CN112330635A - A method for detecting internal cracks in corn seeds based on correlation filtering - Google Patents

A method for detecting internal cracks in corn seeds based on correlation filtering Download PDF

Info

Publication number
CN112330635A
CN112330635A CN202011227942.5A CN202011227942A CN112330635A CN 112330635 A CN112330635 A CN 112330635A CN 202011227942 A CN202011227942 A CN 202011227942A CN 112330635 A CN112330635 A CN 112330635A
Authority
CN
China
Prior art keywords
seed
crack
correlation
correlation filtering
template
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202011227942.5A
Other languages
Chinese (zh)
Other versions
CN112330635B (en
Inventor
李佳
吕程序
姚丙南
毛文华
汪凤珠
李亚硕
李阳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sinotest Equipment Co ltd
Chinese Academy of Agricultural Mechanization Sciences
Original Assignee
Sinotest Equipment Co ltd
Chinese Academy of Agricultural Mechanization Sciences
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Sinotest Equipment Co ltd, Chinese Academy of Agricultural Mechanization Sciences filed Critical Sinotest Equipment Co ltd
Priority to CN202011227942.5A priority Critical patent/CN112330635B/en
Publication of CN112330635A publication Critical patent/CN112330635A/en
Application granted granted Critical
Publication of CN112330635B publication Critical patent/CN112330635B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • A—HUMAN NECESSITIES
    • A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
    • A01C—PLANTING; SOWING; FERTILISING
    • A01C1/00—Apparatus, or methods of use thereof, for testing or treating seed, roots, or the like, prior to sowing or planting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Soil Sciences (AREA)
  • Environmental Sciences (AREA)
  • Quality & Reliability (AREA)
  • Image Analysis (AREA)
  • Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)

Abstract

一种基于相关滤波的玉米种子内部裂纹检测方法，包括如下步骤：利用彩色相机和高亮度光源获取种子透射图像，并根据所述种子透射图像构建种子裂纹数据集；利用所述种子裂纹数据集构建种子裂纹相关滤波模板和相关阈值；利用所述相关滤波模板对待检测种子的非种子区域进行边缘检测，将与种子裂纹具有高度相关性的候选区域提取出来；裂纹筛选，比较所述候选区域的相关滤波响应峰值与相关阈值，并根据比较结果判断所述候选区域是否为种子裂纹区域；以及将筛选出的种子裂纹区域的像素点连线，作为种子裂纹检测结果。本发明可快速高效进行玉米内部裂纹无损检测，满足种子精细化、智能化及标准化需求，为用种安全、机械化高效播种及作物增产提供保障。

Figure 202011227942

A method for detecting internal cracks in corn seeds based on correlation filtering, comprising the steps of: acquiring a seed transmission image by using a color camera and a high-brightness light source, and constructing a seed crack data set according to the seed transmission image; constructing a seed crack data set using the seed crack data set Seed crack correlation filtering template and correlation threshold; use the correlation filtering template to perform edge detection on the non-seed area of the seed to be detected, and extract the candidate area with high correlation with the seed crack; Crack screening, compare the correlation of the candidate area Filter the response peak value and the relevant threshold, and judge whether the candidate area is a seed crack area according to the comparison result; and connect the pixel points of the screened seed crack area as the seed crack detection result. The invention can quickly and efficiently perform nondestructive detection of internal cracks in corn, meet the needs of seed refinement, intelligence and standardization, and provide guarantee for safe seed use, mechanized and efficient sowing and crop yield increase.

