Corn seed nondestructive testing method based on infrared imaging technology

Abstract

The invention discloses a corn seed nondestructive testing method based on an infrared imaging technology, which comprises the following steps: adding distilled water into a germination box so that the water surface just contacts each to-be-tested corn seed, overlooking and collecting an infrared thermal imaging image in a germination disc, subtracting the reference water temperature mean value from the temperature mean value at the same time for collecting the infrared thermal imaging image to obtain a difference value, and substituting the difference value and the time for collecting the infrared thermal imaging image into a coordinate system to form a temperature difference change curve; converting the filtered temperature difference change curve of the to-be-tested corn seed into a one-dimensional array, taking the one-dimensional array as a training data set, performing label classification on the training data set, and inputting the training data set into a support vector machine (SVM) model for training to obtain a trained SVM model; and inputting the test data set into the trained SVM model for prediction to obtain a prediction result of the test data set. After the corn seed nondestructive testing method is used for testing, the corn seeds to be tested can be continuously planted and used after being dried.

Description

Technical Field

[0001] The invention belongs to the technical field of non-destructive detection of crops, and specifically relates to a non-destructive detection method for corn seeds based on infrared imaging technology. Background Art

[0002] As the main food and feed, corn occupies an important position in my country's economy and production life. Among them, corn seeds are important means of production in the agricultural production process. The quality of corn seeds is a very important influencing factor in corn production. At the same time, the detection and identification of corn seed quality is a key link in the agricultural breeding process. Therefore, the market demand for a method that can quickly and non-destructively detect the quality of corn seeds is increasingly strong.

[0003] Corn seed quality evaluation, vitality and aging detection, purity and authenticity identification, and variety classification research are common problems in corn seed quality testing. Traditional methods for testing corn seed quality include: ATP content determination, conductivity and germination rate determination. Most traditional testing methods require destructive analysis of corn seeds. The tested corn seeds cannot be used after testing. In addition, the testing time is long and the process is complicated, which makes it difficult to meet the requirements of modern agriculture for non-destructive and rapid production of corn seeds. Summary of the invention

[0004] In view of the deficiencies in the prior art, an object of the present invention is to provide a non-destructive detection method for corn seeds based on infrared imaging technology, which is used to quickly detect whether corn seeds are aging.

[0005] A nondestructive detection method for corn seeds based on infrared imaging technology comprises the following steps:

[0006] S1, dividing a plurality of corn seeds into two portions to obtain a first portion of corn seeds and a second portion of corn seeds, aging and sterilizing the first portion of corn seeds in sequence to obtain a first group of corn seeds to be tested consisting of a plurality of corn seeds to be tested; sterilizing the second portion of corn seeds to obtain a second group of corn seeds to be tested consisting of a plurality of corn seeds to be tested;

[0007] In step S1 , the aging is performed by a high temperature and high humidity aging method.

[0008] S2, placing the first group of corn seeds to be tested and the second group of corn seeds to be tested in a germination tray of a germination box, placing one corn seed to be tested in each grid of the germination tray, adding distilled water to the germination box so that the water surface just touches each corn seed to be tested, placing the germination box in a constant temperature water bath, and setting the temperature of the constant temperature water bath to 25-30° C., placing the constant temperature water bath at room temperature, and performing step S3 after 10-30 minutes;

[0009] S3, using an infrared thermal imager to collect infrared thermal imaging images in the germination tray at intervals of 0.5 to 2 minutes, the time for collecting infrared thermal imaging images: starting from the start time of collecting infrared thermal imaging images, a total of T hours are collected, T = 8 to 12;

[0010] S4, including S4-1 and S4-2:

[0011] S4-1, select a frame in the infrared thermal imaging image to obtain a first area and a reference water temperature area: take the area in the grid where each corn seed to be tested is located in the infrared thermal imaging image as the first area, and take the area in the germination tray of the infrared thermal imaging image of the corn seed to be tested where there is only water but no corn seed to be tested as the reference water temperature area; calculate the grayscale mean of the first area of ​​each corn seed to be tested and use it as the temperature mean, calculate the grayscale mean of the reference water temperature area and use it as the reference water temperature mean, subtract the reference water temperature mean from the temperature mean of each corn seed to be tested at the same time of collecting the infrared thermal imaging image to obtain a difference, substitute the difference of each corn seed to be tested and the time of collecting the infrared thermal imaging image into the coordinate system to form a temperature difference change curve;

[0012] S4-2, the temperature difference change curve of each corn seed to be tested is filtered using the EMD algorithm to obtain the filtered temperature difference change curve

[0013] In step S4 - 1 , the area of ​​the reference water temperature region is greater than three times the area of ​​the first region.

