Seed activity prediction method based on comprehensive characteristics of internal components and external phenotypes

Abstract

The application relates to the field of seed activity detection, and in particular to a seed activity prediction method based on comprehensive characteristics of internal components and external phenotypes, which comprises the following steps: arranging multiple cameras around seeds, taking pictures of the seeds through the multiple cameras, and obtaining seed images representing external phenotypes; performing external phenotype analysis on the seed images to obtain image analysis results containing external phenotype characteristics of the seeds; emitting a detection spectrum to the seeds through a light source system, and forming a transmission spectrum carrying component information of the seeds through the seeds; collecting the transmission spectrum by a spectrometer system; analyzing the transmission spectrum to obtain spectral analysis results containing seed components; comprehensively evaluating the image analysis results and the spectral analysis results to obtain quantitative results, and classifying the activity of the seeds according to a set target threshold value. The application combines the external phenotype and the internal molecular components of the seeds, comprehensively analyzes and judges the activity of the seeds, and improves the accuracy of the seed activity judgment.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of seed activity detection, and particularly relates to a seed activity prediction method based on comprehensive characteristics of internal components and external phenotypes. BACKGROUND

[0002] At present, seed activity detection mainly uses a spectrum measurement method to obtain component information of seeds, and further predicts seed activity through threshold discrimination, algorithm analysis and other means. The existing technology takes spectrum information as the basis for judgment, which actually analyzes seed components at the molecular level. However, there are a certain amount of seeds with low activity and germination defects, which are not caused by molecular level reasons, such as a seed being broken in the middle, resulting in seed damage and inability to germinate, but the parts thereof do not change at the molecular level, resulting in spectrum information being almost identical to that of high-activity seeds. In addition, the activity of some deformed, moldy and too small seeds is very low, but the change at the molecular level is not obvious, resulting in no obvious spectrum characteristics, and it is difficult to distinguish using the spectrum method. Therefore, the single spectrum detection method has the problem of poor accuracy of seed activity detection. SUMMARY

[0003] Therefore, the application creates a positioning and feeding device and method suitable for seed activity detection to solve the problem of poor accuracy of seed activity detection caused by single spectrum detection.

[0004] To achieve the above-mentioned purpose, the technical solution of the application creates is as follows:

[0005] A seed activity prediction method based on comprehensive characteristics of internal components and external phenotypes, comprising the following steps:

[0006] External phenotype shooting: a plurality of cameras are arranged around the seed, and the seed is shot by the plurality of cameras to obtain a seed image representing the external phenotype;

[0007] Image analysis: the seed image is analyzed for the external phenotype of the seed to obtain an image analysis result containing the external phenotype characteristics of the seed; the analysis content includes image enhancement processing, target positioning processing and feature extraction processing in sequence, the image enhancement processing is used to optimize the quality of the seed image, the target positioning processing is used to position the position of the seed from the optimized seed image, and the feature extraction processing is used to obtain the feature information of the seed by using the methods of frequency domain analysis, connected domain color subdivision and edge derivation;

[0008] Spectrum acquisition: a detection spectrum is emitted to the seed by a light source system, the detection spectrum forms a transmission spectrum carrying component information of the seed after penetrating the seed, and the transmission spectrum is collected by a spectrometer system;

[0009] Spectrum analysis: the transmission spectrum is analyzed to obtain a spectrum analysis result containing seed components; the analysis content includes data cleaning processing, denoising processing, and normalization processing; the data cleaning processing is used to screen qualified spectrum information, the denoising processing is used to filter unqualified spectrum information, and the normalization processing is used to normalize the filtered spectrum information;

[0010] Comprehensive processing analysis: the image analysis result and the spectrum analysis result are comprehensively evaluated to obtain a quantitative result, and the activity of the seed is classified according to a set target threshold.

[0011] Further, the data cleaning processing uses the following conditions to screen the spectrum information:

[0012]

[0013] wherein A MAX , A MIN are the maximum value and the minimum value in the spectrum curve respectively, and p, q, m, n are constants;

[0014] The spectrum information that does not meet the above conditions is determined as unqualified.