Figure 202011227942

Description

Corn seed internal crack detection method based on related filtering
Technical Field
The invention relates to a nondestructive testing technology of agricultural products, in particular to a corn seed internal crack detection method based on related filtering.
Background
The problems of high damage, nonstandard and the like exist in the processing process of crop seeds, wherein the internal cracks are important factors influencing the quality of corns. Internal cracks are often difficult to detect and are more potentially harmful because the seed coat is intact and unbroken in appearance. The field needs a quick and efficient method for detecting the internal cracks of the corns urgently, the requirements of seed refinement, intellectualization and standardization are met, support is provided for guaranteeing safe and mechanical efficient seeding of seeds and crop yield increase, and technical guarantee is provided for nondestructive detection of agricultural seeds.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a corn seed internal crack detection method based on correlation filtering aiming at the problems in the prior art.
In order to achieve the purpose, the invention provides a corn seed internal crack detection method based on correlation filtering, which comprises the following steps:
s100, acquiring a seed transmission image by using a color camera and a high-brightness light source, and constructing a seed crack data set according to the seed transmission image;
s200, constructing a seed crack related filtering template and a related threshold value by using the seed crack data set;
s300, performing edge detection on a non-seed region of the seed to be detected by using the related filtering template, and extracting a candidate region with high correlation with the seed crack;
s400, crack screening, namely comparing a correlation filtering response peak value and a correlation threshold value of the candidate region, and judging whether the candidate region is a seed crack region according to a comparison result; and
s500, connecting the pixel points of the screened seed crack area to obtain a seed crack detection result.
In the method for detecting the internal crack of the corn seed based on the correlation filtering, in step S200, the seed crack data set is operated through fourier transform and gaussian transform to obtain the seed crack correlation filtering template.
The maize seed internal crack detection method based on the correlation filtering is characterized in that the construction of the seed crack correlation filtering template further comprises the following steps:
s201, randomly extracting seed cracks in the seed crack data set, and forming a set S ═ f by taking k × k as an arbitrary region of a template size and taking the seed crack pixel points as centers1,f2,...,fNN is a seed crack sample;
s202, according to a related filtering principle, inputting the seed transmission image and the related filtering template to calculate to obtain a following related graph g:
Figure BDA0002764201250000021
wherein f is the seed transmission image, h is the correlation filtering template,
Figure BDA0002764201250000022
performing convolution operation;
s203, Fourier transform is carried out on the above expression, and the Fourier transform of the function cross correlation is equal to the product of the function Fourier transform:
Figure BDA0002764201250000023
wherein, denotes the conjugate calculation, F (-) is the fourier transform;
s204, simplifying the formula as follows: g ═ F · H*Obtaining:
Figure BDA0002764201250000024
wherein H*Is the seed crack correlation filter template, G is the expected response.
The maize seed internal crack detection method based on the correlation filtering is characterized in that the construction of the seed crack correlation filtering template further comprises the following steps:
s205, solving by adopting a minimized sum of squares error for N seed crack samples in the set S:
Figure BDA0002764201250000025
where G is a standard Gaussian function with a peak value of 1.
The maize seed internal crack detection method based on the correlation filtering is characterized in that the edge detection in step S300 further includes:
adopting scharr operator to carry out edge calculation to obtain an edge image ci：
Figure BDA0002764201250000026
Where x is the seed transmission image sample with internal cracks, xiRepresenting the ith sample data (i ═ 1, 2.., N), where N is the number of crack samples; kxIs the filter coefficient in the x direction, KyIs the filter coefficient in the y direction.
The corn seed internal crack detection method based on the correlation filtering is characterized in that Kx，KyRespectively as follows:
Figure BDA0002764201250000031
Figure BDA0002764201250000032
in the corn seed internal crack detection method based on correlation filtering, only the edge image of the seed in the circular light-transmitting area is calculated, and the radius of the circular light-transmitting area is R.
The maize seed internal crack detection method based on correlation filtering is characterized in that the crack screening in the step S400 further comprises:
with ciTaking an edge pixel point as a center, acquiring a rectangular image area P with the size of k multiplied by k, and performing Fourier transform on the rectangular image area P to obtain P, wherein the relevant filtering response is as follows:
Y＝H*·P
wherein, ciFor the output image after edge detection, H*P is a candidate region for the filter template.
In the method for detecting internal cracks of corn seeds based on correlation filtering, a threshold T is set, and if the peak value of the correlation filtering response is greater than or equal to the threshold T, the candidate region is a crack region; and if the peak value of the correlation filtering response is smaller than a threshold value T, the candidate region is a non-crack region.
In order to better achieve the above object, the present invention further provides a corn seed internal crack detection method based on correlation filtering, wherein the method comprises the following steps:
constructing a seed crack data set: acquiring a crack region image with the size of k multiplied by k by taking a crack pixel point as a center according to the seed transmission image to form a seed crack data set;
and (3) calculating a filtering template: setting the target response as a standard Gaussian function with the peak value of 1, carrying out Fourier transform on the crack image, obtaining a corresponding related filtering template according to a filtering template calculation formula, and carrying out iterative solution through a least square sum error to obtain a final filtering template H*；
Carrying out edge detection on the seed transmission image to obtain a candidate region of the seed crack;
taking the edge detection result as a mask, and carrying out pixel correspondence with the original seed transmission image;
taking candidate edge pixel points in the edge detection result as a center, intercepting a k multiplied by k original seed transmission image as candidate region input, and performing Fourier transform on the candidate region to obtain a relevant filtering response of the candidate region;
judging the peak value of the relevant filtering response, and if the peak value is greater than or equal to a threshold value T, determining that the candidate region is a crack region; if the peak value is smaller than the threshold value T, the candidate area is a non-crack area;
circularly traversing all candidate regions of edge detection, and returning to obtain relevant filtering response of the candidate regions again if the traversal is not completed; and if traversing is finished, connecting the crack areas of the seeds in sequence to finish the detection of the cracks in the seeds.
The invention has the technical effects that:
the method can quickly and efficiently detect the internal cracks of the corn, meets the requirements of refinement, intellectualization and standardization of seeds, provides support for guaranteeing the safety of seed utilization, mechanized efficient sowing and crop yield increase, and provides technical support for nondestructive detection of agricultural seed utilization.
The invention is described in detail below with reference to the drawings and specific examples, but the invention is not limited thereto.
Drawings
FIG. 1 is a schematic view of crack seed transmission according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of crack seed edge detection according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of crack screening according to an embodiment of the present invention.
Detailed Description
The invention will be described in detail with reference to the following drawings, which are provided for illustration purposes and the like:
the invention provides a corn seed internal crack detection method based on relevant filtering based on an image processing technology, a machine learning technology and a signal filtering technology, and a seed transmission image is obtained by utilizing a color camera and a high-brightness light source. And according to the actual situation, removing the non-seed area by using the template. The method comprises the steps of converting time domain calculation of an image into frequency domain calculation by utilizing a correlation filtering principle, extracting a region with high correlation with a seed crack part by constructing a correlation filtering template and a correlation threshold value, and finally connecting screened pixel points to realize detection and identification of the crack part. The method specifically comprises the following steps:
s100, acquiring a seed transmission image (see figure 1) by using a color camera and a high-brightness light source, and constructing a seed crack data set according to the seed transmission image;
s200, constructing a seed crack related filtering template and a related threshold value by using the seed crack data set;
s300, performing edge detection on a non-seed region of the seed to be detected by using the related filtering template, and extracting a candidate region with high correlation with the crack of the seed;
s400, crack screening, namely comparing a correlation filtering response peak value and a correlation threshold value of the candidate region, and judging whether the candidate region is a seed crack region or not according to a comparison result; and
and S500, connecting the pixel points of the screened seed crack area as a seed crack detection result.
In this embodiment, in step S200, the seed crack data set is operated through fourier transform and gaussian transform to obtain the seed crack related filtering template. Further comprising:
step S201, in the seed crack data set, randomly extracting seed cracks and calculating the seed cracksThe seed crack pixel point is taken as the center, and an arbitrary region with the size of k multiplied by k as a template forms a set S ═ f1,f2,...,fNN is a seed crack sample;
step S202, according to the relevant filtering principle, inputting the seed transmission image f and the relevant filtering template h to calculate to obtain the following relevant graph g:
Figure BDA0002764201250000051
wherein f is the seed transmission image, h is the correlation filtering template,
Figure BDA0002764201250000052
performing convolution operation;
step S203, in order to increase the calculation speed, fourier transform is performed on the above equation, and it can be known from the convolution theorem that the fourier transform of the functional cross-correlation is equal to the product of the functional fourier transform:
Figure BDA0002764201250000053
wherein, denotes the conjugate calculation, F (-) is the fourier transform;
step S204, the above formula is simplified as follows:
G＝F·H*(3) then, we can get:
Figure BDA0002764201250000054
wherein H*Is a seed crack correlation filter template, G is an expected response, where the peak is a standard Gaussian function of 1; and