[0014] In step S4-2, the EMD algorithm decomposes the temperature difference change curve of a single corn seed to be tested into 11 intrinsic mode functions IMFs of different frequencies in order from low to high frequencies of 0-1kHz, removes the intrinsic mode function IMF of the lowest frequency point and the intrinsic mode function IMF of the highest frequency point among the 11 intrinsic mode functions IMFs of different frequencies, and superimposes and normalizes the intrinsic mode functions IMFs of the remaining 9 different frequencies to obtain the filtered temperature difference change curve of the single corn seed to be tested.

[0015] S5, converting the filtered temperature difference change curve of each corn seed to be tested in the first group of corn seeds to be tested and the second group of corn seeds to be tested into a one-dimensional array, using the one-dimensional array of part of the corn seeds to be tested in the first group of corn seeds to be tested and part of the corn seeds to be tested in the second group of corn seeds to be tested as a training data set and performing label classification on it, using the one-dimensional array of the remaining corn seeds to be tested in the first group of corn seeds to be tested and the remaining corn seeds to be tested in the second group of corn seeds to be tested as a test data set, inputting the training data set into a support vector machine SVM model for training, and after 25 iterations, obtaining a trained SVM model;

[0016] S6, input the test data set into the trained SVM model for prediction, and obtain the prediction result of the test data set.

[0017] Compared with the prior art, the present invention has the following beneficial effects:

[0018] 1. Compared with the traditional conductivity determination method and TTC quantitative measurement method, the corn seed nondestructive detection method of the present invention causes less damage to the corn seeds to be tested. After being tested by the corn seed nondestructive detection method of the present invention, the corn seeds to be tested can continue to be planted and used after being dried.

[0019] 2. Compared with various existing near-infrared and multi-spectral detection methods, the detection equipment involved in the corn seed non-destructive detection method of the present invention is low-cost and does not require the addition of expensive detection reagents and detection consumables.

[0020] 3. Compared with the existing traditional seed detection methods, the corn seed nondestructive detection method of the present invention has lower requirements on the detection environment, shorter detection time and higher detection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a photo of the corn seeds to be tested placed in a germination tray;

[0022] Figure 2 is the temperature difference variation curve;

[0023] Figure 3 It is an infrared thermal imaging image;

[0024] Figure 4 These are the waveforms of the intrinsic mode functions (IMFs) at 11 different frequencies. DETAILED DESCRIPTION

[0025] The nondestructive detection method for corn seeds of the present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0026] Example 1

[0027] A nondestructive detection method for corn seeds based on infrared imaging technology comprises the following steps:

[0028] S1, dividing a plurality of corn seeds that are not shrunken, damaged, or moldy into two portions to obtain a first portion of corn seeds and a second portion of corn seeds, aging and sterilizing the first portion of corn seeds in sequence to obtain a first group of corn seeds to be tested consisting of M1 corn seeds to be tested; sterilizing the second portion of corn seeds to obtain a second group of corn seeds to be tested consisting of M2 corn seeds to be tested;

[0029] S2, placing the first group of corn seeds to be tested and the second group of corn seeds to be tested in the germination tray of the germination box (the germination tray is kept horizontal), as shown in FIG. Figure 1As shown, a corn seed to be tested is placed in each grid of a germination tray (a filter paper is placed under the germination tray to ensure moisture after adding water), distilled water is added to the germination box so that the water surface just touches each corn seed to be tested, the germination box is placed in a constant temperature water bath, and the temperature of the constant temperature water bath is set to 27°C, the constant temperature water bath is placed in an environment of 25°C, and the germination box in the constant temperature water bath is located directly below the infrared thermal imager lens, the distance between the infrared thermal imager lens and the germination tray is 30 cm (there is no obstruction between the corn seed to be tested and the infrared thermal imager lens), and step S3 is performed after 20 minutes;

[0030] S3, using an infrared thermal imager to collect infrared thermal imaging images in the germination tray at intervals of 1 minute. The time for collecting infrared thermal imaging images is 12 hours starting from the start time of collecting infrared thermal imaging images.

