A sprouting seed screening device and method based on machine vision

Abstract

The application provides a kind of sprouting seed screening device and method based on machine vision, comprising: collecting several stage images of seed tray, marking seed position in each stage image respectively, determining several key analysis positions in seed tray, respectively acquiring image information of each key analysis position in different stage images, determining the growth trend of the seed corresponding to the key analysis position, then analyzing the estimated germination time period corresponding to each seed, acquiring the seed state feature in the estimated germination time period, further determining the planting position corresponding to each germinated seed in the seed tray using the seed state feature, marking the planting position and generating a seed germination report for transmission to a designated terminal for display. The application can identify the germinated seeds among many seeds using image recognition technology, reducing the pressure and workload of manual work, and improving the accuracy of screening.

Description

Technical Field

[0001] This invention relates to the field of agricultural industry technology, and in particular to a machine vision-based germination seed screening device and method. Background Technology

[0002] The main methods for identifying crop germination are observing changes in seeds, soil, and their environment, as well as using various tools and methods for detection. Germination is a crucial stage in plant growth, and timely identification of germination is essential for agricultural management and crop growth regulation. The following are some common methods for identifying crop germination:

[0003] 1. Visual inspection method

[0004] Method: This is the most common and direct method, determining germination status by observing the surface condition of the seeds and soil. Seeds in the soil or a moist environment can be checked periodically for a period after sowing.

[0005] Identification mark:

[0006] Seed coat cracking: After the seed absorbs water and swells, cracks appear in the outer shell.

[0007] Root growth: The seed roots begin to break through the seed coat and grow downwards, appearing as white or transparent roots.

[0008] Bud protrusion: The plumule breaks through the seed coat, usually a small green bud or stem.

[0009] Soil surface observation: If you see small green shoots or bud tips emerging from the soil, it means that the seeds have germinated.

[0010] 2. Water soaking method (seed soaking method)

[0011] Method: Soak the seeds in water and observe their swelling. Typically, the seeds will begin to swell after absorbing water, and the seed coat may crack. This is a physiological phenomenon before seed germination.

[0012] Operating steps:

[0013] Soak the seeds in clean water for 24 hours (the soaking time may vary depending on the type of seed).

[0014] Observe whether the seeds are swollen and note whether any cracks appear.

[0015] If some seeds begin to crack and reveal small roots or buds, it means that the seeds have entered the germination process.

[0016] 3. Wet cloth method

[0017] Method: Place the seeds in a damp cloth or paper towel to simulate a natural humid environment. Maintain the humidity of the cloth and a suitable temperature, and check regularly to see if the seeds have begun to germinate.

[0018] Operating steps:

[0019] Take an appropriate amount of seeds and place them on a damp paper towel or cloth.

[0020] Place the seeds and a damp cloth in a warm environment (e.g., 25-30°C).

[0021] Regularly check the seeds to see if there are any protruding embryos or roots.

[0022] Once you see the root or bud tip break through the seed coat, you can be sure that the seed has germinated.

[0023] 4. Temperature and humidity monitoring method

[0024] Methods: Germination status was inferred by monitoring the temperature and humidity of the soil or sowing environment, combined with the physiological characteristics of the seeds. Different crops have different germination temperature and humidity requirements; changes in these conditions can be used to indirectly determine the germination status.

[0025] Operating steps:

[0026] Use thermometers and hygrometers to monitor the temperature and humidity of the sowing environment to ensure they are within the suitable range for seed germination.

[0027] Under suitable conditions, regularly check the sowing area and judge the germination progress by combining changes in temperature and humidity.

[0028] Observe the soil surface or seedling tray for signs of seeds protruding.

[0029] 5. Transmissive light inspection method

[0030] Method: Use a light-transmitting device (such as a portable magnifying glass or microscope) to examine the internal structure of the seed and see if the embryo has begun to emerge.

[0031] Operating steps:

[0032] Remove the seeds and gently open them or use a transparent container to examine them.

[0033] Observe the inside of the seed for signs of roots or buds extending.

[0034] If the embryo has broken through the seed coat, it can be confirmed that the seed has begun to germinate.

[0035] 6. Soil detection method (probe method)

[0036] Method: Use specialized soil testing tools, such as soil probes or detectors, to check the condition of seeds deep in the soil. This method is typically used for large areas of farmland.

[0037] Operating steps:

[0038] Gently probe the depth of seeds in the soil using a soil probe.

[0039] If the probe touches the roots or buds that have broken through the soil, it means that the seed has germinated.

[0040] At the same time, soil temperature and humidity are monitored to ensure that seeds germinate in a suitable environment.

[0041] 7. Seed Vigor Test Method

[0042] Method: The germination ability of seeds is checked through a professional seed vigor test. This usually needs to be conducted under laboratory conditions using standardized germination testing methods.

[0043] Operating steps:

[0044] A batch of seeds was taken out and processed according to standard testing procedures.

[0045] Under suitable temperature and humidity conditions, observe the germination of the seeds.

[0046] Record the germination rate and germination time to determine seed vigor and suitability for planting.

[0047] While the above methods can identify germinated seeds among many seeds, they all require manual observation and screening. Since different seeds have different germination cycles, and the observation experiments also require professional personnel, identifying whether a seed has germinated has become a high-pressure and high-workload task.

[0048] Therefore, the present invention provides a machine vision-based germination seed screening device and method. Summary of the Invention

[0049] This invention provides a machine vision-based germination seed screening device and method, which can use image recognition technology to identify germinating seeds among many seeds, reducing manual labor and improving screening accuracy.

[0050] This invention provides a machine vision-based germination seed screening device, comprising:

[0051] The visual acquisition module is used to acquire several stage images of the seed tray, mark the seed position in each stage image, and determine several key analysis positions in the seed tray.

[0052] The key analysis module is used to acquire image information of each key analysis location in images at different stages, and determine the seed growth trend corresponding to the key analysis location based on the image information.

[0053] The germination monitoring module is used to analyze the estimated germination time period corresponding to each seed based on the seed growth and germination trend, and to obtain the seed state characteristics within the estimated germination time period.

