict.

[0034] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those within the scope of this description. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0035] Those skilled in the art should understand that, in the disclosure of the present invention, the orientation or position relationship indicated by the terms "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc. are based on the orientation or position relationship shown in the drawings, which are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the above terms should not be understood as limiting the present invention.

[0036] It is to be understood that the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element may be one, while in another embodiment, the number of the element may be multiple, and the term "one" should not be understood as a limitation on the quantity.

[0037] Embodiment 1;

[0038] Please refer to Figure 2 , Figure 2 The present invention provides a Chuanxiong Lingzhi seed production system based on machine vision, and the system comprises: The loading unit is used to load the ling stalks to the first transmission unit; wherein the loading unit is a device in the prior art that can realize automatic loading of ling stalks, such as a storage bin and a conveying mechanism, wherein the storage bin design: inclined hopper: automatically arranges the ling stalks by gravity, suitable for short stalks (<1 meter); or, multi-layer vibrating rack: long stalks can be laid flat in layers and conveyed by motor-driven chains or rollers. Conveying mechanism: belt / roller conveyor: suitable for horizontal conveying of long stalks, and side baffles need to be installed to prevent deviation; or, robotic arm grasping: six-axis robotic arm with adaptive grippers (such as pneumatic V-type grippers), suitable for irregular or fragile materials; or, vacuum suction cup system: if the surface of the ling stalks is flat, vacuum adsorption can be used for transportation; The first transmission unit is used to transmit the ling stalks to the cutting unit. The first transmission unit can be an automatic transmission device or a transmission belt or other device with a transmission function, which is not limited accordingly in the embodiment of the present invention. The first transmission device can transmit the loaded ling stalks to the subsequent unit. A cutting unit, used for cutting the ling stems to obtain ling seeds; An image acquisition unit, used to acquire the image; the image acquisition unit may be an industrial camera or a video camera, etc., and the embodiment of the present invention does not limit it accordingly; if the ambient light does not meet the use requirements, it may also include a light source and other equipment; The first recognition unit is used to classify and identify the types of Ling on the Ling stem before cutting, and obtain the type detection result of the Ling; An appearance inspection unit is used to perform appearance defect inspection on the ling seeds based on the ling seeds image to obtain appearance defect inspection results; A first screening unit is used to transmit the Ling species with qualified appearance defect detection results to the odor detection unit, and transmit the Ling species with unqualified appearance defect detection results to the first collection unit; wherein the screening unit can be implemented by a robotic arm, such as the robotic arm grabbing and placing unqualified Ling species on a conveyor belt of the first collection unit, or can be a corresponding screening device, which is not limited accordingly in the embodiments of the present invention; An odor detection unit is used to detect the odor of the Ling seeds and obtain the odor detection result; The second screening unit is used to transmit the Ling species with qualified odor detection results to the internal detection unit, and transmit the Ling species with unqualified odor detection results to the first collection unit; the principle and implementation method of the second screening unit are similar to those of the first screening unit, and the corresponding description is not repeated in the embodiment of the present invention; An internal detection unit, used for performing internal corruption detection on the Ling seeds and obtaining internal corruption detection results; The third screening unit is used to sort the ling seeds that are qualified in the internal corruption detection results into ling seed storage boxes corresponding to different categories according to the category detection results of the ling seeds, and to transmit the ling seeds that are unqualified in the internal corruption detection results to the first collection unit. The principle and implementation method of the third screening unit are similar to those of the first screening unit, and will not be described in detail in the embodiment of the present invention.