step S205, solving by using a minimum sum of squares error for N seed crack samples in the set S:
Figure BDA0002764201250000061
where G is a standard Gaussian function with a peak value of 1.
Wherein, the edge detection in step S300 further includes:
according to the practical situation, the seed crack is a gap with edge characteristics, so that the suspected region of the seed crack is firstly deleted by using an edge detection method to obtain an edge image ciFor further investigation. Since the seed crack region is often weak, the edge calculation is performed using the scharr operator, and the filter coefficient is shown as KxIs the filter coefficient in the x direction, KyFor the filter coefficients in the y-direction, an edge image c is obtainedi：
Figure BDA0002764201250000062
Referring to fig. 2, fig. 2 is a schematic diagram of crack seed edge detection according to an embodiment of the invention. Where x is the seed transmission image sample with internal cracks, let x beiRepresenting the ith sample data (i ═ 1, 2.., N), where N is the number of crack samples; kxIs the filter coefficient in the x direction, KyThe filter coefficient in the y direction is, according to the actual situation, the seed is in the circular light-transmitting area, so only the edge image of the seed in the circular light-transmitting area is calculated, and the radius of the circular light-transmitting area is R.
Wherein Kx，KyRespectively as follows:
Figure BDA0002764201250000063
Figure BDA0002764201250000064
referring to fig. 3, fig. 3 is a schematic diagram of crack screening according to an embodiment of the invention. Wherein the crack screening in step S400 further comprises:
from the above formula (6) to obtain ciFor the output image after edge detection, H*Is a filtering template. With ciAs a mask, taking an edge pixel point as a center, obtaining a rectangular image region P with a size of k × k and performing fourier transform to obtain P, as shown in fig. 3, if an objective function is a gaussian function according to formula (4), a relevant filter response is as follows:
Y＝H*·P (9)；
wherein, ciFor the output image after edge detection, H*P is a candidate region for the filter template.
From the above equation, if the input p is a crack region, the output filter response should be a gaussian-like function; if the input is not a seed crack region, but is other regions with edge features, the output response is not a gaussian function. Setting a threshold T for accelerating judgment, and if the peak value of the relevant filtering response is greater than or equal to the threshold T, determining that the candidate region is a crack region; and if the peak value of the correlation filtering response is smaller than a threshold value T, the candidate region is a non-crack region. And finally, connecting the crack pixel points to obtain a seed crack detection result.
The corn seed internal crack detection method based on the correlation filtering comprises the following steps of:
step 1, constructing a crack part data set: acquiring a crack region image with the size of k multiplied by k (k is 7) by taking a crack pixel point as a center according to the seed transmission image, and constructing a seed crack data set;
step 2, calculating a filtering template: setting the target response as a standard Gaussian function with the peak value of 1, carrying out Fourier transform on the crack image, obtaining a corresponding related filtering template according to a filtering template calculation formula, and carrying out iterative solution through a least square sum error to obtain a final filtering template H*；
And 3, according to the actual situation, the seed is in the circular light-transmitting area, so that only the image in the circular area participates in calculation, the image outside the circular area does not participate in calculation, and the radius of the circle in the image is R (R is 112). Carrying out edge detection on the seed transmission image to obtain a candidate region of the seed crack;
step 4, taking the edge detection result as a mask, and carrying out pixel correspondence with the original seed transmission image;
step 5, taking the candidate edge pixel point in the edge detection result as the center, intercepting the k multiplied by k original seed transmission image as the input of the candidate area, carrying out Fourier transform on the candidate area, and then solving by using a formula (9) to obtain the relevant filtering response of the candidate area;
step 6, judging a peak value of the correlation filtering response, and if the peak value is greater than or equal to a threshold value T (T is 0.7), determining that the candidate region is a crack region; if the peak value is smaller than the threshold value T, the candidate area is a non-crack area;
7, circularly traversing all candidate regions of edge detection, and returning to the step 5 to obtain relevant filtering responses of the candidate regions again if traversing is not finished; if the traversal is finished, entering the step 8;
and 8, connecting the crack regions of the seeds in sequence to complete the detection of the internal cracks of the seeds.
The invention utilizes a color camera and a high-brightness point light source to obtain the transmission image of the corn seed. And (4) utilizing correlation filtering analysis to count crack parts in the crack seeds, and obtaining a crack correlation filtering template through operations such as Fourier transform, Gaussian transform and the like. And (3) converting time domain calculation into frequency domain calculation, and quickly positioning the crack part of the seed, thereby completing the internal crack detection of the corn seed and realizing the positioning and identification of the internal crack of the corn seed.
The present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof, and it should be understood that various changes and modifications can be effected therein by one skilled in the art without departing from the spirit and scope of the invention as defined in the appended claims.