[0054] The germination screening module is used to determine the planting position corresponding to each germinated seed in the seed tray based on the seed state characteristics, mark the planting position, generate a seed germination report, and transmit it to a designated terminal for display.

[0055] In one feasible approach

[0056] The visual acquisition module includes:

[0057] The cycle adjustment unit is used to acquire the seed attributes in the seed tray, search for the seed attributes in big data, determine the growth data of the corresponding seed, simulate the growth of the seed based on the growth data and the environmental data of the seed tray, determine the growth cycle of the seed in the seed tray, and determine the image acquisition frequency.

[0058] The acquisition execution unit is used to acquire several stage images of the seed plate according to the image acquisition frequency, and to encode the corresponding stage image according to the acquisition time corresponding to each stage image to generate a periodic image set of the seed plate.

[0059] The key positioning unit is used to perform pixel segmentation on each stage image, determine the seed distribution position of the corresponding stage image based on the segmentation result, obtain the distribution information corresponding to each stage image, and perform fusion processing on the distribution information in the periodic image set to determine several key analysis positions of the seed plate.

[0060] In one feasible approach

[0061] The key analysis module includes:

[0062] The preliminary processing unit is used to establish a planting matrix for the seed tray based on the distribution of key analysis positions in the seed tray, mark the key analysis positions in each stage image, divide each stage image into several unit images, and obtain a set of unit images corresponding to each key analysis position.

[0063] An information recognition unit is used to identify the image change features in each of the unit image sets, map each image change feature to the element position corresponding to the planting matrix, obtain the planting change matrix of the seed plate, and analyze the influence features between different adjacent key analysis positions based on the planting change matrix.

[0064] The information processing unit is used to correct the corresponding image change features according to several influencing features corresponding to each key analysis location, generate image information corresponding to each key analysis location, and draw a dynamic map of the region corresponding to the key analysis location based on the image information.

[0065] The trend analysis unit is used to identify the fixed position of the seed corresponding to the key analysis position in the dynamic map of the region, obtain the dynamic change characteristics corresponding to the fixed position of the seed, perform growth analysis on the dynamic change characteristics based on the growth data of the seed, and determine the seed growth trend of the corresponding key analysis position.

[0066] In one feasible approach

[0067] The germination monitoring module includes:

[0068] The trend analysis unit is used to determine the current growth quality of the corresponding seed based on the seed growth trend, classify the seeds with the same current growth quality into the same seed class, obtain the planting time corresponding to each seed, and determine the growth time of the corresponding seed based on the planting time.

[0069] The deep analysis unit is used to analyze the nutrient supply rate of different seeds in each seed class based on the growth time, analyze the estimated germination time of the corresponding seeds in combination with the current growth quality, and add category tags and time period tags to each seed according to the estimated germination time of different seeds in each seed class.

[0070] The state refinement unit is used to determine the current growth state and remaining germination time of the corresponding seed according to the category label and time period label of each seed, draw the growth concept chart of the corresponding seed, and identify the seed growth information corresponding to each unit time in the growth concept chart according to the corresponding growth trend and nutrient supply rate.

[0071] The overall analysis unit is used to collect real-time growth information corresponding to each seed, determine the current growth deviation information of the corresponding seed according to the unit time corresponding to the monitoring collection time, adjust the estimated germination time period of the corresponding seed according to the current growth deviation information, and obtain the precise germination time period corresponding to each seed. When the monitoring collection time coincides with the precise germination time period, the seed state characteristics of the corresponding seed are obtained.

[0072] In one feasible approach

[0073] The process of adjusting the estimated germination time period of the corresponding seeds based on the current growth deviation information to obtain the precise germination time period for each seed includes:

[0074] When the current growth deviation information is positive, it is determined that the corresponding seed is currently in a positive stimulation state, and the current positive stimulation level of the seed is analyzed based on the deviation amount corresponding to the growth deviation information.

[0075] The estimated germination time period is shifted forward based on the current positive stimulus level to obtain the temporary germination time period of the seed.

[0076] The updated growth deviation information of the seed at the next supervised acquisition time is obtained. When the updated growth deviation is positive, the stable stimulation level of the external environment on the corresponding seed is determined according to the updated positive stimulation level corresponding to the updated growth deviation information and the current positive stimulation level.

[0077] Based on the stable stimulation level and the corresponding seed growth trend, the growth stimulation rate of the external environment on the seed is analyzed, and the duration of the temporary germination time period is adjusted according to the growth stimulation rate to obtain the precise germination time period of the seed.

[0078] When the current growth deviation information is negative, the estimated germination time period is regarded as the precise germination time period of the corresponding seed.

[0079] When the current growth deviation information is positive and the corresponding updated growth deviation information is negative, the estimated germination time period is regarded as the precise germination time period of the corresponding seed.

[0080] In one feasible approach

[0081] The germination screening module includes:

[0082] A germination positioning unit is used to determine, based on the seed state, a number of germinating seeds in the seed tray that exhibit germination characteristics, and to obtain the planting position corresponding to each germinating seed.

[0083] A germination marking unit is used to select a promoting light of a corresponding wavelength according to the seed attribute, establish a marking cursor using the promoting light, and mark the planting position using the marking cursor;

[0084] The report generation unit is used to acquire the appearance information corresponding to each germinated seed, generate a seed germination report for the seed tray, and transmit it to a designated terminal for display.

[0085] In one feasible approach

[0086] The report generation unit includes:

[0087] The visual analysis subunit is used to acquire the germination image corresponding to each germinating seed, project the image of each germination image to obtain the seed image contained in each germination image, and use the peak detection method to identify corner points in each seed image to obtain the connection line of several corner points contained in each seed image.

[0088] A germination classification subunit is used to map the corner point connection line onto the corresponding germination image to obtain seed crack information and seed water emergence information corresponding to each germination image. The first germinating seed containing both the seed crack information and the seed water emergence information is regarded as a type I germinating seed, the second germinating seed containing the seed crack information is regarded as a type II germinating seed, and the third germinating seed containing the seed water emergence information is regarded as a type III germinating seed.