[0039] Wherein, in the embodiment of the present invention, the odor detection unit includes: The first ling species obtaining unit is used to obtain a number of first ling species that can volatilize corresponding organic compounds due to diseases or insect pests, and to obtain a number of normal second ling species; in practical applications, ling species that have been affected by diseases or insect pests can be selected, and then tested to see whether they have emitted the corresponding smell. If they have emitted the smell, they are selected as the first ling species, and then the normal ling species are used as the second ling species; The infrared spectrum analysis unit is used to perform infrared spectrum analysis on the first and second Ling species to obtain the first and second infrared spectrum analysis results respectively; wherein, the normal smell of Ling species is: Chuanxiong Ling species has a slight spicy aroma (containing volatile oil); the abnormal smell of Ling species: corrupt sour smell, musty smell or rancid smell, which may be due to the influence of diseases or pests, such as bacterial or fungal infection, powdery mildew or storage mold, etc. The difference in the normal and abnormal smell of Ling species can be reflected in the infrared spectrum analysis results; wherein, the change of the chemical composition of the surface or inside of the Ling species can be detected by infrared spectroscopy, which can indirectly reflect the abnormal smell. It is suitable for rapid screening. It is a non-destructive test and easy to operate.

[0040] A first training sample obtaining unit, used for annotating the first and second infrared spectrum analysis results to obtain a first training sample; An odor detection model obtaining unit, used to obtain an odor detection model based on the first training sample training model; the model may adopt an AI model (such as a convolutional neural network CNN), etc., and the embodiment of the present invention does not make corresponding limitations; The odor detection unit detects the odor of the species to be detected in the following manner: Using an infrared spectrum analysis unit to detect the smell of the tested Poria species to obtain an infrared spectrum analysis result of the tested Poria species; The infrared spectrum analysis result of the to-be-detected Ling species is input into the odor detection model, and the odor detection model outputs whether the odor of the to-be-detected Ling species is qualified or unqualified.

[0041] Wherein, in the embodiment of the present invention, the appearance detection unit includes: A second ling species obtaining unit, used for obtaining a plurality of third ling species with diseases, and used for obtaining a plurality of fourth ling species with insect pests; The feature extraction unit is used to extract disease features from the image of the third type of ling, and to extract pest features from the image of the fourth type of ling; wherein, healthy ling: the surface color is uniform (such as Poria cocos species are mostly milky white, light yellow or light brown). Disease signs: the appearance of black spots, brown spots, mildew (green, gray, black mold), yellowing or large-area discoloration. Pest signs: holes, moth marks or insect eggs attached to the surface. Therefore, the features of color, holes, moth marks and insect eggs can be used as key features for appearance detection; The second training sample obtaining unit is used to mark all disease features (color) and insect pest features (holes, borer marks and insect eggs) to obtain the second training sample; An appearance detection model obtaining unit, used to obtain an appearance detection model based on the second training sample training model; the model may adopt an AI model (such as a convolutional neural network CNN), etc., and the embodiment of the present invention does not make corresponding limitations; The appearance inspection unit performs appearance defect inspection on the inspected species in the following manner: Acquire and obtain images of the species to be detected; The image of the ling species to be detected is input into the appearance detection model, and the appearance detection model outputs the appearance pest and disease detection results of the ling species. If the appearance pest and disease detection results include pests or diseases, the appearance defect detection result is unqualified; if the appearance pest and disease detection results do not include pests and diseases, the appearance defect detection result is qualified.

[0042] Among them, in the embodiment of the present invention, since some Ling species have no abnormal appearance and no peculiar smell, but have been rotten inside, the abnormality cannot be detected by the appearance detection unit and the odor detection unit. Therefore, the present invention designs an internal detection unit, which includes: A third ling species obtaining unit, used to obtain a plurality of fifth ling species having a corruption region therein; Spectral camera, used to collect the spectral characteristics of the fifth type of Ling; the principle is: use different bands of light (visible light, near infrared, short-wave infrared) to penetrate the surface of Ling to capture the spectral characteristics of internal chemical components (such as water, starch, and spoilage metabolites); in practice, a portable hyperspectral camera (such as Specim IQ) can be used with a multi-band LED light source; A third training sample obtaining unit is used to annotate the spectral features of the fifth Ling species to obtain a third training sample; An internal corruption detection model acquisition unit is used to obtain an internal corruption detection model based on the third training sample training model; an AI model (such as a convolutional neural network CNN) can be used to train spectral features to identify corruption area features; Among them, the internal detection unit conducts internal corruption detection on Ling seeds in the following ways: Use a spectral camera to obtain the spectral characteristics of the species to be detected; The spectral characteristics of the to-be-detected corm species are input into the internal corruption detection model to obtain the internal corruption detection results of the to-be-detected corm species.