[0031] S4, including S4-1 and S4-2:

[0032] S4-1, select a frame in the infrared thermal imaging image to obtain a first area and a reference water temperature area: the area in the grid where each corn seed to be tested is located in the infrared thermal imaging image is taken as the first area, and the area in the germination tray of the infrared thermal imaging image of the corn seed to be tested where there is only water but no corn seed to be tested is taken as the reference water temperature area (in this embodiment, the middle long strip of the germination tray is selected), and the area of ​​the reference water temperature area is greater than 3 times the area of ​​the first area, such as Figure 3 shown.

[0033] The grayscale mean of the first area of ​​each corn seed to be tested is calculated and used as the temperature mean. The grayscale mean of the reference water temperature area of ​​each corn seed to be tested is calculated and used as the reference water temperature mean. The reference water temperature mean is subtracted from the temperature mean of each corn seed to be tested at the same time of collecting infrared thermal imaging images to obtain the difference of the corn seed to be tested. The difference of each corn seed to be tested and the time of collecting infrared thermal imaging images are substituted into the coordinate system to form a temperature difference change curve, such as Figure 2 As shown, each curve is a temperature difference change curve of a single corn seed to be tested within 12 hours, the horizontal axis is the time for collecting infrared thermal imaging images, and the vertical axis is the difference value, among which the upper one is the temperature difference change curve of some corn seeds to be tested in the second group of corn seeds to be tested, and the lower one is the temperature difference change curve of some corn seeds to be tested in the first group of corn seeds to be tested.

[0034] S4-2, the temperature difference change curve of each corn seed to be tested is filtered using the EMD algorithm (Ne.H, Sr.L, Mlc.W, et al. The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis [J]. Proceedings of the Royal Society. Mathematical, physical and engineering sciences, 1998 (1971): 454.) to obtain a filtered temperature difference change curve, wherein the EMD algorithm decomposes the temperature difference change curve of a single corn seed to be tested into 11 eigenmode functions IMF at different frequencies in the order of 0-1kHz frequency from low to high, wherein the waveforms of the eigenmode functions IMF at 11 different frequencies are as follows: Figure 4 As shown, Figure 4 The middle horizontal axis is the time of collecting infrared thermal imaging images, the left vertical axis is the difference of each intrinsic mode function IMF after decomposition, and the right vertical axis is the frequency point; the intrinsic mode function IMF of the lowest frequency point (0Hz) and the intrinsic mode function IMF of the highest frequency point (1kHz) among the 11 intrinsic mode functions IMF of different frequencies are removed, and the intrinsic mode functions IMF of the remaining 9 different frequencies are superimposed and normalized to obtain the filtered temperature difference change curve of the single corn seed to be tested.

[0035] S5, converting the filtered temperature difference change curve of each corn seed to be tested in the first group of corn seeds to be tested and the second group of corn seeds to be tested into a one-dimensional array, taking the one-dimensional array of part of the corn seeds to be tested in the first group of corn seeds to be tested and part of the corn seeds to be tested in the second group of corn seeds to be tested as a training data set and performing label classification on them, taking the one-dimensional array of the remaining corn seeds to be tested in the first group of corn seeds to be tested and the remaining corn seeds to be tested in the second group of corn seeds to be tested as a test data set, inputting the training data set into a support vector machine SVM model (Wang Haiyan, Li Jianhui, Yang Fenglei. A review of support vector machine theory and algorithm research [J]. Computer Application Research, 2014, 31(05): 1281-1286.) for training, and obtaining a trained SVM model after 25 iterations;

[0036] S6, input the test data set into the trained SVM model for prediction, and obtain the prediction result of the test data set.