[0015] Further, the filtering mode used in the denoising processing is as follows:

[0016]

[0017] wherein A n is the value of a point in the unqualified spectrum curve, A iF is the filtered value of the point in the unqualified spectrum curve, and 2t+1 is the range of the filter window, and the value of t is preferably 10-15 nm.

[0018] Further, the formula used in the normalization processing is as follows:

[0019]

[0020] wherein A iFN is the normalized spectrum intensity, A IfMAX , A IFMIN are the maximum value and the minimum value of the filtered spectrum curve respectively.

[0021] Further, the formula used in the comprehensive evaluation is as follows:

[0022]

[0023]

[0024] wherein J is the image analysis result, K is the spectrum analysis result, JNML K is a value processed by image analysis result NML J is a value processed by spectrum analysis result MAX K is a total maximum value of image analysis result MAX J is a total maximum value of spectrum analysis result MIM K is a total minimum value of image analysis result MIM J is a total minimum value of spectrum analysis result, X is a quantization result, and a and b are weight factor constants of spectrum analysis and image analysis, respectively.

[0025] Further, the camera adopts a color camera with a resolution of 720p or above, the color camera has a field of view of 1.5-3 times the size of the seed cross section, and the number of color cameras is 6-8.

[0026] Further, the light source system is coaxial with the spectrometer system, and the axis passes through the center of the seed.

[0027] Further, the light source system is composed of a light emitter, an optical fiber and a lens.

[0028] Further, the spectrometer system is composed of a spectrometer, an optical fiber and a lens.

[0029] Further, the image enhancement processing uses an adaptive image parameter adjustment algorithm or a Gaussian image filtering method to optimize the quality of the seed image, and the target positioning processing uses a gradient method, a threshold method or a connected domain algorithm to extract the edge information of the seed.

[0030] Compared with the prior art, the present application can achieve the following beneficial effects:

[0031] The present application combines the external phenotype and internal molecular composition of the seed by obtaining the image information and spectrum information of the seed, comprehensively analyzes and judges the activity of the seed, makes the analysis basis more comprehensive and the analysis process more advanced and reasonable, and finally realizes more accurate activity judgment results. DETAILED DESCRIPTION

[0032] The accompanying drawings, which form a part of the present application, are included to provide a further understanding of the application and are incorporated herein for purposes of illustration. The embodiments of the present application, as well as a summary of the application, are described below in conjunction with the accompanying drawings. In the drawings:

[0033] Fig. 1 FIG. 1 is a flowchart of a seed activity prediction method based on comprehensive characteristics of internal composition and external phenotype according to an embodiment of the present application;

[0034] Fig. 2 FIG. 2 is a distribution diagram of a camera according to an embodiment of the present application;

[0035] Fig. 3A distribution diagram of a light source system and a spectrometer system for creating an embodiment of the present invention.

[0036] Explanation of reference signs:

[0037] Light source system 1, probe spectrum 2, seed 3, transmission spectrum 4, spectrometer system 5, camera 6. DETAILED DESCRIPTION

[0038] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and do not constitute a limitation on the present invention.

[0039] It should be noted that the embodiments in the present invention and the features in the embodiments can be combined with each other without conflict.

[0040] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, and therefore cannot be understood as indicating or implying that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present invention. In addition, the terms "first", "second" and the like are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined with "first", "second" and the like can explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, the meaning of "a plurality of" is two or more.

[0041] In the description of the present invention, it should be noted that unless otherwise explicitly specified and limited, the terms "assembly", "connection", "connection" should be understood broadly, for example, it can be a fixed connection, or a detachable connection, or an integral connection; it can be a mechanical connection, or an electrical connection; it can be directly connected, or indirectly connected through an intermediate medium, or the internal communication of two elements. For those skilled in the art, the specific meaning of the above terms in the present invention can be understood according to the specific circumstances.

[0042] The present invention will be described in detail below with reference to the drawings and embodiments.