[0043] In the embodiment of the present invention, the first identification unit classifies and identifies the types of soya on the soya stem in the following manner: Taking pictures to obtain the image of the ling rod; Extract from the ling stem to obtain the size of each ling seed, the size of the ling stem connected to the ling seed, and the position information of the ling seed on the ling stem; Based on the size of each type of Ling, the size of the Ling stem connected to the type of Ling and the position information of the type of Ling on the Ling stem, the classification information of each type of Ling is obtained.

[0044] Among them, the embodiment of the present invention divides the Ling species into the following three categories: Category 1: The first type is the best quality type of Chuanxiong Lingzi, which refers to the Lingzi with thick stems connected to the Lingzi, full nodes and strong growth. This type of Lingzi has uniform seedlings and strong growth potential, and is the preferred material for planting. In production practice, the yield and survival rate of the first type are better than other types.

[0045] Category 2: Definition: The second category refers to the Ling species near the base of the stem. Its quality is inferior to the first category, but it still has a certain planting value. Its growth potential and emergence rate are slightly weaker than the first category, which is related to its proximity to the soil and susceptibility to pests and diseases or environmental influences.

[0046] The third category: The third category refers to the ling species with thinner stems and smaller ling nodes, which are of poorer quality. This type of ling species has thin stems, less nutrients, and weaker growth after germination, so it is usually necessary to plant it separately from other grades to ensure uniform field management.

[0047] The present invention adopts the above classification method to take into account three factors related to the growth and health of the Ling seeds, such as the thickness of the stems of the Ling rods connected to the Ling seeds. Thicker stems can provide better nutrition and growth and development conditions, and the size of the node disks of the Ling seeds can reflect the size of the Ling seeds, which affects the subsequent germination of tooth germs and thus affects their survival rate. The position of the Ling seeds on the Ling rods can reflect whether they are close to the ground and whether they are prone to insect pests or diseases. Therefore, by considering the above three points, the Ling seeds can be accurately classified from the perspective of affecting the growth and development of the Ling seeds.

[0048] Among them, please refer to Figure 3 , Figure 3 Schematic diagram of cutting of a cutting unit, wherein the cutting unit comprises: A second identification unit, a first clamping part 3, a second clamping part 4, a plurality of third clamping parts 5, a controller, a processor and a plurality of cutting modules; The second recognition unit is used to capture a first overall image of the ling stalk to be cut, recognize the ling stalk head, ling stalk tail and ling stalk species in the first overall image, and obtain position information of the ling stalk head, position information of the ling stalk tail and the number M of ling stalk species; The controller is used to control the first clamping part and the second clamping part to clamp and fix the head and tail of the ling stem to be cut respectively based on the position information of the head and tail of the ling stem, so that the ling stem is suspended and fixed above all cutting modules; The processor is used to calculate the number N of third clamping parts involved in clamping and fixing the ling stems to be cut based on the number M of ling species, where M and N are both integers greater than 1; After the ling stem is suspended and fixed above all cutting modules, the second identification unit is used to capture a second overall image of the ling stem to be cut, identify the ling species in the second overall image, and obtain the position information of each ling species; The controller is used to control the third clamping part to clamp the middle part of the shaft between two adjacent types of ling based on the position information of each type of ling, and one third clamping part corresponds to the middle part between two adjacent types of ling; After the N third clamping parts have completed clamping, the controller is used to control the cutting module to cut the shafts on both sides of the Ling seeds along a cutting path from bottom to top to obtain the Ling seeds.

[0049] The sizes of M and N are adjusted according to actual needs, and are not limited in the embodiment of the present invention. The calculation method between M and N is: N=M-1.