[0089] The report generation unit is used to calculate the germination rate of the seeds in the seed tray, as well as the number of germinations for each type of germinating seed, to create a seed germination report for the seed tray and transmit it to a designated terminal for display.

[0090] In one feasible approach

[0091] Also includes:

[0092] The analysis and reminder module is used to collect the non-germination time corresponding to each key analysis location. When the non-germination time is longer than a preset time, the corresponding seed germination is identified and a reminder message is generated.

[0093] This invention provides a machine vision-based method for screening germinating seeds, comprising:

[0094] Step 1: Collect several phase images of the seed tray, mark the seed position in each phase image, and determine several key analysis positions in the seed tray;

[0095] Step 2: Obtain image information of each key analysis location in images at different stages, and determine the seed growth trend corresponding to the key analysis location based on the image information;

[0096] Step 3: Analyze the estimated germination time period for each seed based on the seed growth and germination trend, and obtain the seed state characteristics within the estimated germination time period;

[0097] Step 4: Determine the planting position corresponding to each germinated seed in the seed tray based on the seed state characteristics, mark the planting position, generate a seed germination report, and transmit it to the designated terminal for display.

[0098] In one feasible approach

[0099] Step 3 includes:

[0100] Step 31: Determine the current growth quality of the corresponding seed based on the seed growth trend, classify the seeds with the same current growth quality into the same seed class, obtain the planting time corresponding to each seed, and determine the growth time of the corresponding seed based on the planting time;

[0101] Step 32: Analyze the nutrient supply rate corresponding to different seeds in each seed class based on the growth time, and combine it with the corresponding current growth quality to analyze the estimated germination time period of the corresponding seeds. Add category tags and time period tags to each seed according to the estimated germination time period corresponding to different seeds in each seed class.

[0102] Step 33: Determine the current growth status and remaining germination time of the corresponding seed according to the category label and time period label of each seed, draw the growth concept chart of the corresponding seed, and identify the seed growth information corresponding to each unit of time in the growth concept chart according to the corresponding growth trend and nutrient supply rate.

[0103] Step 34: Collect real-time growth information for each seed, determine the current growth deviation information of the seed based on the unit time corresponding to the monitoring collection time, adjust the estimated germination time period of the seed based on the current growth deviation information, and obtain the precise germination time period for each seed. When the monitoring collection time coincides with the precise germination time period, obtain the seed state characteristics of the seed.

[0104] The beneficial effects of the above technical solution are as follows: To reduce the workload of staff and improve the efficiency of screening germinating seeds, several stage images of the seed tray are first collected. The seed positions are determined in the stage images, thereby identifying several key analysis positions in the seed tray. Further image analysis is performed on the key analysis positions to determine the seed growth trend of the seeds in each key analysis position, thereby estimating the estimated germination time period of the seeds. When the estimated germination time period is reached, the seed state characteristics of the seeds during this period are collected. Based on the seed state characteristics, the planting positions of the germinated seeds in the seed tray are determined, and corresponding reminders are given. Finally, a seed germination report is generated based on the germination data and transmitted to a designated terminal for display. Staff can quickly determine the positions of the germinated seeds based on the report, achieving the purpose of rapid screening. Moreover, using image analysis for screening can also improve the accuracy of screening, reduce the randomness of manual screening, and improve the speed of agricultural development.

[0105] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0106] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0107] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0108] Figure 1 This is a schematic diagram of the composition of a machine vision-based seed germination screening device according to an embodiment of the present invention;

[0109] Figure 2 This is a schematic diagram illustrating the workflow of a machine vision-based seed germination screening method according to an embodiment of the present invention. Detailed Implementation

[0110] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0111] Example 1

[0112] This embodiment provides a machine vision-based germination seed screening device, such as... Figure 1 As shown, it includes:

[0113] The visual acquisition module is used to acquire several stage images of the seed tray, mark the seed position in each stage image, and determine several key analysis positions in the seed tray.

[0114] The key analysis module is used to acquire image information of each key analysis location in images at different stages, and determine the seed growth trend corresponding to the key analysis location based on the image information.

[0115] The germination monitoring module is used to analyze the estimated germination time period corresponding to each seed based on the seed growth and germination trend, and to obtain the seed state characteristics within the estimated germination time period.

[0116] The germination screening module is used to determine the planting position corresponding to each germinated seed in the seed tray based on the seed state characteristics, mark the planting position, generate a seed germination report, and transmit it to a designated terminal for display.

[0117] In this example, the stage image represents images acquired at different times;

[0118] In this example, several seeds are planted in the seed tray, and the seeds may be planted at different times;

[0119] In this example, the key analysis position represents the location of the seed. One key analysis position contains one or more seeds. When the interval between seeds is greater than 2 cm, one key analysis position corresponds to one seed. When the interval between seeds is less than 2 cm, one key analysis position corresponds to multiple seeds.

[0120] In this example, the image information contains information about the next key analysis location at different times;

[0121] In this example, the seed growth trend indicates the growth of seeds planted at the key analysis location;

[0122] In this example, the estimated germination time period means that the seed will germinate at any point within that time period.

[0123] The working principle and beneficial effects of the above technical solution are as follows: To reduce the workload of staff and improve the efficiency of screening germinated seeds, several stage images of the seed tray are first collected. The seed positions are determined in the stage images, thereby identifying several key analysis positions in the seed tray. Further image analysis is performed on the key analysis positions to determine the seed growth trend of the seeds in each key analysis position, thereby estimating the estimated germination time period of the seeds. When the estimated germination time period is reached, the seed state characteristics of the seeds during this period are collected. Based on the seed state characteristics, the planting positions of the germinated seeds in the seed tray are determined, and corresponding reminders are given. Finally, a seed germination report is generated based on the germination data and transmitted to a designated terminal for display. Staff can quickly determine the positions of the germinated seeds based on the report, achieving the purpose of rapid screening. Moreover, using image analysis for screening can also improve the accuracy of screening, reduce the randomness of manual screening, and improve the speed of agricultural development.