[0043] As Figs. 1-3As shown, the embodiment of the present application provides a seed activity prediction method based on the comprehensive characteristics of internal components and external phenotypes, comprising the following steps:

[0044] External phenotype shooting: a plurality of cameras 6 are arranged around the seed 3, the seed 3 is shot by the plurality of cameras 6, and a seed image representing the external phenotype is obtained.

[0045] Each camera 6 is directly opposite the seed 3, the number of cameras 6 is preferably 2-6, and the seed 3 is centered and surrounded by the cameras 6. The field of view of the camera 6 should be adapted to the size of the seed 3, and the size of the cross section of the seed 3 is preferably 1.5-3 times. The camera 6 can be driven by using a programmable device such as a computer, a PLC, a microcontroller, an FPGA, etc. The camera 6 should use an RGB format color camera with a resolution of 720p or above, and should have a data communication interface to transmit data.

[0046] Image analysis: the external phenotype of the seed 3 is analyzed on the seed image to obtain an image analysis result containing the external phenotype characteristics of the seed; the analysis content includes image enhancement processing, target positioning processing, and feature extraction processing in sequence, the image enhancement processing is used to optimize the quality of the seed image, the target positioning processing is used to locate the position of the seed 3 from the optimized seed image, and the feature extraction processing is used to obtain the feature information of the seed by using the methods of frequency domain analysis, connected domain color subdivision, and edge derivation.

[0047] The image analysis step is realized by using an algorithm, and can rely on a computer, a workstation, a cloud computer, an embedded computer, a programmable logic system, and a programmable logic controller.

[0048] The image enhancement processing is used to optimize the imaging quality, and can be realized by using an adaptive image parameter adjustment method and a Gaussian image filtering method.

[0049] The target positioning processing is used to determine the area where the seed 3 is located from the whole seed image, that is, to locate the position of the seed 3 in the seed image, and the target positioning processing can use a gradient method, a threshold method, and a connected domain method to extract the edge information of the seed 3.

[0050] The feature extraction processing is used to obtain the feature information of the seed 3 by using the methods of frequency domain analysis, connected domain color subdivision, and edge derivation after determining the area where the seed 3 is located, and the feature information of the seed includes shape, color, size, etc. Through the feature information of the seed 3, whether the seed has problems such as cracking, deformation, mildew, and being too small can be further analyzed.

[0051] The image analysis step can also include machine learning analysis, which is a result analysis of the seed 3 through the seed image itself and the feature information of the seed 3, which can be implemented using one-dimensional or two-dimensional neural networks. In particular, for a two-dimensional context correlation neural network algorithm, target positioning and feature extraction can also not be performed. For seeds 3 with some obvious features, the results of feature extraction can be directly quantified as image analysis results without using the step of machine learning analysis.

[0052] Spectrum acquisition: The probe spectrum 2 is emitted to the seed 3 by the light source system 1, the probe spectrum 2 transmits through the seed 3 to form a transmission spectrum 4 carrying seed component information, and the transmission spectrum 4 is collected by the spectrometer system 5.

[0053] The light source system 1 and the spectrometer system 5 are distributed on both sides of the seed 3, the light paths of the light source system 1 and the spectrometer system 5 are coaxial and the axis passes through the center of the seed 3.

[0054] The light source system 1 is a light-emitting system composed of a light-emitting body, an optical fiber, a lens, etc., and needs to meet the ability to emit a continuous spectrum light beam, and its power should be between 0.2-1.5mW. Its spectral range should be one of visible-near infrared, near infrared, near infrared-middle infrared, near infrared-middle infrared-far infrared, and middle infrared-far infrared.

[0055] The spectrometer system 5 is a spectral measurement system composed of a spectrometer, an optical fiber, a lens, etc., and needs to have a spectral detection range suitable for the light source system 1. It should have a data communication interface to transmit data.