[0050] Among them, the second recognition unit includes a camera or a video camera and a corresponding processor, wherein the camera or the video camera is used to acquire images of the ling stems to be cut, and the processor is equipped with a corresponding recognition model that can identify the head, tail and type of the ling stems. The recognition model can be pre-trained and installed in the processor and run when used. The recognition model is obtained by obtaining images of several ling stems, and then marking the head, tail and type of the ling stems in the images, and then obtaining a training set based on the marked images, and then using the training set to train the model to obtain the recognition model.

[0051] Among them, the head and tail of the ling stem are located at the two ends of the ling stem respectively, and the tail of the ling stem is usually darker in color due to being in the soil for a long time, so it can be distinguished by color.

[0052] Among them, the first clamping part, the second clamping part and the third clamping part can be a robotic arm and the front end of the robotic arm is a mechanical claw. The clamping and releasing operations are realized by opening and closing the mechanical claw. They can also be a clamper, an end effector, a pneumatic clamp, an electric clamp and an adaptive clamp, etc. The embodiments of the present invention do not make corresponding limitations.

[0053] Among them, in the embodiment of the present invention, after cutting and obtaining all the Ling species, the controller is also used to control the first clamping part, the second clamping part and the third clamping part to move the clamped object to the top of the first collection unit, and then control the first clamping part, the second clamping part and the third clamping part to end clamping, release the clamped object to the first collection unit, and then control the first clamping part, the second clamping part and the third clamping part to return to their respective initial positions. The above method realizes the rapid collection and management of different parts after cutting.

[0054] Please refer to Figure 3 , the cutting module comprises: A mechanical arm 6, a first cutter 7, a second cutter 8 and a collection box 9; Wherein, the robotic arm, the first cutter and the second cutter are all connected to the controller signal, the front end of the robotic arm is fixedly connected to the mounting end at the bottom of the collection box, the upper part of the collection box is open and used to collect the ling seeds obtained by cutting, the first cutter and the second cutter are fixedly connected to both sides of the collection box, respectively, and the cutting parts of the first cutter and the second cutter extend to the top of the collection box respectively to cut the shafts on both sides of the ling seeds to obtain the ling seeds; wherein the cutting method of the cutting module is as follows: the controller controls the robotic arm to carry the collection box and move from bottom to top toward the target ling seeds, and adjusts the position of the collection box after moving to a preset distance from the target ling seeds, so that the center of the collection box is aligned with the center of the ling seeds, and then the controller turns on the first cutter and the second cutter, and then controls the robotic arm to carry the first cutter and the second cutter to cut the shafts on both sides of the target ling seeds and collect the target ling seeds into the collection box.

[0055] Among them, the collection box can be a rectangular parallelepiped or a cube, the first cutter and the second cutter can be corresponding electric cutting machines or automatic cutting machines, and the outer shell of the cutting machine is fixedly connected to the outer wall of the collection box, and the cutting end of the cutting machine extends to the top of the collection box. A corresponding cache layer can be provided in the collection box to buffer and protect the fallen ling seeds.

[0056] Among them, after identifying the target ling seeds, the controller controls the robotic arm to carry the collection box and move from bottom to top toward the target ling seeds, and adjusts the position of the collection box after moving to a preset distance from the target ling seeds. The preset distance can be adjusted according to actual needs, such as 0.3 to 0.5 meters. After the cutting is completed, the first cutter and the second cutter are closed. After the collection box collects the ling seeds, the robotic arm can carry the collection box toward the conveyor belt, and then rotate the collection box so that the opening of the collection box faces downward, and pour the ling seeds onto the conveyor belt, thereby realizing the cutting and transfer of the ling seeds.

[0057] The exposed incision after cutting is prone to microbial contamination, and a spray unit is provided inside the collection box to spray a disinfectant solution on the cut ling seeds. A biological preservative (such as chitosan solution) is sprayed immediately after cutting.

[0058] In the embodiment of the present invention, the first identification unit is further used to encode each type of ling to obtain the encoding result of each type of ling, and the appearance detection unit sequentially performs appearance defect detection on each type of ling based on the encoding result of each type of ling. The encoding can be digital or alphabetic, and the present invention does not specifically limit it.