Claims (10)

1.一种基于相关滤波的玉米种子内部裂纹检测方法，其特征在于，包括如下步骤：1. a method for detecting internal cracks in corn seeds based on correlation filtering, is characterized in that, comprises the steps: S100、利用彩色相机和高亮度光源获取种子透射图像，并根据所述种子透射图像构建种子裂纹数据集；S100, using a color camera and a high-brightness light source to obtain a seed transmission image, and constructing a seed crack data set according to the seed transmission image; S200、利用所述种子裂纹数据集构建种子裂纹相关滤波模板和相关阈值；S200, using the seed crack data set to construct a seed crack correlation filter template and a related threshold; S300、利用所述相关滤波模板对待检测种子的非种子区域进行边缘检测，将与种子裂纹具有高度相关性的候选区域提取出来；S300, using the correlation filtering template to perform edge detection on the non-seed region of the seed to be detected, and extracting the candidate region that is highly correlated with the seed crack; S400、裂纹筛选，比较所述候选区域的相关滤波响应峰值与相关阈值，并根据比较结果判断所述候选区域是否为种子裂纹区域；以及S400. Crack screening, comparing the correlation filter response peak value of the candidate region with the correlation threshold, and determining whether the candidate region is a seed crack region according to the comparison result; and S500、将筛选出的种子裂纹区域的像素点连线，作为种子裂纹检测结果。S500 , connecting the screened pixel points of the seed crack region as a seed crack detection result. 2.如权利要求1所述的基于相关滤波的玉米种子内部裂纹检测方法，其特征在于，步骤S200中通过傅里叶变换和高斯变换对所述种子裂纹数据集进行运算获得所述种子裂纹相关滤波模板。2. The method for detecting internal cracks in corn seeds based on correlation filtering according to claim 1, wherein in step S200, the seed crack correlation is obtained by performing operations on the seed crack dataset by Fourier transform and Gaussian transform. Filter template. 3.如权利要求2所述的基于相关滤波的玉米种子内部裂纹检测方法，其特征在于，构建所述种子裂纹相关滤波模板进一步包括：3. the corn seed internal crack detection method based on correlation filtering as claimed in claim 2, is characterized in that, constructing described seed crack correlation filtering template further comprises: S201、在所述种子裂纹数据集中，随机提取种子裂纹并以所述种子裂纹像素点为中心，以k×k为模板大小的任意区域构成集合S＝{f1,f2,...,fN}，N为种子裂纹样本；S201. From the seed crack data set, randomly extract seed cracks and form a set S={f 1 , f 2 , . . . f N }, N is the seed crack sample; S202、根据相关滤波原理，输入所述种子透射图像与所述相关滤波模板进行计算可得如下相关图g：S202, according to the correlation filtering principle, input the seed transmission image and the correlation filtering template for calculation to obtain the following correlation diagram g:
Figure FDA0002764201240000011
Figure FDA0002764201240000011
 其中，f为种子透射图像，h为相关滤波模板，
Figure FDA0002764201240000012
为卷积运算；where f is the seed transmission image, h is the correlation filter template,
Figure FDA0002764201240000012
is the convolution operation; S203、对上式进行傅里叶变换，函数互相关的傅里叶变换等于函数傅里叶变换的乘积：S203. Perform Fourier transform on the above formula, and the Fourier transform of the function cross-correlation is equal to the product of the function Fourier transform:
Figure FDA0002764201240000013
Figure FDA0002764201240000013
 其中，*表示共轭计算，F(·)为傅里叶变换；Among them, * represents the conjugate calculation, F( ) is the Fourier transform; S204、将上式化简为：G＝F·H*，得到：
Figure FDA0002764201240000014
S204. Simplify the above formula into: G=F·H * , and obtain:
Figure FDA0002764201240000014
 其中，H*为种子裂纹相关滤波模板，G为期望响应。where H * is the seed crack correlation filter template and G is the expected response. 4.如权利要求3所述的基于相关滤波的玉米种子内部裂纹检测方法，其特征在于，构建所述种子裂纹相关滤波模板还包括：4. the corn seed internal crack detection method based on correlation filtering as claimed in claim 3, is characterized in that, constructing described seed crack correlation filtering template also comprises: S205、针对集合S中的N个种子裂纹样本，采用最小化平方和误差进行求解：S205. For the N seed crack samples in the set S, use the minimized sum of squares error to solve:
Figure FDA0002764201240000021
Figure FDA0002764201240000021
 其中G为峰值为1的标准高斯函数。where G is a standard Gaussian function with a peak value of 1. 5.如权利要求1-4中任意一项所述的基于相关滤波的玉米种子内部裂纹检测方法，其特征在于，步骤S300中边缘检测进一步包括：5. The method for detecting internal cracks in corn seeds based on correlation filtering according to any one of claims 1-4, wherein in step S300, edge detection further comprises: 采用scharr算子进行边缘计算，获得边缘图像ci：Use scharr operator to perform edge calculation to obtain edge image c i :
Figure FDA0002764201240000022
Figure FDA0002764201240000022