[0124] Example 2

[0125] Based on Example 1, the machine vision-based germination seed screening device is characterized in that the vision acquisition module includes:

[0126] The cycle adjustment unit is used to acquire the seed attributes in the seed tray, search for the seed attributes in big data, determine the growth data of the corresponding seed, simulate the growth of the seed based on the growth data and the environmental data of the seed tray, determine the growth cycle of the seed in the seed tray, and determine the image acquisition frequency.

[0127] The acquisition execution unit is used to acquire several stage images of the seed plate according to the image acquisition frequency, and to encode the corresponding stage image according to the acquisition time corresponding to each stage image to generate a periodic image set of the seed plate.

[0128] The key positioning unit is used to perform pixel segmentation on each stage image, determine the seed distribution position of the corresponding stage image based on the segmentation result, obtain the distribution information corresponding to each stage image, and perform fusion processing on the distribution information in the periodic image set to determine several key analysis positions of the seed plate.

[0129] In this example, the seed attribute includes the seed's name and seed origin, which are used to identify the seed;

[0130] In this example, the growth data represents the data generated when the seed grows in an adapted environment;

[0131] In this example, the growth cycle represents the period from planting to germination of a seed in the seed tray. This cycle includes the seed's water absorption stage (early germination), activation stage (enzyme activation stage), radicle breakthrough stage, plumule breakthrough stage, first true leaf unfolding stage, and root development and nutrient absorption stage.

[0132] In this example, the image acquisition frequency refers to the frequency at which images are taken of the seed tray. The image acquisition frequency is related to the seed's growth cycle, and three images need to be taken at each stage of the growth cycle.

[0133] In this example, the periodic image set represents the result of encoding and sorting the phase images in chronological order;

[0134] In this example, the distribution information represents the distribution of seeds in a stage image;

[0135] In this example, the purpose of the fusion process is to verify the position of the seed.

[0136] The working principle and beneficial effects of the above technical solution are as follows: To improve the screening efficiency of germinating seeds, it is necessary to determine the planting location of the seeds in the early stage of screening, and then focus on analyzing the planting location. First, the growth data of the seeds is found according to their attributes. The growth cycle of the seeds in the seed tray is determined by simulation, and the frequency of image acquisition is determined. Then, stage images of the seed tray are acquired according to this frequency. By encoding and sorting the stage images, a periodic image set of the seed tray is established. The seed distribution information presented in each periodic image is fused, thereby calibrating and verifying the seed position multiple times and determining several key analysis locations in the seed tray. Therefore, when conducting analysis and screening, it is only necessary to focus on analyzing the key analysis locations to determine the location of germinating seeds, reducing a lot of unnecessary identification work and improving the efficiency of the screening work.

[0137] Example 3

[0138] Based on Example 1, the key analysis module of the machine vision-based germination seed screening device includes:

[0139] The preliminary processing unit is used to establish a planting matrix for the seed tray based on the distribution of key analysis positions in the seed tray, mark the key analysis positions in each stage image, divide each stage image into several unit images, and obtain a set of unit images corresponding to each key analysis position.

[0140] An information recognition unit is used to identify the image change features in each of the unit image sets, map each image change feature to the element position corresponding to the planting matrix, obtain the planting change matrix of the seed plate, and analyze the influence features between different adjacent key analysis positions based on the planting change matrix.

[0141] The information processing unit is used to correct the corresponding image change features according to several influencing features corresponding to each key analysis location, generate image information corresponding to each key analysis location, and draw a dynamic map of the region corresponding to the key analysis location based on the image information.

[0142] The trend analysis unit is used to identify the fixed position of the seed corresponding to the key analysis position in the dynamic map of the region, obtain the dynamic change characteristics corresponding to the fixed position of the seed, perform growth analysis on the dynamic change characteristics based on the growth data of the seed, and determine the seed growth trend of the corresponding key analysis position.

[0143] In this example, the planting matrix represents a blank matrix established based on the distribution of key analysis locations;

[0144] In this example, the unit image represents the image used to indicate the key analysis location within a phase image;

[0145] In this example, the unit image set refers to the image set generated by combining unit images of a key analysis location in different phase images;

[0146] In this example, the image variation feature represents the feature that changes with respect to the seed, which consists of the differences between different unit images in a unit image set;

[0147] In this example, a unit image set corresponds to an image change feature, and also to an element bit;

[0148] In this example, the influence feature represents the seed growth influence between key analysis locations that have an adjacent relationship;

[0149] In this example, the regional dynamic map represents a dynamic image of the changes occurring within the key analysis area;

[0150] In this example, the seed's position in the key analysis location is fixed.

[0151] The working principle and beneficial effects of the above technical solution are as follows: Since the environment affects seed germination, in order to better analyze the seed growth trend, a planting matrix is ​​first established based on the distribution of key analysis positions in the seed tray. Each stage image is then divided into several unit images, generating a unit image set corresponding to each key analysis position. The image change characteristics of the unit image set are then mapped onto the planting matrix to analyze the influence characteristics between different adjacent key analysis positions. The influence characteristics are then used to correct the image change characteristics, determining the image information at the key analysis positions and thus drawing a regional dynamic map of the key analysis positions. By focusing on the fixed seed positions in the regional dynamic map, the dynamic change characteristics of those positions are obtained. Combined with the seed growth data, growth analysis is performed to determine the seed growth trend at the key analysis positions. In this way, all seeds in the seed tray can be analyzed simultaneously and their growth trends can be determined, facilitating subsequent screening.

[0152] Example 4

[0153] Based on Example 1, the germination monitoring module of the machine vision-based seed screening device includes:

[0154] The trend analysis unit is used to determine the current growth quality of the corresponding seed based on the seed growth trend, classify the seeds with the same current growth quality into the same seed class, obtain the planting time corresponding to each seed, and determine the growth time of the corresponding seed based on the planting time.