[0059] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0060] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. The Chuanxiong Ling seed production system based on machine vision is characterized by: The system comprises: A feeding unit, used for feeding the ling rods to the first transmission unit; A first transmission unit, used for transmitting the ling stem to the cutting unit; A cutting unit, used for cutting the ling stems to obtain ling seeds; An image acquisition unit, used for acquiring images of the species; The first recognition unit is used to classify and identify the types of Ling on the Ling stem before cutting, and obtain the type detection result of the Ling; An appearance inspection unit is used to perform appearance defect inspection on the ling seeds based on the ling seeds image to obtain appearance defect inspection results; A first screening unit is used to transmit the Ling species with qualified appearance defect detection results to the odor detection unit, and transmit the Ling species with unqualified appearance defect detection results to the first collection unit; An odor detection unit is used to detect the odor of the Ling seeds and obtain the odor detection result; A second screening unit is used to transmit the Ling species with qualified smell detection results to the internal detection unit, and transmit the Ling species with unqualified smell detection results to the first collection unit; An internal detection unit, used for performing internal corruption detection on the Ling seeds and obtaining internal corruption detection results; The third screening unit is used to sort the Ling seeds that are qualified in the internal corruption detection result into Ling seed storage boxes corresponding to different categories according to the Ling seed category detection result, and transmit the Ling seeds that are unqualified in the internal corruption detection result to the first collection unit.

2. The machine vision-based Chuanxiong Ling seed production system according to claim 1, characterized in that: The odor detection unit comprises: A first gu ling seed obtaining unit, used to obtain a number of first gu ling seeds that can volatilize corresponding organic compounds due to the influence of diseases or insect pests, and to obtain a number of normal second gu ling seeds; An infrared spectrum analysis unit, used for performing infrared spectrum analysis on the first and second Ling species to obtain first and second infrared spectrum analysis results respectively; A first training sample obtaining unit, used for annotating the first and second infrared spectrum analysis results to obtain a first training sample; an odor detection model obtaining unit, configured to obtain an odor detection model based on the first training sample training model; The odor detection unit detects the odor of the species to be detected in the following manner: Using an infrared spectrum analysis unit to detect the smell of the tested Poria species to obtain an infrared spectrum analysis result of the tested Poria species; The infrared spectrum analysis result of the to-be-detected Ling species is input into the odor detection model, and the odor detection model outputs whether the odor of the to-be-detected Ling species is qualified or unqualified.

3. The machine vision-based Chuanxiong Ling seed production system according to claim 1, characterized in that: The appearance detection unit comprises: A second ling species obtaining unit, used for obtaining a plurality of third ling species with diseases, and used for obtaining a plurality of fourth ling species with insect pests; A feature extraction unit, used for extracting disease features from the image of the third type of ling, and for extracting insect pest features from the image of the fourth type of ling; A second training sample obtaining unit is used to mark all disease characteristics and insect pest characteristics to obtain a second training sample; an appearance detection model obtaining unit, configured to obtain an appearance detection model based on the second training sample training model; The appearance inspection unit performs appearance defect inspection on the inspected species in the following manner: Acquire and obtain images of the species to be detected; The image of the ling species to be detected is input into the appearance detection model, and the appearance detection model outputs the appearance pest and disease detection results of the ling species. If the appearance pest and disease detection results include pests or diseases, the appearance defect detection result is unqualified; if the appearance pest and disease detection results do not include pests and diseases, the appearance defect detection result is qualified.

4. The machine vision-based Chuanxiong Ling seed production system according to claim 1, characterized in that: The internal detection unit comprises: A third ling species obtaining unit, used to obtain a plurality of fifth ling species having a corruption region therein; Spectral camera, used to collect the spectral characteristics of the fifth species of Ling; A third training sample obtaining unit is used to annotate the spectral features of the fifth Ling species to obtain a third training sample; An internal corruption detection model obtaining unit, configured to obtain an internal corruption detection model based on the third training sample training model; Among them, the internal detection unit conducts internal corruption detection on Ling seeds in the following ways: Use a spectral camera to obtain the spectral characteristics of the species to be detected; The spectral characteristics of the to-be-detected corm species are input into the internal corruption detection model to obtain the internal corruption detection results of the to-be-detected corm species.