[0056] Spectral analysis: The transmission spectrum is analyzed to obtain spectral analysis results containing seed components; the analysis content includes data cleaning processing, denoising processing, and normalization processing, the data cleaning processing is used to screen qualified spectral information, the denoising processing is used to filter unqualified spectral information, and the normalization processing is used to normalize the filtered spectral information.

[0057] The spectral analysis step is implemented using algorithms, which can rely on computers, workstations, cloud computers, embedded computers, programmable logic systems, and programmable logic controllers.

[0058] The data cleaning processing uses the following conditions to screen the spectral information:

[0059]

[0060] Where, A MAX , A MIN are the maximum and minimum values in the spectral curve, respectively, and p, q, m, n are constants defined by the maximum and minimum values of a large number of valid and correct spectral statistical values.

[0061] Spectrum information not meeting the above conditions is determined as unqualified, and then the spectrum curve is filtered by the following formula:

[0062]

[0063] Wherein, A n is the value of a point in the unqualified spectrum curve, A iF is the filtered value of a point in the unqualified spectrum curve, and 2t+1 is the range of the filter window, and the value of t is preferably selected to cover the spectrum range of 10-15nm.

[0064] The filtered spectrum data is normalized, and the formula used for normalization is as follows:

[0065]

[0066] Wherein, A iFN is the normalized spectrum intensity, A IfMAX and A IFMIN are the maximum and minimum values of the filtered spectrum curve respectively.

[0067] Then the preliminary conclusion of the spectrum analysis step can be obtained by machine learning. Machine learning can use methods such as support vector machine, partial least squares, decision tree, principal component analysis classification, neural network learning, etc.

[0068] Comprehensive processing and analysis: The image analysis result and the spectrum analysis result are comprehensively evaluated to obtain a quantitative result, and the activity of the seeds is classified according to the set target threshold.

[0069] The comprehensive processing and analysis step is realized by using algorithms, which can rely on computers, workstations, cloud computers, embedded computers, programmable logic systems, programmable logic controllers.

[0070] Let the image analysis result obtained by image analysis be J, and the spectrum analysis result obtained by spectrum analysis be K. First, the image analysis result J and the spectrum analysis result K need to be processed to obtain the same evaluation scale, and the following calculation method is used for implementation:

[0071]

[0072] Wherein, J NML is the processed value of the image analysis result, K NML is the processed value of the spectrum analysis result, J MIM is the total minimum value of the image analysis result, K MIM is the total minimum value of the spectrum analysis result, J MAX is the total maximum value of the image analysis result, and K MAXTotal maximum value of the spectrum analysis result.

[0073] Then the above results are analyzed comprehensively by using the following formula:

[0074]

[0075] Wherein, X is the quantification result, a and b are weight factor constants of spectrum analysis and spectrum analysis respectively, and the weights of a and b are determined according to the good or bad of the sorting result of different weights.

[0076] For the 'germination' and 'non-germination' classification according to the active quantification prediction result, the final quantification result X can be compared and judged in size, if the quantification result X exceeds the set target threshold, the seed is defined as germinable, if the quantification result X does not exceed the set target threshold, the seed is defined as non-germinable.

[0077] The two varieties of rice seeds are sorted by germinability using the technical solution, and compared with the prior art, the germination rates can be respectively increased from 85.0%, 89.3% to 94.1%, 96.5%, it can be seen that the technical solution can indeed improve the accuracy of activity prediction, and achieve the beneficial technical purpose.

[0078] It should be understood that various forms of the flow shown above can be used to reorder, add or delete steps. For example, each step recorded in the present disclosure can be executed in parallel, sequentially or in different order, as long as the desired results of the technical solution of the present disclosure can be achieved, which is not limited herein.