[0155] The deep analysis unit is used to analyze the nutrient supply rate of different seeds in each seed class based on the growth time, analyze the estimated germination time of the corresponding seeds in combination with the current growth quality, and add category tags and time period tags to each seed according to the estimated germination time of different seeds in each seed class.

[0156] The state refinement unit is used to determine the current growth state and remaining germination time of the corresponding seed according to the category label and time period label of each seed, draw the growth concept chart of the corresponding seed, and identify the seed growth information corresponding to each unit time in the growth concept chart according to the corresponding growth trend and nutrient supply rate.

[0157] The overall analysis unit is used to collect real-time growth information corresponding to each seed, determine the current growth deviation information of the corresponding seed according to the unit time corresponding to the monitoring collection time, adjust the estimated germination time period of the corresponding seed according to the current growth deviation information, and obtain the precise germination time period corresponding to each seed. When the monitoring collection time coincides with the precise germination time period, the seed state characteristics of the corresponding seed are obtained.

[0158] In this example, current growth quality indicates the current growth stage of the seed and the degree of growth at that stage;

[0159] In this example, the faster the nutrient supply rate, the closer the estimated germination time of the seed is to the current moment;

[0160] In this example, the purpose of adding category labels and time period labels is to further classify the seeds;

[0161] In this example, the current growth status is related to the current growth quality, and they are synchronous.

[0162] In this example, the remaining germination time represents the duration between the current moment and the last moment of the estimated germination time period;

[0163] In this example, the current growth deviation information represents the information corresponding to the inconsistency between the real-time growth information and the seed growth information at the corresponding unit time.

[0164] In this example, seed growth information represents the analysis of seed growth at different time points using a predictive approach.

[0165] In this example, the monitoring data collection time corresponds one-to-one with the unit time.

[0166] In this example, the precise germination time period represents the result of further refining and adjusting the estimated germination time period based on the determination of the impact of the external environment on the seed according to the real-time growth information of the seed.

[0167] The working principle and beneficial effects of the above technical solution are as follows: First, the current growth quality of the seeds is analyzed based on their growth trend. Then, the seeds are classified, and the growth time of each seed is determined by combining the planting time of each seed. This allows for the analysis of the nutrient supply rate of different seeds within the same seed category, determining the estimated germination time period. Corresponding category and time period labels are added to each seed to facilitate the subsequent determination of the current growth status and remaining germination time of each seed. A growth concept chart for each seed is drawn, and the nutrient supply rate of the seed is analyzed through this chart to determine the seed growth information corresponding to each unit of time. Then, the seeds are monitored in real time to determine their current growth deviation information at different times. Finally, this current growth deviation information is used to further refine the estimated germination time period of the seeds, determining the precise germination time period. Within this time period, the state characteristics of the seeds are collected. This method allows for in-depth monitoring of the seed situation during the period before germination, enabling the collection of germination information at the first moment of germination, improving the working efficiency of the germinating seed screening device, and reducing errors from manual screening.

[0168] Example 5

[0169] Based on Example 4, the process of adjusting the estimated germination time period of the corresponding seeds according to the current growth deviation information to obtain the precise germination time period for each seed in the machine vision-based seed selection device includes:

[0170] When the current growth deviation information is positive, it is determined that the corresponding seed is currently in a positive stimulation state, and the current positive stimulation level of the seed is analyzed based on the deviation amount corresponding to the growth deviation information.

[0171] The estimated germination time period is shifted forward based on the current positive stimulus level to obtain the temporary germination time period of the seed.

[0172] The updated growth deviation information of the seed at the next supervised acquisition time is obtained. When the updated growth deviation is positive, the stable stimulation level of the external environment on the corresponding seed is determined according to the updated positive stimulation level corresponding to the updated growth deviation information and the current positive stimulation level.

[0173] Based on the stable stimulation level and the corresponding seed growth trend, the growth stimulation rate of the external environment on the seed is analyzed, and the duration of the temporary germination time period is adjusted according to the growth stimulation rate to obtain the precise germination time period of the seed.

[0174] When the current growth deviation information is negative, the estimated germination time period is regarded as the precise germination time period of the corresponding seed.

[0175] When the current growth deviation information is positive and the corresponding updated growth deviation information is negative, the estimated germination time period is regarded as the precise germination time period of the corresponding seed.

[0176] In this example, the positive stimulus state indicates that the external environment promotes seed growth;

[0177] In this example, the current positive stimulus level is related to the external environment. For each level increase in the current positive stimulus level, the positive stimulus moves forward by one unit of time, with a maximum of 4 units of time. The unit of time is set in advance by the staff, usually 24 hours.

[0178] In this example, the direction of forward movement is the direction at the current moment;

[0179] In this example, the updated growth deviation information represents the deviation between the real-time growth information and the seed growth information at the next monitoring time.

[0180] In this example, the growth stimulation rate is related to the external environment, and the precise germination time can be obtained by multiplying the growth stimulation rate by the corresponding temporary germination time period.

[0181] In this example, when adjusting the estimated germination time period, the odd-numbered monitoring collection time and its adjacent even-numbered monitoring collection time are regarded as a deviation adjustment combination. The purpose is to determine the stable external stimulation of the seeds and reduce randomness.

[0182] The working principle and beneficial effects of the above technical solution are as follows: In the process of adjusting the estimated germination time period, the stimulation of the external environment on the seeds is taken into account. By analyzing the growth deviation information corresponding to different monitoring and collection times, a combined analysis is performed to respond to the positive stimulation of the external environment on the seeds, thereby scaling up or down the range of the estimated germination time period of the seeds, determining the accurate germination time period, and improving the efficiency of screening germinating seeds.

[0183] Example 6

[0184] Based on Example 1, the germination screening module of the machine vision-based germination seed screening device includes:

[0185] A germination positioning unit is used to determine, based on the seed state, a number of germinating seeds in the seed tray that exhibit germination characteristics, and to obtain the planting position corresponding to each germinating seed.