5. The machine vision-based Chuanxiong Ling seed production system according to claim 1, characterized in that: The first identification unit classifies and identifies the types of soya on the soya stem in the following manner: Taking pictures to obtain the image of the ling rod; Extract from the ling stem to obtain the size of each ling seed, the size of the ling stem connected to the ling seed, and the position information of the ling seed on the ling stem; Based on the size of each type of Ling, the size of the Ling stem connected to the type of Ling and the position information of the type of Ling on the Ling stem, the classification information of each type of Ling is obtained.

6. The machine vision-based Chuanxiong Ling seed production system according to claim 1, characterized in that: The cutting unit comprises: A second identification unit, a first clamping portion, a second clamping portion, a plurality of third clamping portions, a controller, a processor, and a plurality of cutting modules; The second recognition unit is used to capture a first overall image of the ling stalk to be cut, recognize the ling stalk head, ling stalk tail and ling stalk species in the first overall image, and obtain position information of the ling stalk head, position information of the ling stalk tail and the number M of ling stalk species; The controller is used to control the first clamping part and the second clamping part to clamp and fix the head and tail of the ling stem to be cut respectively based on the position information of the head and tail of the ling stem, so that the ling stem is suspended and fixed above all cutting modules; The processor is used to calculate the number N of third clamping parts involved in clamping and fixing the ling stems to be cut based on the number M of ling species, where M and N are both integers greater than 1; After the ling stem is suspended and fixed above all cutting modules, the second identification unit is used to capture a second overall image of the ling stem to be cut, identify the ling species in the second overall image, and obtain the position information of each ling species; The controller is used to control the third clamping part to clamp the middle part of the shaft between two adjacent types of ling based on the position information of each type of ling, and one third clamping part corresponds to the middle part between two adjacent types of ling; After the N third clamping parts have completed clamping, the controller is used to control the cutting module to cut the shafts on both sides of the Ling seeds along a cutting path from bottom to top to obtain the Ling seeds.

7. The machine vision-based Chuanxiong Ling seed production system according to claim 6, characterized in that: After all the Ling seeds are cut, the controller is also used to control the first clamping part, the second clamping part and the third clamping part to move the clamped object to above the first collection unit, and then control the first clamping part, the second clamping part and the third clamping part to end clamping, release the clamped object into the first collection unit, and then control the first clamping part, the second clamping part and the third clamping part to return to their respective initial positions.

8. The machine vision-based Chuanxiong Ling seed production system according to claim 6, characterized in that: The cutting module comprises: A robotic arm, a first cutter, a second cutter and a collection box; Wherein, the robotic arm, the first cutter and the second cutter are all connected to the controller signal, the front end of the robotic arm is fixedly connected to the mounting end at the bottom of the collection box, the upper part of the collection box is open and used to collect the ling seeds obtained by cutting, the first cutter and the second cutter are fixedly connected to both sides of the collection box, respectively, and the cutting parts of the first cutter and the second cutter extend to the top of the collection box respectively to cut the shafts on both sides of the ling seeds to obtain the ling seeds; wherein the cutting method of the cutting module is as follows: the controller controls the robotic arm to carry the collection box and move from bottom to top toward the target ling seeds, and adjusts the position of the collection box after moving to a preset distance from the target ling seeds, so that the center of the collection box is aligned with the center of the ling seeds, and then the controller turns on the first cutter and the second cutter, and then controls the robotic arm to carry the first cutter and the second cutter to cut the shafts on both sides of the target ling seeds and collect the target ling seeds into the collection box.

9. The machine vision-based Chuanxiong Ling seed production system according to claim 8, characterized in that: A spray unit is provided inside the collection box for spraying a disinfectant solution on the cut ling seeds.

10. The machine vision-based Chuanxiong Ling seed production system according to claim 6, characterized in that: The first recognition unit is further used to encode each type of ling to obtain an encoding result of each type of ling, and the appearance detection unit performs appearance defect detection on each type of ling in turn based on the encoding result of each type of ling.

Images