[0079] The above specific embodiments do not constitute a limitation on the scope of protection of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for predicting seed viability based on the comprehensive characteristics of internal components and external phenotypes, characterized in that, Comprising the following steps: External phenotype shooting: arranging multiple cameras around the seed, shooting the seed through the multiple cameras, and obtaining a seed image representing the external phenotype; Image analysis: analyzing the seed image for the external phenotype of the seed, and obtaining an image analysis result containing the external phenotype characteristics of the seed; the analysis includes image enhancement processing, target positioning processing, and feature extraction processing in sequence, the image enhancement processing is used to optimize the quality of the seed image, the target positioning processing is used to locate the position of the seed from the optimized seed image, and the feature extraction processing is used to obtain the feature information of the seed by using the methods of frequency domain analysis, connected domain color subdivision, and edge derivation; Spectrum acquisition: emitting a detection spectrum to the seed through a light source system, the detection spectrum passing through the seed to form a transmission spectrum carrying the component information of the seed, and the transmission spectrum being collected by a spectrometer system; Spectrum analysis: analyzing the transmission spectrum, and obtaining a spectrum analysis result containing the components of the seed; the analysis includes data cleaning processing, denoising processing, and normalization processing, the data cleaning processing is used to filter qualified spectrum information, the denoising processing is used to filter unqualified spectrum information, and the normalization processing is used to normalize the filtered spectrum information; Comprehensive processing and analysis: comprehensively evaluating the image analysis result and the spectrum analysis result, obtaining a quantitative result, and classifying the activity of the seed according to a set target threshold; the formula used in the comprehensive evaluation is as follows: wherein J is the image analysis result, K is the spectral analysis result, J NML is the processed value of the image analysis result, K NML is the processed value of the spectral analysis result, J MAX is the total statistical maximum value of the image analysis result, K MAX is the total statistical maximum value of the spectral analysis result, J MIM is the total statistical minimum value of the image analysis result, K MIM is the total statistical minimum value of the spectral analysis result, X is the quantification result, and a and b are weight factor constants of the spectral analysis and the spectral analysis, respectively.

2. The method for predicting seed activity based on the integrated characteristics of internal components and external phenotypes according to claim 1, characterized in that, The data cleaning processing uses the following conditions to filter the spectrum information: where A MAX , A MIN are the maximum and minimum values in the spectral curve, respectively, and p, q, m, n are constants. Spectrum information that does not meet the above conditions is determined as unqualified.

3. The method for predicting seed activity based on the integrated characteristics of internal components and external phenotypes according to claim 2, characterized in that, The filtering method used in the denoising processing is as follows: Wherein, A n is the value of a certain point in the unqualified spectral curve, A iF is the filtered value of a certain point in the unqualified spectral curve, 2t+1 is the range of the filter window, and the value of t is selected such that the filter window covers the spectral range of 10-15 nm.

4. The method for predicting seed activity based on the comprehensive characteristics of internal components and external phenotypes according to claim 3, characterized in that, The formula used in the normalization processing is as follows: wherein A iFN is the normalized spectral intensity, A IfMAX , A IFMIN are the maximum and minimum values of the filtered spectral curve, respectively.

5. The method for predicting seed activity based on the integrated characteristics of internal components and external phenotypes according to claim 1, characterized in that, The camera uses a color camera with a resolution of 720p or above, the field of view of the color camera is 1.5-3 times the size of the seed cross section, and the number of color cameras is 6-8.

6. The method for predicting seed activity based on the integrated characteristics of internal components and external phenotypes according to claim 1, characterized in that, The light source system and the spectrometer system are coaxial, and the axis passes through the center of the seed.

7. The method for predicting seed activity based on the integrated characteristics of internal components and external phenotypes according to claim 6, characterized in that, The light source system is composed of a light emitter, an optical fiber, and a lens.

8. The method for predicting seed activity based on the integrated characteristics of internal components and external phenotypes according to claim 6, characterized in that, The spectrometer system is composed of a spectrometer, an optical fiber, and a lens.

9. The method for predicting seed activity based on the integrated characteristics of internal components and external phenotypes according to claim 1, characterized in that, The image enhancement processing uses an adaptive image parameter adjustment algorithm or a Gaussian image filtering method to optimize the quality of the seed image, and the target positioning processing uses a gradient method, a threshold method, or a connected domain algorithm to extract the edge information of the seed.

Images