[0186] A germination marking unit is used to select a promoting light of a corresponding wavelength according to the seed attribute, establish a marking cursor using the promoting light, and mark the planting position using the marking cursor;

[0187] The report generation unit is used to acquire the appearance information corresponding to each germinated seed, generate a seed germination report for the seed tray, and transmit it to a designated terminal for display.

[0188] In this example, promoting light refers to light that promotes seed growth. The wavelength of promoting light is determined by the seed properties, and different wavelengths of promoting light have a promoting effect on different seeds.

[0189] The working principle and beneficial effects of the above technical solution are as follows: the germination status of seeds in the seed tray is determined by the seed status, and then the location of the germinating seeds is reminded by the light marking method. At the same time, a seed germination report is generated for relevant personnel to view, so that the growth status of the seeds in the seed tray can be understood at any time.

[0190] Example 7

[0191] Based on Example 6, the report generation unit of the machine vision-based germination seed screening device includes:

[0192] The visual analysis subunit is used to acquire the germination image corresponding to each germinating seed, project the image of each germination image to obtain the seed image contained in each germination image, and use the peak detection method to identify corner points in each seed image to obtain the connection line of several corner points contained in each seed image.

[0193] A germination classification subunit is used to map the corner point connection line onto the corresponding germination image to obtain seed crack information and seed water emergence information corresponding to each germination image. The first germinating seed containing both the seed crack information and the seed water emergence information is regarded as a type I germinating seed, the second germinating seed containing the seed crack information is regarded as a type II germinating seed, and the third germinating seed containing the seed water emergence information is regarded as a type III germinating seed.

[0194] The report generation unit is used to calculate the germination rate of the seeds in the seed tray, as well as the number of germinations for each type of germinating seed, to create a seed germination report for the seed tray and transmit it to a designated terminal for display.

[0195] In this example, the corner line represents an irregular line segment composed of edges, edge intersections, and line corners contained in the seed image;

[0196] In this example, the seed crack information indicates that the seed is cracked;

[0197] In this example, the seed water emergence information indicates that the seed has germinated and there is no water emerging from the sprout;

[0198] In this example, different types of germinating seeds represent seeds at different stages of germination.

[0199] The working principle and beneficial effects of the above technical solution are as follows: By projecting the germination image of the germinated seeds, corner points in the seed image are identified. Then, the corner points are continuously mapped onto the germination image to identify seed crack information and seed water emergence information. Based on these two pieces of information, the germinated seeds are classified. Combined with the germination rate of the seeds in the seed tray, a seed germination report is constructed for relevant personnel to view. Relevant personnel can use this report to understand the germination status of the seeds at any time, providing a reference for their next technical work.

[0200] Example 8

[0201] Based on Example 1, the machine vision-based germination seed screening device further includes:

[0202] The analysis and reminder module is used to collect the non-germination time corresponding to each key analysis location. When the non-germination time is longer than a preset time, the corresponding seed germination is identified and a reminder message is generated.

[0203] The working principle and beneficial effects of the above technical solution are as follows: when seeds fail to germinate for a long time, it reminds relevant personnel to check and investigate the cause as soon as possible.

[0204] Example 9

[0205] This embodiment provides a machine vision-based method for screening germinating seeds, such as... Figure 2 As shown, it includes:

[0206] Step 1: Collect several phase images of the seed tray, mark the seed position in each phase image, and determine several key analysis positions in the seed tray;

[0207] Step 2: Obtain image information of each key analysis location in images at different stages, and determine the seed growth trend corresponding to the key analysis location based on the image information;

[0208] Step 3: Analyze the estimated germination time period for each seed based on the seed growth and germination trend, and obtain the seed state characteristics within the estimated germination time period;

[0209] Step 4: Determine the planting position corresponding to each germinated seed in the seed tray based on the seed status characteristics, mark the planting position, generate a seed germination report, and transmit it to the designated terminal for display.

[0210] In this example, the stage image represents images acquired at different times;

[0211] In this example, several seeds are planted in the seed tray, and the seeds may be planted at different times;

[0212] In this example, the key analysis position represents the location of the seed. One key analysis position contains one or more seeds. When the interval between seeds is greater than 2 cm, one key analysis position corresponds to one seed. When the interval between seeds is less than 2 cm, one key analysis position corresponds to multiple seeds.

[0213] In this example, the image information contains information about the next key analysis location at different times;

[0214] In this example, the seed growth trend indicates the growth of seeds planted at the key analysis location;

[0215] In this example, the estimated germination time period means that the seed will germinate at any point within that time period.

[0216] The working principle and beneficial effects of the above technical solution are as follows: To reduce the workload of staff and improve the efficiency of screening germinated seeds, several stage images of the seed tray are first collected. The seed positions are determined in the stage images, thereby identifying several key analysis positions in the seed tray. Further image analysis is performed on the key analysis positions to determine the seed growth trend of the seeds in each key analysis position, thereby estimating the estimated germination time period of the seeds. When the estimated germination time period is reached, the seed state characteristics of the seeds during this period are collected. Based on the seed state characteristics, the planting positions of the germinated seeds in the seed tray are determined, and corresponding reminders are given. Finally, a seed germination report is generated based on the germination data and transmitted to a designated terminal for display. Staff can quickly determine the positions of the germinated seeds based on the report, achieving the purpose of rapid screening. Moreover, using image analysis for screening can also improve the accuracy of screening, reduce the randomness of manual screening, and improve the speed of agricultural development.

[0217] Example 10

[0218] Based on Example 9, the step 3 of the machine vision-based germination seed screening method includes:

[0219] Step 31: Determine the current growth quality of the corresponding seed based on the seed growth trend, classify the seeds with the same current growth quality into the same seed class, obtain the planting time corresponding to each seed, and determine the growth time of the corresponding seed based on the planting time;

[0220] Step 32: Analyze the nutrient supply rate corresponding to different seeds in each seed class based on the growth time, and combine it with the corresponding current growth quality to analyze the estimated germination time period of the corresponding seeds. Add category tags and time period tags to each seed according to the estimated germination time period corresponding to different seeds in each seed class.