 其中，x为具有内部裂纹的种子透射图像样本，xi表示第i个样本数据(i＝1,2,...,N),N为裂纹样本数量；Kx为x方向的滤波系数，Ky为y方向的滤波系数。Among them, x is the seed transmission image sample with internal cracks, x i represents the ith sample data (i=1,2,...,N), N is the number of crack samples; Kx is the filter coefficient in the x direction, Ky is the filter coefficient in the y direction. 6.如权利要求5所述的基于相关滤波的玉米种子内部裂纹检测方法，其特征在于，Kx，Ky分别为：6. the corn seed internal crack detection method based on correlation filtering as claimed in claim 5 is characterized in that, K x , K y are respectively:
Figure FDA0002764201240000023
Figure FDA0002764201240000023
Figure FDA0002764201240000024
Figure FDA0002764201240000024
 7.如权利要求6所述的基于相关滤波的玉米种子内部裂纹检测方法，其特征在于，仅计算种子在圆形透光区域内的边缘图像，所述圆形透光区域的半径为R。7 . The method for detecting internal cracks in corn seeds based on correlation filtering according to claim 6 , wherein only the edge image of the seeds in a circular light-transmitting area is calculated, and the radius of the circular light-transmitting area is R. 8 . 8.如权利要求7所述的基于相关滤波的玉米种子内部裂纹检测方法，其特征在于，步骤S400中裂纹筛选进一步包括：8. The method for detecting internal cracks in corn seeds based on correlation filtering as claimed in claim 7, wherein in step S400, crack screening further comprises: 以ci为蒙版，以边缘像素点为中心，获取k×k大小的矩形图像区域p并进行傅里叶变换为P，则其相关滤波响应为：Taking c i as the mask and the edge pixel as the center, obtaining a rectangular image area p of size k×k and performing Fourier transform to P, then the relevant filter response is: Y＝H*·PY=H * ·P 其中，ci为边缘检测后的输出图像，H*为滤波模板，p为候选区域。Among them, ci is the output image after edge detection, H * is the filtering template, and p is the candidate region. 9.如权利要求8所述的基于相关滤波的玉米种子内部裂纹检测方法，其特征在于，设阈值T，若所述相关滤波响应的峰值大于或等于阈值T，则所述候选区域为裂纹区域；若所述相关滤波响应的峰值小于阈值T，则所述候选区域为非裂纹区域。9. The method for detecting internal cracks in corn seeds based on correlation filtering according to claim 8, wherein a threshold value T is set, and if the peak value of the correlation filtering response is greater than or equal to the threshold value T, the candidate region is a crack region ; If the peak value of the correlation filter response is less than the threshold T, the candidate region is a non-crack region. 10.一种基于相关滤波的玉米种子内部裂纹检测方法，其特征在于，包括如下步骤：10. A method for detecting internal cracks in corn seeds based on correlation filtering, comprising the steps of: 根据种子透射图像，以裂纹像素点为中心，获取k×k大小的裂纹区域图像，构建种子裂纹数据集；According to the seed transmission image, taking the crack pixel as the center, the crack area image of size k × k is obtained, and the seed crack data set is constructed; 设目标响应为峰值为1的标准高斯函数，对裂纹图像进行傅里叶变换，并根据滤波模板计算公式，获得对应的相关滤波模板，并通过最小平方和误差进行迭代求解，获得最终的滤波模板H*；Set the target response as a standard Gaussian function with a peak value of 1, perform Fourier transform on the crack image, and obtain the corresponding correlation filter template according to the filter template calculation formula, and iteratively solve through the least square sum error to obtain the final filter template H * ; 将种子透射图像进行边缘检测，得到种子裂纹的候选区域；Perform edge detection on the seed transmission image to obtain the candidate area of the seed crack; 将边缘检测结果作为蒙版，与原种子透射图像进行像素对应；The edge detection result is used as a mask, which corresponds to the pixel of the original seed transmission image; 以边缘检测结果中候选的边缘像素点为中心，截取k×k大小的原种子透射图像作为候选区域输入，将该候选区域进行傅里叶变换后，获得该候选区域的相关滤波响应；Taking the candidate edge pixel point in the edge detection result as the center, intercepting the original sub-transmission image of size k × k as the input of the candidate region, and after performing Fourier transform on the candidate region, the correlation filter response of the candidate region is obtained; 判断相关滤波响应的峰值，若峰值大于或等于阈值T，则该候选区域为裂纹区域；若峰值小于阈值T，则该候选区域为非裂纹区域；Determine the peak value of the correlation filter response. If the peak value is greater than or equal to the threshold value T, the candidate area is a crack area; if the peak value is less than the threshold value T, the candidate area is a non-crack area; 循环遍历所有边缘检测的候选区域，若未遍历完成，则返回重新获得该候选区域的相关滤波响应；若遍历完成，则依次连接种子裂纹区域完成种子内部裂纹检测。Loop through all the candidate regions for edge detection, if the traversal is not completed, return to re-obtain the relevant filter response of the candidate region; if the traversal is completed, connect the seed crack regions in turn to complete the seed internal crack detection.
CN202011227942.5A 2020-11-06 2020-11-06 Corn seed internal crack detection method based on correlation filtering Active CN112330635B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011227942.5A CN112330635B (en) 2020-11-06 2020-11-06 Corn seed internal crack detection method based on correlation filtering

