CN103810522B - Counting method and device for corn ear grains - Google Patents

Abstract

The invention provides a counting method and a counting device for corn ear grains. The method comprises the following steps: step S1, obtaining an estimation model for the corn ear grains, wherein the estimation model for the corn ear grains is used for showing a function relation between the number of the ear grains on the single surface of corn and the number of all ear grains; step 2, obtaining the single sided image of the corn, and preprocessing the single sided image to obtain a corresponding gray level image; step S3, performing multi-threshold segmentation on the gray level image obtained in the step S2, so as to obtain a corresponding binary image; step S4, counting communicating areas with the same gray values, and obtaining the number of the communication areas corresponding to the grey values for showing the corn grains; and step S5, determining all ear grains according to the number obtained in the step S4 and the corn grain estimation model in the step S1. By adopting the counting method for the corn ear grains, the corn ear grains can be quickly and conveniently counted.

Description

Corn ear and grain counting method and device

Technical Field

The invention relates to an image processing technology, in particular to a method and a device for counting corn ear grains.

Background

The corn breeding is to perform hybridization test on a large amount of inbred lines and hybrid combinations by utilizing the heterozygosity of corn cross pollination and genetic basis, and finally select individual excellent hybrid seeds for production and use. Wherein the seed yield of the fruit cluster is an important index for judging the quality of the corn variety.

At present, the corn ear and seed counting work in China mainly adopts manual work. The method has huge workload and large error of reading data, and can not repeatedly read the read data, thereby causing problems and being difficult to verify the correctness of the data. Research in developed countries such as Europe and America uses a seed testing instrument. However, the instrument has the disadvantages of high price, large volume and weight, difficult flow and use and is not suitable for popularization and use in production practice.

Therefore, it is necessary to develop an automatic computer to improve the counting efficiency of corn ear and grain, improve the counting method and accelerate the variety examination and popularization.

Disclosure of Invention

The invention provides a method and a device for counting corn ear grains, which can realize the quick and simple counting of the corn ear grains,

the invention provides a corn ear seed counting method, which comprises the following steps:

a corn ear kernel counting method is characterized by comprising the following steps:

step S1, obtaining a corn ear and grain estimation model; wherein the corn ear kernel estimation model is used for expressing the functional relation between the number of the corn ear seeds on a single surface of the corn and the number of all the corn ear seeds;

s2, acquiring a single-side image of the corn, and performing image preprocessing on the single-side image to obtain a corresponding gray image;

step S3, carrying out multi-threshold segmentation on the gray level image obtained in the step S2 to obtain a corresponding binary image;

step S4, counting the connected areas with the same gray value, and acquiring the number of the connected areas corresponding to the gray value for representing the corn kernels;

and step S5, determining the number of all the ear seeds according to the number acquired in the step S4 and the corn kernel estimation model in the step S1.

Preferably, the step S1 specifically includes:

obtaining the following formula as a corn ear kernel estimation model:

y=ax+b；

wherein x represents the number of corn seeds in a single-sided image of the corn ear, a and b are training values, and y represents the number of all corn ear seeds.

Preferably, the step S2 includes:

step S21, performing median filtering processing on the single-sided image;

step S22, performing Wallis filtering processing on the image obtained in step S21;

step S23, performing histogram enhancement processing on the image obtained in step S22 to obtain a grayscale image corresponding to the single-sided image.

Preferably, the step S3 specifically includes:

and (4) carrying out image segmentation on the gray level image obtained in the step S2 by adopting a multi-threshold segmentation method combining RBGM and OTSU to further obtain a binary image of the corn kernels.

Preferably, the image segmentation is performed on the corn ear kernel gray level image by adopting a multi-threshold segmentation method combining RBGM and OTSU, so as to obtain a binary image of the corn kernels, and the method specifically comprises the following steps:

step S31, selecting initial iteration point t_iAnd iteration step η_mCarrying out iterative solution on the following multi-threshold segmentation iterative solution formula to obtain the optimal segmentation thresholdThe optimal segmentation thresholdTo make it possible toMaximum segmentation threshold:

$x_i^{\mathrm{mN}+j+1}=x_i^{\mathrm{mN}+j}-{\mathrm{\η}}_m\frac{\∂{\mathrm{\σ}}_B^2}{\∂t_i},i=\mathrm{1,2},...,N;$

wherein,

j＝0,1,...,N-1;m＝0,1,...

${\mathrm{\σ}}_B^2=\underset{k=0}{\overset{n-1}{\mathrm{\Σ}}}\underset{j=k+1}{\overset{n-1}{\mathrm{\Σ}}}{\mathrm{\ω}}_k{\mathrm{\ω}}_j{({\mathrm{\μ}}_k-{\mathrm{\μ}}_j)}^2;$

wherein ω is_k、μ_kRespectively representing the probability and the mean square error corresponding to the kth problem when the n problems are segmented;

step S32, using the optimal segmentation threshold obtained in step S31The grayscale image obtained in step S2 is subjected to multi-threshold segmentation.

The invention provides a corn ear and grain counting device, which comprises:

the modeling module is used for obtaining a corn ear and grain estimation model; wherein the corn ear kernel estimation model is used for expressing the functional relation between the number of the corn ear seeds on a single surface of the corn and the number of all the corn ear seeds;

the image preprocessing module is used for acquiring a single-side image of the corn and preprocessing the single-side image to obtain a corresponding gray image;

the image segmentation module is used for carrying out multi-threshold segmentation on the gray level image obtained by the image preprocessing module to obtain a corresponding binary image;

the counting module is used for counting the connected regions with the same gray value and acquiring the number of the connected regions corresponding to the gray value for representing the corn kernels;

and the calculating module is used for determining the number of all the ear seeds according to the number acquired in the counting module and the corn kernel estimation model acquired by the modeling module.

Preferably, the modeling module is specifically configured to obtain the following formula as the corn ear kernel estimation model:

y=ax+b；

wherein x represents the number of corn seeds in a single-sided image of the corn ear, a and b are training values, and y represents the number of all corn ear seeds.

Preferably, the image preprocessing module includes:

the first processing submodule is used for carrying out median filtering processing on the single-sided image;

the second processing submodule is used for performing Wallis filtering processing on the image obtained by the first processing submodule;

and the third processing submodule is used for performing histogram enhancement processing on the image obtained by the second processing submodule to obtain a gray level image corresponding to the single-side image.

Preferably, the image segmentation module is specifically configured to perform image segmentation on the grayscale image obtained by the image preprocessing module by using a multi-threshold segmentation method combining RBGM and OTSU, so as to obtain a binary image of the corn kernels.

Preferably, the image segmentation module is specifically configured to perform, when performing image segmentation on the grayscale image obtained by the image preprocessing module by using a multi-threshold segmentation method combining RBGM and OTSU to obtain a binary image of the corn kernel:

step S31, selecting initial iteration point t_iAnd iteration step η_mCarrying out iterative solution on the following multi-threshold segmentation iterative solution formula to obtain the optimal segmentation thresholdThe optimal segmentation thresholdTo make it possible toMaximum segmentation threshold:

$x_i^{\mathrm{mN}+j+1}=x_i^{\mathrm{mN}+j}-{\mathrm{\η}}_m\frac{\∂{\mathrm{\σ}}_B^2}{\∂t_i},i=\mathrm{1,2},...,N;$

wherein,

j＝0,1,...,N-1;m＝0,1,...

${\mathrm{\σ}}_B^2=\underset{k=0}{\overset{n-1}{\mathrm{\Σ}}}\underset{j=k+1}{\overset{n-1}{\mathrm{\Σ}}}{\mathrm{\ω}}_k{\mathrm{\ω}}_j{({\mathrm{\μ}}_k-{\mathrm{\μ}}_j)}^2;$

wherein ω is_k、μ_kRespectively representing the probability and the mean square error corresponding to the kth problem when the n problems are segmented;

step S32, using the optimal segmentation threshold obtained in step S31The grayscale image obtained in step S2 is subjected to multi-threshold segmentation.

By adopting the method for counting the corn ear grains, provided by the invention, the corn ear grains can be quickly, simply and conveniently counted.

Drawings

Fig. 1 is a schematic flow chart of a method for counting corn ear grains according to an embodiment of the present invention;

fig. 2 is a schematic flow chart of gray processing in the corn ear kernel counting method according to the embodiment of the present invention;

fig. 3 is a schematic flow chart of image segmentation in the corn ear kernel counting method according to the embodiment of the present invention;

FIG. 4a is a single-sided corn image after the ear of corn has been identified using the method for counting corn ears and kernels provided by embodiments of the present invention;

FIG. 4b is a partial single-sided corn image after the corn ear has been identified using the corn ear kernel counting method provided by embodiments of the present invention;

fig. 5 is a schematic structural view of a corn ear kernel counting device provided in the embodiment of the present invention.

Detailed Description

The following description of the embodiments of the present invention will be made with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.

The following describes in further detail embodiments of the present invention with reference to the accompanying drawings. The following examples are intended to illustrate the invention, but not to limit the scope thereof.

Example one

There is provided a corn ear kernel count, as shown in fig. 1, the method comprising:

step S1, obtaining a corn ear and grain estimation model; wherein the corn ear kernel estimation model is used for expressing the functional relation between the number of the corn ear seeds on a single surface of the corn and the number of all the corn ear seeds;

s2, acquiring a single-side image of the corn, and performing image preprocessing on the single-side image to obtain a corresponding gray image;

step S3, carrying out multi-threshold segmentation on the gray level image obtained in the step S2 to obtain a corresponding binary image;

step S4, counting the connected areas with the same gray value, and acquiring the number of the connected areas corresponding to the gray value for representing the corn kernels;

and step S5, determining the number of all the ear seeds according to the number acquired in the step S4 and the corn kernel estimation model in the step S1.

By adopting the method for counting the corn ear grains, provided by the invention, the corn ear grains can be quickly, simply and conveniently counted.

In the process of implementing the invention, the inventor of the application finds that: (1) the number of rows of corn ear seeds is generally fixed, and is usually 14 rows or 16 rows; (2) in the corn image sampling, an orthographic projection method is used, and corn kernels are fixed in the corn kernel region occupied by the visible part of the image, and on this basis, in a preferred embodiment of the present invention, the step S1 may specifically be: obtaining the following formula (1) as a corn ear kernel estimation model:

y=ax+b； （1）

wherein x represents the number of corn seeds in a single-sided image of the corn ear, a and b are training values, and y represents the number of all corn ear seeds.

The values a and b can be obtained through training, namely, manual counting is carried out on partial corn ears, and the values a and b are assigned. It should be noted that the training model established here can be used for a plurality of times in steps S2-S5, which facilitates the realization of fast counting of other corn ear grains.

Since step S4 is the result of processing the single-sided image of the corn, the obtained kernels are also single-sided, i.e. corresponding to x in the above-mentioned relational expression, and at this time, the total number of corn ear kernels can be obtained by substituting x into the above-mentioned relational expression.

Of course, in practical applications, other mathematical models may be used by those skilled in the art, and the specific model used does not affect the scope of the present invention.

Preferably, the method for performing image preprocessing on the single-sided image in step S2 may include, as shown in fig. 2:

step S21, performing median filtering processing on the single-sided image;

step S22, performing Wallis filtering processing on the image obtained in step S21;

step S23, performing histogram enhancement processing on the image obtained in step S22 to obtain a grayscale image corresponding to the single-sided image.

In practical applications, before step S21, a grayscale image f (x, y) of corn ear and kernel is also taken, where f (x, y) is a grayscale value corresponding to the (x, y) pixel point.

The formula used in the median filtering process in step S21 is the following formula (2):

$f^{\text{'}}(x_0,y_0)={\left[\underset{(x_i,y_i)\∈S}{\mathrm{Sort}}{\mathrm{\ω}}_{\mathrm{ij}}f(x_i,y_i)\right]}_{\frac{\left|S\right|+1}{2}}---\left(2\right)$

wherein S is a pixel (x)₀,y₀) Contains therein (x)₀,y₀) (x, y) represents an arbitrary element in S, f (x, y) represents a gray value of a point (x, y), | S | represents the number of elements in the set S, Sort represents the ranking, and ω represents a weight value added when calculating the intermediate pixel value.

The formula adopted in the Wallis filtering process in step S22 is the following formula (3):

$g(i,j)=[\mathrm{\α}{m}_d+(1-\mathrm{\α})\stackrel{\‾}{f}(i,j)]+[f(i,j)-\stackrel{\‾}{f}(i,j)]\frac{A{\mathrm{\σ}}_d}{\mathrm{A\σ}(i,j)+{\mathrm{\σ}}_d}---\left(3\right)$

wherein [ f (i, j)]_M×NRepresenting the original image, [ g (i, j)]_M×NThe enhanced image is represented by a representation of the image,denotes the local mean, m_dDenotes the mean value, σ_dIndicating the standard deviation, a the gain factor, α is a constant of proportionality that controls the composition of edges and background in the enhanced image.

The formula adopted for the histogram enhancement processing in step S23 is the following formula (4)

z＝G^-1[T(r)]（4）

r represents the gray level of the pixel before image enhancement, the transform function s-t (r) represents the gray level of the pixel after enhancement, and the inverse transform function z-G^-1(s) represents the desired new gray level.

Preferably, the step S3 specifically includes:

and (4) carrying out image segmentation on the gray level image obtained in the step S2 by adopting a multi-threshold segmentation method combining RBGM and OTSU to further obtain a binary image of the corn kernels.

Specifically, step S3 may include the steps shown in fig. 3:

step S31, selecting initial iteration point t_iAnd iteration step η_mCarrying out iterative solution on the following multi-threshold segmentation iterative solution formula (5) to obtain the optimal segmentation thresholdThe optimal segmentation thresholdTo make it possible toMaximum segmentation threshold:

$x_i^{\mathrm{mN}+j+1}=x_i^{\mathrm{mN}+j}-{\mathrm{\η}}_m\frac{\∂{\mathrm{\σ}}_B^2}{\∂t_i},i=\mathrm{1,2},...,N;---\left(5\right)$

wherein,

j＝0,1,...,N-1;m＝0,1,...

${\mathrm{\σ}}_B^2=\underset{k=0}{\overset{n-1}{\mathrm{\Σ}}}\underset{j=k+1}{\overset{n-1}{\mathrm{\Σ}}}{\mathrm{\ω}}_k{\mathrm{\ω}}_j{({\mathrm{\μ}}_k-{\mathrm{\μ}}_j)}^2;$

wherein ω is_k、μ_kThe probabilities and mean differences corresponding to the kth class of problems when the n classes of problems are divided are shown.

Step S32, using the optimal segmentation threshold obtained in step S31The grayscale image obtained in step S2 is subjected to multi-threshold segmentation.

The principle of the above step S31 is:

assume for the n-class segmentation problem that the total variance is labeled σ²The probability, mean and variance of each class are marked as omega_k、μ_k、(k-0, 1.., n-1), and a combination of the respective segmentation thresholds is denoted as t-t_k1, 2., n-1}, then:

wherein k is 0, 1.. times, n-1 (6)

t₀＝0,t_nL is the histogram border (7)

Wherein 1 < t_kL (k ═ 1, 2.., n-1) is each segmentation threshold, (8)

Then the multi-threshold intra-class variance is:

${\mathrm{\&sigma;}}_w^2=\underset{k=0}{\overset{n-1}{\mathrm{\&Sigma;}}}{\mathrm{\&omega;}}_k{\mathrm{\&sigma;}}_k^2---\left(9\right)$

similar to the single threshold problem, a constraint equation for the multi-threshold problem is established:

${\mathrm{\&sigma;}}^2=\underset{i=1}{\overset{L}{\mathrm{\&Sigma;}}}{(i-\mathrm{\&mu;})}^2{P}_i=\underset{i=1}{\overset{L}{\mathrm{\&Sigma;}}}{i}^2{P}_i-2\mathrm{\&mu;}\underset{i=1}{\overset{L}{\mathrm{\&Sigma;}}}i{P}_i+{\mathrm{\&mu;}}^2\underset{i=1}{\overset{L}{\mathrm{\&Sigma;}}}{P}_i=\underset{i=1}{\overset{L}{\mathrm{\&Sigma;}}}{i}^2{P}_i-{\mathrm{\&mu;}}^2;$

from the above formula can be derived ${\mathrm{\&sigma;}}^2-{\mathrm{\&sigma;}}_w^2=\underset{k=0}{\overset{n-1}{\mathrm{\&Sigma;}}}{\mathrm{\&omega;}}_k{\mathrm{\&mu;}}_k^2+\underset{k=0}{\overset{n-1}{\mathrm{\&Sigma;}}}{\mathrm{\&omega;}}_k{\mathrm{\&mu;}}_k\underset{j\&NotEqual;k}{\mathrm{\&Sigma;}}{\mathrm{\&omega;}}_j{\mathrm{\&mu;}}_j;$

Through sorting derivation, can obtain ${\mathrm{\&sigma;}}_B^2={\mathrm{\&sigma;}}^2-{\mathrm{\&sigma;}}_w^2=\underset{k=0}{\overset{n-1}{\mathrm{\&Sigma;}}}\underset{j=k+1}{\overset{n-1}{\mathrm{\&Sigma;}}}{\mathrm{\&omega;}}_k{\mathrm{\&omega;}}_j{({\mathrm{\&mu;}}_k-{\mathrm{\&mu;}}_j)}^2---\left(10\right);$ It is obvious thatRelative to an independent variable t_iAs a function of (c).

Then x in the formula (11) of gradient descent method (RBGM)_iIs replaced by t_iReplacing f (x) withAvailable drop solution equation (12):

wherein,

j＝0,1,...,N-1;m＝0,1,...。

$x_i^{\mathrm{mN}+j+1}=x_i^{\mathrm{mN}+j}-{\mathrm{\&eta;}}_m\frac{\&PartialD;{\mathrm{\&sigma;}}_B^2}{\&PartialD;t_i},i=\mathrm{1,2},...,N;---\left(12\right).$

the multi-threshold segmentation method combining RBGM and OTSU is adopted to solve the problem that the calculation amount for solving the optimal threshold increases geometrically as the number of segmentation thresholds increases from a single threshold to multiple thresholds in the traditional OSTU multi-threshold segmentation method, and by adopting the method, the segmentation calculation amount can be greatly reduced under the same precision, so that the segmentation calculation time is shortened.

And S3, obtaining a binary image of the corn ear grains, wherein each corn grain is divided into a connected region, counting the connected regions with the same gray value to obtain the number of the connected regions corresponding to the gray value for representing the corn grains, and then calculating the total number of the corn ear grains according to a pre-established estimation model of the corn ear grains. As shown in fig. 4 (a) (b), each corn kernel in the binary image of corn ear kernels is divided into a connected region, i.e. adjacent pixels in the region have the same gray value of 0 or 1. Before the identification, a counter C is initialized, the initial value is 1, the counter C is used for counting the number of the single-sided corn kernels, and p (x, y) represents the pixel value of the pixel at the (x, y) point. From any initial point (x) of the image₀,y₀) Starting to scan each pixel line by line, and when p (x, y) is not 1, indicating that the pixel is an image background or an identified pixel skips the pixel; when a pixel with p (x, y) being 1 is encountered, the pixel is identified as a counter C value, all pixels with the pixel adjacent to the pixel and the pixel with the pixel value being 1 are scanned, the value of the pixel is identified as C, after the identification is completed, the C value is added with 1, and the identification of the next pixel is continued. And marking the pixels one by one until all the pixels are scanned, and finally obtaining the single-side corn seed number.

The beneficial effects of the invention include: (1) the computer graphic processing technology is used for replacing a manual corn ear seed counting method, the total number of the corn ear seeds can be quickly, simply and accurately obtained, and the defects of long counting time, strong subjectivity, large data reading error and the like of the corn ear seeds in the prior art are overcome; (2) the method comprises the steps of taking a part of collected single-sided corn ear seed images as samples, performing regression analysis according to the number of the single-sided corn ear seeds in the images and the actual total number of the corn ear seeds to obtain a corn ear seed estimation model y = ax + b, estimating the total number of the corn ear seeds according to the corn seed estimation model and the number of the single-sided corn seeds in the images, wherein the accuracy rate reaches 93%; (3) the invention adopts a multi-threshold segmentation method combining RBGM and OTSU, aiming at solving the problem that the calculation amount for solving the optimal threshold increases geometrically as the number of segmentation thresholds increases from a single threshold to a multi-threshold in the traditional OSTU multi-threshold segmentation method, the method can greatly reduce the segmentation calculation amount under the same precision, thereby shortening the segmentation calculation time.

Example two

Based on the same concept, the invention also provides a corn ear kernel counting device, as shown in fig. 5, the device comprises:

the modeling module 501 is used for obtaining a corn ear and grain estimation model; wherein the corn ear kernel estimation model is used for expressing the functional relation between the number of the corn ear seeds on a single surface of the corn and the number of all the corn ear seeds;

the image preprocessing module 502 is used for acquiring a single-side image of the corn and preprocessing the single-side image to obtain a corresponding gray image;

the image segmentation module 503 is configured to perform multi-threshold segmentation on the grayscale image obtained by the image preprocessing module 502 to obtain a corresponding binary image;

the counting module 504 is configured to count the connected regions with the same gray value, and obtain the number of the connected regions corresponding to the gray value used for representing the corn kernels;

and a calculating module 505 for determining the total number of the ear seeds according to the number obtained in the counting module 504 and the corn kernel estimation model obtained in the modeling module 501.

Preferably, the modeling module 501 is specifically configured to obtain the following formula as the corn ear kernel estimation model:

y=ax+b；

wherein x represents the number of corn seeds in a single-sided image of the corn ear, a and b are training values, and y represents the number of all corn ear seeds.

Preferably, the image preprocessing module 502 specifically includes:

the first processing submodule is used for carrying out median filtering processing on the single-sided image;

the second processing submodule is used for performing Wallis filtering processing on the image obtained by the first processing submodule;

and the third processing submodule is used for performing histogram enhancement processing on the image obtained by the second processing submodule to obtain a gray level image corresponding to the single-side image.

Preferably, the image segmentation module 503 is specifically configured to perform image segmentation on the grayscale image obtained by the image preprocessing module 502 by using a multi-threshold segmentation method combining RBGM and OTSU, so as to obtain a binary image of the corn kernels.

Preferably, the segmentation module is configured to, when performing image segmentation on the grayscale image obtained by the image preprocessing module 502 by using a multi-threshold segmentation method combining RBGM and OTSU to obtain a binary image of the corn kernel, specifically perform:

step S31, selecting initial iteration point t_iAnd iteration step η_mCarrying out iterative solution on the following multi-threshold segmentation iterative solution formula to obtain the optimal segmentation thresholdThe optimal segmentation thresholdTo make it possible toMaximum segmentation threshold:

$x_i^{\mathrm{mN}+j+1}=x_i^{\mathrm{mN}+j}-{\mathrm{\&eta;}}_m\frac{\&PartialD;{\mathrm{\&sigma;}}_B^2}{\&PartialD;t_i},i=\mathrm{1,2},...,N;$

wherein,

j＝0,1,...,N-1;m＝0,1,...

${\mathrm{\&sigma;}}_B^2=\underset{k=0}{\overset{n-1}{\mathrm{\&Sigma;}}}\underset{j=k+1}{\overset{n-1}{\mathrm{\&Sigma;}}}{\mathrm{\&omega;}}_k{\mathrm{\&omega;}}_j{({\mathrm{\&mu;}}_k-{\mathrm{\&mu;}}_j)}^2;$

wherein ω is_k、μ_kRespectively representing the probability and the mean square error corresponding to the kth problem when the n problems are segmented;

step S32, using the optimal segmentation threshold obtained in step S31The grayscale image obtained in step S2 is subjected to multi-threshold segmentation.

The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and decorations can be made without departing from the technical principle of the present invention, and these modifications and decorations should also be regarded as the protection scope of the present invention.

Claims (8)

1. A corn ear kernel counting method is characterized by comprising the following steps:

step S1, obtaining a corn ear and grain estimation model; wherein the corn ear kernel estimation model is used for expressing the functional relation between the number of the corn ear seeds on a single surface of the corn and the number of all the corn ear seeds;

s2, acquiring a single-side image of the corn, and performing image preprocessing on the single-side image to obtain a corresponding gray image;

step S3, carrying out multi-threshold segmentation on the gray level image obtained in the step S2 to obtain a corresponding binary image;

step S4, counting the connected areas with the same gray value, and acquiring the number of the connected areas with the gray value for representing the corn kernels;

step S5, determining the number of all the ears and seeds according to the number acquired in the step S4 and the corn kernel estimation model in the step S1;

wherein, the step S1 specifically includes:

obtaining the following formula as a corn ear kernel estimation model:

y＝ax+b；

wherein x represents the number of corn seeds in a single-sided image of the corn ear, a and b are training values, and y represents the number of all corn ear seeds.

2. The method of claim 1, wherein the step S2 includes:

step S21, performing median filtering processing on the single-sided image;

step S22, performing Wallis filtering processing on the image obtained in step S21;

step S23, performing histogram enhancement processing on the image obtained in step S22 to obtain a grayscale image corresponding to the single-sided image.

3. The method according to claim 1, wherein the step S3 specifically includes:

and (4) performing image segmentation on the gray level image obtained in the step S2 by adopting a multi-threshold segmentation method combining a gradient descent method RBGM and a maximum inter-class variance method OTSU to obtain a binary image of the corn kernels.

4. The method of claim 3, wherein the image segmentation is performed on the corn ear kernel gray scale image by a multi-threshold segmentation method combining RBGM and OTSU to obtain a binary image of the corn kernels, and specifically comprises:

step S31, selecting initial iteration point t_iAnd iteration step η_mCarrying out iterative solution on the following multi-threshold segmentation iterative solution formula to obtain the optimal segmentation thresholdThe optimal segmentation thresholdTo make it possible toMaximum segmentation threshold:

${x_i}^{mN+j+1}={x_i}^{mN+j}-{\mathrm{\&eta;}}_m\frac{\&part;{\mathrm{\&sigma;}}_B^2}{\&part;t_i},i=1,2,...,N;$

wherein x is_i ^mN+j+1＝x_i ^mN+j-η_mΔ_i(x_i ^mN+j),i＝1,2,...,N

${\mathrm{\&Delta;}}_i\left(x^{mn+j}\right)=\left\{\begin{array}{cc}\frac{\&part;f\left(x^{mn+j}\right)}{\&part;x^i},& j=i-1\\ 0,& j\&NotEqual;i-1\end{array}\right.,j=0,1,...,N-1;m=0,1,...$

${\mathrm{\&sigma;}}_B^2=\underset{k=0}{\overset{n-1}{\&Sigma;}}\underset{j=k+1}{\overset{n-1}{\&Sigma;}}{\mathrm{\&omega;}}_k{\mathrm{\&omega;}}_i{({\mathrm{\&mu;}}_k-{\mathrm{\&mu;}}_j)}^2;$

Wherein ω is_k、μ_kRespectively representing the probability and the mean square error corresponding to the kth problem when the n problems are segmented;

step S32, using the optimal segmentation threshold obtained in step S31The grayscale image obtained in step S2 is subjected to multi-threshold segmentation.

5. The utility model provides a maize ear seed grain counting assembly which characterized in that includes:

the modeling module is used for obtaining a corn ear and grain estimation model; wherein the corn ear kernel estimation model is used for expressing the functional relation between the number of the corn ear seeds on a single surface of the corn and the number of all the corn ear seeds;

the image preprocessing module is used for acquiring a single-side image of the corn and preprocessing the single-side image to obtain a corresponding gray image;

the image segmentation module is used for carrying out multi-threshold segmentation on the gray level image obtained by the image preprocessing module to obtain a corresponding binary image;

the counting module is used for counting the connected regions with the same gray value to obtain the number of the connected regions with the gray value for representing the corn kernels;

the calculation module is used for determining the number of all the ear seeds according to the number acquired in the counting module and the corn kernel estimation model acquired by the modeling module;

the modeling module is specifically used for obtaining the following formula as a corn ear kernel estimation model:

y＝ax+b；

wherein x represents the number of corn seeds in a single-sided image of the corn ear, a and b are training values, and y represents the number of all corn ear seeds.

6. The apparatus of claim 5, wherein the image pre-processing module comprises:

the first processing submodule is used for carrying out median filtering processing on the single-sided image;

the second processing submodule is used for performing Wallis filtering processing on the image obtained by the first processing submodule;

and the third processing submodule is used for performing histogram enhancement processing on the image obtained by the second processing submodule to obtain a gray level image corresponding to the single-side image.

7. The apparatus according to claim 5, wherein the image segmentation module is specifically configured to perform image segmentation on the grayscale image obtained by the image preprocessing module by using a multi-threshold segmentation method combining a gradient descent method (RBGM) and a maximum inter-class variance method (OTSU) to obtain a binary image of the corn kernel.

8. The apparatus of claim 7, wherein the image segmentation module is configured to, when performing image segmentation on the grayscale image obtained by the image preprocessing module by using a multi-threshold segmentation method combining RBGM and OTSU to obtain a binary image of the corn kernel, specifically configured to perform:

step S31, selecting initial iteration point t_iAnd iteration step η_mCarrying out iterative solution on the following multi-threshold segmentation iterative solution formula to obtain the optimal segmentation thresholdThe optimal segmentation thresholdTo make it possible toMaximum segmentation threshold:

${x_i}^{mN+j+1}={x_i}^{mN+j}-{\mathrm{\&eta;}}_m\frac{\&part;{\mathrm{\&sigma;}}_B^2}{\&part;t_i},i=1,2,...,N;$

wherein x is_i ^mN+j+1＝x_i ^mN+j-η_mΔ_i(x_i ^mN+j),i＝1,2,...,N

${\mathrm{\&Delta;}}_i\left(x^{mn+j}\right)=\left\{\begin{array}{cc}\frac{\&part;f\left(x^{mn+j}\right)}{\&part;x^i},& j=i-1\\ 0,& j\&NotEqual;i-1\end{array}\right.,j=0,1,...,N-1;m=0,1,...$

${\mathrm{\&sigma;}}_B^2=\underset{k=0}{\overset{n-1}{\&Sigma;}}\underset{j=k+1}{\overset{n-1}{\&Sigma;}}{\mathrm{\&omega;}}_k{\mathrm{\&omega;}}_i{({\mathrm{\&mu;}}_k-{\mathrm{\&mu;}}_j)}^2;$

Wherein ω is_k、μ_kRespectively representing the probability and the mean square error corresponding to the kth problem when the n problems are segmented;

step S32, using the optimal segmentation threshold obtained in step S31The grayscale image obtained in step S2 is subjected to multi-threshold segmentation.