[0221] Step 33: Determine the current growth status and remaining germination time of the corresponding seed according to the category label and time period label of each seed, draw the growth concept chart of the corresponding seed, and identify the seed growth information corresponding to each unit of time in the growth concept chart according to the corresponding growth trend and nutrient supply rate.

[0222] Step 34: Collect real-time growth information for each seed, determine the current growth deviation information of the seed based on the unit time corresponding to the monitoring collection time, adjust the estimated germination time period of the seed based on the current growth deviation information, and obtain the precise germination time period for each seed. When the monitoring collection time coincides with the precise germination time period, obtain the seed state characteristics of the seed.

[0223] In this example, current growth quality indicates the current growth stage of the seed and the degree of growth at that stage;

[0224] In this example, the faster the nutrient supply rate, the closer the estimated germination time of the seed is to the current moment;

[0225] In this example, the purpose of adding category labels and time period labels is to further classify the seeds;

[0226] In this example, the current growth status is related to the current growth quality, and they are synchronous.

[0227] In this example, the remaining germination time represents the duration between the current moment and the last moment of the estimated germination time period;

[0228] In this example, the current growth deviation information represents the information corresponding to the inconsistency between the real-time growth information and the seed growth information at the corresponding unit time.

[0229] In this example, seed growth information represents the analysis of seed growth at different time points using a predictive approach.

[0230] In this example, the monitoring data collection time corresponds one-to-one with the unit time.

[0231] In this example, the precise germination time period represents the result of further refining and adjusting the estimated germination time period based on the determination of the impact of the external environment on the seed according to the real-time growth information of the seed.

[0232] The working principle and beneficial effects of the above technical solution are as follows: First, the current growth quality of the seeds is analyzed based on their growth trend. Then, the seeds are classified, and the growth time of each seed is determined by combining the planting time of each seed. This allows for the analysis of the nutrient supply rate of different seeds within the same seed category, determining the estimated germination time period. Corresponding category and time period labels are added to each seed to facilitate the subsequent determination of the current growth status and remaining germination time of each seed. A growth concept chart for each seed is drawn, and the nutrient supply rate of the seed is analyzed through this chart to determine the seed growth information corresponding to each unit of time. Then, the seeds are monitored in real time to determine their current growth deviation information at different times. Finally, this current growth deviation information is used to further refine the estimated germination time period of the seeds, determining the precise germination time period. Within this time period, the state characteristics of the seeds are collected. This method allows for in-depth monitoring of the seed situation during the period before germination, enabling the collection of germination information at the first moment of germination, improving the working efficiency of the germinating seed screening device, and reducing errors from manual screening.

[0233] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A machine vision based germinating seed screening apparatus, characterized by, The method comprises the following steps: A visual acquisition module is configured to acquire a plurality of stage images of a seed tray, mark seed positions in each of the stage images, and determine a plurality of key analysis positions in the seed tray; A key analysis module is configured to acquire image information of each of the key analysis positions in different stage images, determine seed growth trends corresponding to the key analysis positions according to the image information, and analyze estimated germination time periods of the seeds corresponding to the key analysis positions according to the seed growth trends; A germination supervision module is configured to acquire seed state features in the estimated germination time periods of the seeds, determine planting positions of each of the seeds in the seed tray according to the seed state features, mark the planting positions, and generate a seed germination report for display on a designated terminal; The germination supervision module comprises: A trend analysis unit is configured to determine current growth qualities of the seeds corresponding to the key analysis positions according to the seed growth trends, list the seeds with the same current growth qualities as a same seed class, acquire planting time of each of the seeds, and determine grown time lengths of the seeds corresponding to the key analysis positions according to the planting time; A deep analysis unit is configured to analyze nutrient supply speeds of different seeds in each of the seed classes according to the grown time lengths, analyze estimated germination time periods of the seeds corresponding to the key analysis positions according to the current growth qualities and the nutrient supply speeds, add class labels and time period labels to each of the seeds according to the estimated germination time periods of the different seeds in each of the seed classes; A state refinement unit is configured to determine current growth states and remaining germination time lengths of the seeds corresponding to the key analysis positions according to the class labels and the time period labels of each of the seeds, draw a growth concept chart of the seeds corresponding to the key analysis positions, and identify seed growth information of each unit time in the growth concept chart according to the growth trends and the nutrient supply speeds; A whole-tray analysis unit is configured to acquire real-time growth information of each of the seeds, determine current growth deviation information of the seeds corresponding to the key analysis positions according to the unit time corresponding to a supervision acquisition time, adjust the estimated germination time periods of the seeds corresponding to the key analysis positions according to the current growth deviation information, obtain accurate germination time periods of each of the seeds, and acquire seed state features of the seeds corresponding to the key analysis positions when the supervision acquisition time coincides with the accurate germination time periods. The visual acquisition module comprises:

2. A machine vision based germinating seed screening device as claimed in claim 1, wherein, A cycle adjustment unit is configured to acquire seed attributes in the seed tray, find the seed attributes in big data, determine growth data of the seeds corresponding to the seed attributes, simulate growth conditions of the seeds according to the growth data and environment data of the seed tray, determine a growth cycle of the seeds in the seed tray, and determine an image acquisition frequency; An acquisition execution unit is configured to acquire a plurality of stage images of the seed tray according to the image acquisition frequency, encode the stage images according to acquisition times corresponding to the stage images, and generate a cycle image set of the seed tray. ​ The focus positioning unit is configured to perform pixel segmentation on each of the stage images, determine seed distribution positions corresponding to the stage images according to segmentation results, obtain distribution information corresponding to each of the stage images, and fuse the distribution information to determine a plurality of focus analysis positions of the seed tray.