[0037] Example 2

[0038] A nondestructive detection method for corn seeds based on infrared imaging technology. On the basis of Example 1, the aging adopts the high temperature and high humidity aging method (Zhao Xinxin, Liu Jiquan, Wang Qi, et al. Study on the response of corn seeds to high temperature and high humidity aging [J]. Jilin Agricultural Science, 2015, 40 (2): 11-15.), and the specific operation is as follows: 40 mL of distilled water is added to a plastic aging box, and a grid is placed in the plastic aging box to ensure that water does not penetrate into the grid and the corn seeds placed later; the corn seeds are weighed and recorded; the corn seeds are placed on the sterilized grid and the corn seeds are flattened into a layer, and each corn seed does not overlap to ensure that each The invention discloses a method for preparing corn seeds for aging treatment. The method comprises the following steps: setting the temperature and humidity of the aging box, in which the temperature is set to 41° C. and the humidity is set to 100% in the present embodiment; placing the plastic aging box in the aging box, and performing aging treatment on the corn seeds for 72 hours; taking out the corn seeds after aging treatment and weighing the corn seeds after aging treatment immediately; comparing the weight of the corn seeds and the corn seeds after aging treatment, and calculating that the water content of the corn seeds after aging treatment is 31-35%; placing the corn seeds after aging treatment at room temperature for 2 days to reduce the water content of the corn seeds after aging treatment to 13%, thereby obtaining corn seeds after aging treatment.

[0039] The disinfection steps include: selecting corn seeds with full grains, consistent size, shape and maturity from aged corn seeds and unaged corn seeds; soaking the selected corn seeds in a 2% sodium hypochlorite (NaClO) solution for 10 minutes; and rinsing the soaked corn seeds with distilled water to complete the disinfection.

[0040] Example 3

[0041] A nondestructive detection method for corn seeds based on infrared imaging technology. On the basis of Example 2, the varieties of corn seeds are the same, all of which are Su Ketian 1506 (purchased from Shouguang Beishu Seed Co., Ltd.).

[0042] In this embodiment, the training data set includes a one-dimensional array of 126 corn seeds to be tested in the second group of corn seeds to be tested and a one-dimensional array of 222 corn seeds to be tested in the first group of corn seeds to be tested;

[0043] The test data set includes a one-dimensional array of 95 corn seeds to be tested in the second group of corn seeds to be tested and a one-dimensional array of 54 corn seeds to be tested in the first group of corn seeds to be tested;

[0044] Among them, the label classification is aged and not aged; the prediction results are aged and not aged.

[0045] After training, the SVM model predicts the test data set and obtains the prediction results of the test data set: the prediction accuracy of the second group of corn seeds to be tested is 97.9%, the prediction accuracy of the first group of corn seeds to be tested is 100%, and the total average accuracy is 98.6577%.

[0046] Example 4

[0047] A nondestructive detection method for corn seeds based on infrared imaging technology, based on Example 2, the varieties of corn seeds are Jintian Shun Liuheliu (purchased from Nanping Xiaofu Seed Co., Ltd.) and Su Ketian 1506 (purchased from Shouguang Beishu Seed Co., Ltd.);

[0048] In this embodiment, the training data set in step S5 includes: a one-dimensional array of a first group of corn seeds to be tested (63 grains of Jintian Shun 666 corn seeds to be tested and 63 grains of Sukotian 1506 corn seeds to be tested) and a second group of corn seeds to be tested (123 grains of Jintian Shun 666 corn seeds to be tested and 99 grains of Sukotian 1506 corn seeds to be tested); the test data set includes: a one-dimensional array of a first group of corn seeds to be tested (27 grains of Jintian Shun 666 corn seeds to be tested and 27 grains of Sukotian 1506 corn seeds to be tested) and a second group of corn seeds to be tested (53 grains of Jintian Shun 666 corn seeds to be tested and 42 grains of Sukotian 1506 corn seeds to be tested).