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011227942.5A CN112330635B (en) 2020-11-06 2020-11-06 Corn seed internal crack detection method based on correlation filtering

Publications (2)

Publication Number Publication Date
CN112330635A true CN112330635A (en) 2021-02-05
CN112330635B CN112330635B (en) 2024-08-02

Family

ID=74315606

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011227942.5A Active CN112330635B (en) 2020-11-06 2020-11-06 Corn seed internal crack detection method based on correlation filtering

Country Status (1)

Country Link
CN (1) CN112330635B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113791078A (en) * 2021-09-02 2021-12-14 中国农业机械化科学研究院 Method and device for batch detection of internal cracks in corn seeds

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2006162583A (en) * 2004-11-10 2006-06-22 Taisei Corp Crack detection method
CN104700417A (en) * 2015-01-19 2015-06-10 湖南大学 Computer image based automatic identification method of timber knot flaws
CN108279294A (en) * 2017-12-27 2018-07-13 江苏省建筑工程质量检测中心有限公司 For steel structure bridge health monitoring without loss automatic monitoring system and method
CN110231341A (en) * 2019-04-29 2019-09-13 中国科学院合肥物质科学研究院 A kind of rice paddy seed underbead crack on-line measuring device and its detection method
CN110412045A (en) * 2019-07-04 2019-11-05 中国农业机械化科学研究院 A device and method for rapid detection of internal cracks in transmissive corn seeds
CN111599006A (en) * 2020-05-22 2020-08-28 山东农业大学 Three-dimensional reconstruction method and system for internal cracks of corn seeds