3. A machine vision based germinating seed screening device as claimed in claim 1, wherein, The focus analysis module comprises: The preliminary processing unit is configured to establish a planting matrix of the seed tray according to a distribution of the focus analysis positions in the seed tray, mark the focus analysis positions in each of the stage images, divide each of the stage images into a plurality of unit images, and obtain a unit image set corresponding to each of the focus analysis positions. The information identification unit is configured to identify image change features in each of the unit image sets, map each of the image change features on an element position corresponding to the planting matrix, obtain a planting change matrix of the seed tray, and analyze influence features corresponding to different adjacent focus analysis positions according to the planting change matrix. The information processing unit is configured to correct corresponding image change features according to a plurality of influence features corresponding to each of the focus analysis positions, generate image information corresponding to each of the focus analysis positions, and draw a regional dynamic image corresponding to the focus analysis positions according to the image information. The trend analysis unit is configured to identify a seed fixed position corresponding to the focus analysis positions in the regional dynamic image, obtain a dynamic change feature corresponding to the seed fixed position, perform growth analysis on the dynamic change feature according to growth data of the seed, and determine a growth trend of the seed corresponding to the focus analysis positions.

4. A machine vision based germinating seed screening device as claimed in claim 1, wherein, The process of adjusting an estimated germination time period of a corresponding seed according to the current growth deviation information to obtain an accurate germination time period of each of the seeds comprises: When the current growth deviation information is positive, it is determined that the corresponding seed is currently in a positive stimulation state, and a current positive stimulation level of the seed is analyzed according to a deviation amount corresponding to the growth deviation information. The estimated germination time period is moved positively according to the current positive stimulation level to obtain a temporary germination time period of the corresponding seed. When the update growth deviation is positive, a stable stimulation level of an external environment on the corresponding seed is determined according to an update positive stimulation level corresponding to the update growth deviation information and the current positive stimulation level. The growth stimulation speed of the external environment on the seed is analyzed according to the stable stimulation level and the growth trend of the corresponding seed, the length of the temporary germination time period is adjusted according to the growth stimulation speed to obtain the accurate germination time period of the corresponding seed. When the current growth deviation information is negative, the estimated germination time period is regarded as the accurate germination time period of the corresponding seed. When the current growth deviation information is positive and the update growth deviation information of the corresponding seed is negative, the estimated germination time period is regarded as the accurate germination time period of the corresponding seed.

5. A machine vision based germinating seed screening device as claimed in claim 1, wherein, The germination screening module comprises: The sprouting positioning unit is configured to determine a plurality of sprouting seeds showing a sprouting feature in the seed tray according to the seed state, and obtain a planting position corresponding to each of the sprouting seeds respectively; The sprouting marking unit is configured to filter a wavelength of promoting light according to a seed attribute corresponding to the seed, establish a marking cursor by using the promoting light, and mark the planting position by using the marking cursor; The report generation unit is configured to obtain appearance information corresponding to each of the sprouting seeds respectively, generate a seed sprouting report of the seed tray, and transmit the seed sprouting report to a designated terminal for display.

6. A machine vision based germinating seed screening device as claimed in claim 5, wherein, The report generation unit comprises: The visual analysis subunit is configured to obtain a sprouting image corresponding to each of the sprouting seeds respectively, perform image projection on each of the sprouting images respectively to obtain a seed image contained in each of the sprouting images, perform corner point identification on each of the seed images by using a peak detection method, and obtain a plurality of corner point connecting lines contained in each of the seed images; The sprouting classification subunit is configured to map the corner point connecting lines in the corresponding sprouting images to obtain seed crack information and seed water information corresponding to each of the sprouting images, regard a first sprouting seed containing both the seed crack information and the seed water information as a type I sprouting seed, regard a second sprouting seed containing the seed crack information as a type II sprouting seed, and regard a third sprouting seed containing the seed water information as a type III sprouting seed; The report generation unit is configured to calculate a sprouting rate of the seeds in the seed tray, and a number of sprouts corresponding to each type of sprouting seed, establish a seed sprouting report of the seed tray, and transmit the seed sprouting report to a designated terminal for display.

7. A machine vision based germinating seed screening device as claimed in claim 1, wherein, Further comprising: The analysis reminding module is configured to collect an un-sprouting duration corresponding to each of the key analysis positions, determine sprouting identification of a corresponding seed when the un-sprouting duration is higher than a preset duration, and generate reminding information.

8. A method of germinating seed screening based on machine vision, characterized by, Comprising: Step 1: collect a plurality of stage images of a seed tray, mark a seed position in each of the stage images respectively, and determine a plurality of key analysis positions in the seed tray; Step 2: obtain image information of each of the key analysis positions in different stage images respectively, and determine a seed growth trend corresponding to the key analysis position according to the image information; Step 3: analyze an estimated sprouting time period of each seed according to the seed growth sprouting trend, and obtain a seed state feature in the estimated sprouting time period; Step 4: determine a planting position corresponding to each sprouting seed in the seed tray according to the seed state feature, mark the planting position, and generate a seed sprouting report and transmit the seed sprouting report to a designated terminal for display; The step 3 comprises: Step 31: determine a current growth quality of a corresponding seed according to the seed growth trend, list the seeds with the same current growth quality as a same seed type, obtain a planting time of each of the seeds respectively, and determine an already-grown duration of the corresponding seed according to the planting time. Step 32: analyzing the nutrition supply speed of different seeds in each seed category according to the grown duration, analyzing the estimated germination time period of the corresponding seed according to the current growth quality, adding a category label and a time period label to each seed in each seed category according to the estimated germination time period of different seeds in each seed category; Step 33: determining the current growth state and the remaining germination duration of the corresponding seed according to the category label and the time period label of each seed, drawing a growth concept chart of the corresponding seed, and identifying the seed growth information at each unit time in the growth concept chart according to the corresponding growth trend and the nutrition supply speed; Step 34: collecting the real-time growth information of each seed respectively, determining the current growth deviation information of the corresponding seed according to the unit time corresponding to the supervision collection time, adjusting the estimated germination time period of the corresponding seed according to the current growth deviation information, obtaining the accurate germination time period of each seed, and obtaining the seed state feature of the corresponding seed when the supervision collection time coincides with the accurate germination time period.

Images