[0049] The labels are divided into four categories according to whether they have been aged and the variety. Label classification method: the one-dimensional array of corn seeds that have not been aged and are tested by Su Ketian 1506 is marked as label 1, the one-dimensional array of corn seeds that have not been aged and are tested by Jin Tian Shun 666 is marked as label 2, the one-dimensional array of corn seeds that have been aged and are tested by Jin Tian Shun 666 is marked as label 3, and the one-dimensional array of corn seeds that have been aged and are tested by Su Ketian 1506 is marked as label 4;

[0050] After training, the SVM model performs classification prediction on the test data set, and the prediction results of distinguishing the variety of corn seeds to be tested and whether they are aged are as follows: the prediction accuracy of label 1 is 100%; the prediction accuracy of label 2 is 91.4%; the prediction accuracy of label 3 is 94.7%; the prediction accuracy of label 4 is 83.9%; and the total average accuracy is 92.6174%.

[0051] The present invention is described above by way of example. It should be noted that, without departing from the core of the present invention, any simple deformation, modification or other equivalent replacement that can be made by those skilled in the art without inventive effort falls within the protection scope of the present invention.

Claims

1. A nondestructive detection method for corn seeds based on infrared imaging technology, characterized in that: The following steps are involved: S1, dividing a plurality of corn seeds into two portions to obtain a first portion of corn seeds and a second portion of corn seeds, and sequentially aging and sterilizing the first portion of corn seeds to obtain a first group of corn seeds to be tested consisting of a plurality of corn seeds to be tested; sterilizing the second batch of corn seeds to obtain a second group of corn seeds to be tested consisting of a plurality of corn seeds to be tested; S2, placing the first group of corn seeds to be tested and the second group of corn seeds to be tested in a germination tray of a germination box, placing one corn seed to be tested in each grid of the germination tray, adding water to the germination box so that the water surface just touches each corn seed to be tested, keeping the germination box at a constant temperature of 25-30° C., and executing step S3 after 10-30 minutes; S3, using an infrared thermal imager to collect infrared thermal imaging images in the germination tray at intervals of 0.5 to 2 minutes, and collecting infrared thermal imaging images for a total of T hours starting from the start time of collecting infrared thermal imaging images, T = 8 to 12; S4, including S4-1 and S4-2: S4-1, select a frame in the infrared thermal imaging image to obtain a first area and a reference water temperature area: the area in the grid where each corn seed to be tested is located in the infrared thermal imaging image is taken as the first area, and the area in the germination tray of the infrared thermal imaging image of the corn seed to be tested where there is only water but no corn seed to be tested is taken as the reference water temperature area; Calculate the grayscale mean of the first area of ​​each corn seed to be tested and use it as the temperature mean, calculate the grayscale mean of the reference water temperature area of ​​each corn seed to be tested and use it as the reference water temperature mean, subtract the reference water temperature mean from the temperature mean of each corn seed to be tested at the same time of collecting infrared thermal imaging images to obtain a difference, substitute the difference of each corn seed to be tested and the time of collecting infrared thermal imaging images into a coordinate system to form a temperature difference change curve; S4-2, the temperature difference change curve of each corn seed to be tested is filtered using the EMD algorithm to obtain the filtered temperature difference change curve S5, converting the filtered temperature difference change curve of each corn seed to be tested in the first group of corn seeds to be tested and the second group of corn seeds to be tested into a one-dimensional array, using the one-dimensional array of the first group of corn seeds to be tested and the second group of corn seeds to be tested as a training data set and performing label classification on the training data set, inputting the training data set into a support vector machine SVM model for training, and obtaining a trained SVM model; S6, inputting the one-dimensional array of the sample to be tested into the trained SVM model for prediction to obtain a prediction result.

2. The method for nondestructive testing of corn seeds according to claim 1, characterized in that: In step S1 , the aging is performed by a high temperature and high humidity aging method.

3. According to the corn seed nondestructive testing method according to claim 1, in step S4-1, the area of ​​the reference water temperature zone is greater than 3 times the area of ​​the first zone.

4. The nondestructive detection method for corn seeds according to claim 1, in step S4-2, the temperature difference change curve of a single corn seed to be tested is evenly decomposed into 11 intrinsic mode functions IMF of different frequencies in the order from 0-1kHz frequency to high, the intrinsic mode function IMF of the lowest frequency and the intrinsic mode function IMF of the highest frequency among the 11 intrinsic mode functions IMF of different frequencies are removed, and the intrinsic mode functions IMF of the remaining 9 different frequencies are superimposed and normalized to obtain the filtered temperature difference change curve of the single corn seed to be tested.

Images