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2006162583A (en) * 2004-11-10 2006-06-22 Taisei Corp Crack detection method
CN104700417A (en) * 2015-01-19 2015-06-10 湖南大学 Computer image based automatic identification method of timber knot flaws
CN108279294A (en) * 2017-12-27 2018-07-13 江苏省建筑工程质量检测中心有限公司 For steel structure bridge health monitoring without loss automatic monitoring system and method
CN110231341A (en) * 2019-04-29 2019-09-13 中国科学院合肥物质科学研究院 A kind of rice paddy seed underbead crack on-line measuring device and its detection method
CN110412045A (en) * 2019-07-04 2019-11-05 中国农业机械化科学研究院 A device and method for rapid detection of internal cracks in transmissive corn seeds
CN111599006A (en) * 2020-05-22 2020-08-28 山东农业大学 Three-dimensional reconstruction method and system for internal cracks of corn seeds

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
挥着斧头的程序员: "【目标跟踪】相关滤波算法之MOSSE", HTTPS://WWW.CNBLOGS.COM/HOATSON/P/8093869.HTML, 23 December 2017 (2017-12-23), pages 1 - 2 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113791078A (en) * 2021-09-02 2021-12-14 中国农业机械化科学研究院 Method and device for batch detection of internal cracks in corn seeds
CN113791078B (en) * 2021-09-02 2023-06-13 中国农业机械化科学研究院 Batch detection method and device for internal cracks of corn seeds

Also Published As

Publication number Publication date
CN112330635B (en) 2024-08-02

Similar Documents

Publication Publication Date Title
Barbedo et al. Detecting Fusarium head blight in wheat kernels using hyperspectral imaging
JP7200113B2 (en) Systems and methods for training and applying defect classifiers on wafers with deeply stacked layers
CN104123542B (en) Device and method for positioning hub workpiece
CN110210434A (en) Pest and disease damage recognition methods and device
Chaugule et al. Identification of paddy varieties based on novel seed angle features
CN109145955B (en) Method and system for wood identification
CN108985237A (en) A kind of detection method and its system of the wheat scab based on depth mixing
WO2022178653A1 (en) Biochip image analysis method and apparatus, and computer device and storage medium
CN109376782B (en) Support vector machine cataract classification method and device based on eye image features
CN112330635A (en) A method for detecting internal cracks in corn seeds based on correlation filtering
Ali et al. Optic Disc Localization in Retinal Fundus Images Based on You Only Look Once Network (YOLO).
Jose et al. Early detection and classification of apple leaf diseases by utilizing ifpa genetic algorithm with mc-svm, svi and deep learning methods
CN109117743A (en) A kind of detection method and its system of the wheat scab based on depth convolution
JP5401605B2 (en) Template matching processing apparatus and template matching processing program
CN112365489A (en) Ultrasonic image blood vessel bifurcation detection method
Abdullah et al. A portable rice disease diagnosis tool basedon bi-level color image thresholding
Qiu et al. Method for measurement of maize stem diameters based on RGB-D camera
Wei et al. A study of image processing on identifying cucumber disease
Singh et al. Automated detection of plant leaf diseases using image processing techniques
CN208857289U (en) Human papillomavirus automatic typing and detection device based on deep learning
CN113588571A (en) Hyperspectral imaging-based method and system for identifying fishy smell of aquatic product
JP2022013285A (en) Machine learning method
Sinada et al. A reflective analysis of image processing operations on Kato-Katz images for the pathological diagnosis of neglected tropical diseases
Zhou et al. Weighted ensemble networks for multiview based tiny object quality assessment
Elakya et al. Enhancing Disease Classification in Paddy Fields: A Stratified Cross-Validation Approach

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant