CN-121982427-A - Popcorn pattern identification method, apparatus and computer program product

CN121982427ACN 121982427 ACN121982427 ACN 121982427ACN-121982427-A

Abstract

The application provides a method, a device and a computer program product for identifying the pattern of popped corn, which relate to the technical field of plant breeding, and the method comprises the steps of acquiring an image of corn kernels to be identified; the method comprises the steps of extracting corn kernel contours of a corn kernel image to be identified to obtain at least one kernel closed contour, calculating the relative length-width ratio of each corn kernel to be identified according to the at least one kernel closed contour, acquiring an anatomic kernel image of each corn kernel to be identified, wherein the anatomic kernel image comprises endosperm structure images of each corn kernel to be identified, calculating the floury endosperm ratio of each corn kernel to be identified according to the anatomic kernel image of each corn kernel to be identified, and determining the flower type of each corn kernel to be identified according to the relative length-width ratio and the floury endosperm ratio of each corn kernel to be identified, so that the problem that the flower type of each corn kernel to be identified is difficult to accurately judge under the condition that the germination capacity of the kernels is not damaged is solved.

Inventors

ZHU XIANWEN
YANG XIAOYUAN
LIU TENG
YU JIALIN
ZHANG XINGPING
LIU ZHIWEI

Assignees

北京大学现代农业研究院
潍坊现代农业山东省实验室

Dates

Publication Date: 20260505
Application Date: 20260327

Claims (10)

1. A method for identifying the pattern of popped corn, comprising: acquiring an image of a popped corn kernel to be identified, wherein the image of the popped corn kernel to be identified comprises at least one image of the popped corn kernel to be identified; extracting the corn kernel contour of the burst corn kernel image to be identified to obtain at least one kernel closed contour; calculating the relative length-width ratio of each burst corn kernel to be identified according to at least one kernel closed contour; Acquiring an anatomic seed image of each burst corn seed to be identified, wherein the anatomic seed image comprises an endosperm structure image of each burst corn seed to be identified; calculating the floury endosperm ratio of each of the popcorn kernels to be identified according to the anatomical kernel images of each of the popcorn kernels to be identified; and determining the flower type of each burst corn seed to be identified according to the relative length-width ratio and the floury endosperm ratio of each burst corn seed to be identified.
2. The method of claim 1, wherein extracting the corn kernel contour from the image of the popped corn kernel to be identified results in at least one kernel closed contour, comprising: Converting the burst corn kernel image to be identified into an HSV color space to obtain a first HSV image; Extracting a foreground image of the first HSV image according to a color threshold range to obtain a corn kernel foreground image; carrying out connected domain analysis on the corn kernel foreground image to obtain at least one candidate kernel region; Removing the candidate grain areas with the contour areas smaller than a first area threshold; performing corrosion-expansion combination operation on at least one of the remaining candidate grain areas to obtain at least one candidate grain independent area; and extracting the grain closed contour corresponding to the at least one candidate grain independent area.
3. The method of claim 1, wherein calculating the relative aspect ratio of each of the pop corn kernels to be identified based on at least one of the kernel closure profiles comprises: extracting the minimum circumscribed rectangle of the closed outline of the grain to obtain at least one minimum circumscribed rectangle of the grain; Determining a first side length of each grain minimum circumscribed rectangle as a pixel level length of the corresponding burst corn grain to be identified, determining a second side length of each grain minimum circumscribed rectangle as a pixel level width of the corresponding burst corn grain to be identified, wherein the first side length is a side length from the top end to the base of the burst corn grain to be identified in the grain minimum circumscribed rectangle, and the second side length is a side length of a side perpendicular to the first side length in the grain minimum circumscribed rectangle; Calculating the product of the pixel-level length of each corn kernel and a size conversion coefficient to obtain the actual length of each burst corn kernel to be identified, and calculating the product of the pixel-level width of each corn kernel and the size conversion coefficient to obtain the actual width of each burst corn kernel to be identified, wherein the size conversion coefficient is the conversion coefficient of the pixel-level size and the actual size; calculating the ratio of the actual width of the to-be-identified popped corn kernels corresponding to the actual length of each to-be-identified popped corn kernel to obtain the relative length-width ratio of each to-be-identified popped corn kernel.
4. The method of claim 3, wherein prior to calculating the relative aspect ratio of each of the popped corn kernels to be identified based on at least one of the kernel closure profile and the size conversion factor, the method further comprises: Extracting a red channel image from the burst corn kernel image to be identified under the condition that the burst corn kernel image to be identified has uniform brightness, and determining the red channel image as a target image; Under the condition that the brightness of the image of the pop corn kernel to be identified is uniform, converting the image of the pop corn kernel to be identified into Lab space to obtain a brightness channel image, converting the image of the pop corn kernel to be identified into a gray level image to obtain a gray level image of the pop corn kernel to be identified, and determining the brightness channel image or the gray level image of the pop corn kernel to be identified as the target image; Threshold segmentation is carried out on the target image to obtain a binary image; Performing contour detection on the binary image to obtain at least one closed region contour; Deleting the closed area contour with the contour area not within a preset range, and processing the rest closed area contour by adopting a polygonal approximation method to obtain a quadrilateral target contour; determining the quadrangular target contour as a reference contour, wherein the pop corn kernel image to be identified comprises a reference image; And calculating the ratio of the actual size of the reference object corresponding to the reference object outline to the pixel size of the reference object outline to obtain the size conversion coefficient.
5. The method of claim 1, wherein calculating a floury endosperm fraction for each of the popped corn kernels to be identified from the anatomical kernel images of each of the popped corn kernels to be identified comprises: a conversion step of converting a target anatomic seed image into an HSV color space to obtain a second HSV image, wherein the target anatomic seed image is an anatomic seed image corresponding to any one of the pop corn seeds to be identified; a first extraction step of extracting a white area of the second HSV image according to a saturation threshold range and a brightness threshold range to obtain at least one white area; an operation step of performing morphological opening and closing operation on at least one white area to obtain at least one independent white area; a second extraction step of removing the independent white region with the contour area smaller than a second area threshold value, and extracting the contour of the rest independent white region to obtain at least one endosperm contour; A third extraction step of extracting the outline of the burst corn kernel to be identified in the target anatomic kernel image to obtain a kernel outline; a first calculation step of calculating the total area of at least one endosperm contour to obtain the total area of the floury endosperm; a second calculation step of calculating the outline area of the outline of the grain to obtain the total area of endosperm of the grain; A third calculation step of calculating the ratio of the total area of the floury endosperm to the total area of the kernel endosperm to obtain the floury endosperm ratio of the pop corn kernels to be identified, which corresponds to the target anatomical kernel image; And sequentially repeating the conversion step, the first extraction step, the operation step, the second extraction step, the third extraction step, the first calculation step, the second calculation step and the third calculation step at least once to obtain the flour endosperm ratio of each of the popcorn kernels to be identified.
6. The method of claim 1, wherein determining the floral shape of each of the identified popped corn kernels based on the relative aspect ratio and the floury endosperm fraction of each of the identified popped corn kernels comprises: determining that the flower type of the pop corn kernel to be identified is a spherical flower type under the conditions that the pop corn kernel to be identified belongs to an inbred line group, the relative length-to-width ratio of the pop corn kernel to be identified is in a first length-to-width ratio interval and the flour endosperm ratio of the pop corn kernel to be identified is in a first flour endosperm ratio interval; Determining that the flower type of the pop corn kernel to be identified is a butterfly flower type under the conditions that the pop corn kernel to be identified belongs to the inbred line population, the relative length-to-width ratio of the pop corn kernel to be identified is in a second length-to-width ratio interval, and the floury endosperm ratio of the pop corn kernel to be identified is in a second floury endosperm ratio interval; Determining that the flower type of the burst corn kernel to be identified is a spherical flower type under the conditions that the burst corn kernel to be identified belongs to a separation population, the relative length-to-width ratio of the burst corn kernel to be identified is in a third length-to-width ratio interval and the flour endosperm ratio of the burst corn kernel to be identified is in a third flour endosperm ratio interval; And determining that the flower type of the burst corn kernel to be identified is a butterfly flower type under the conditions that the burst corn kernel to be identified belongs to the separation population, the relative length-width ratio of the burst corn kernel to be identified is in a fourth length-width ratio interval and the flour endosperm proportion of the burst corn kernel to be identified is in a fourth flour endosperm proportion interval.
7. The method of claim 6, wherein the first aspect ratio interval is [0.96,1.04], the first floury endosperm fraction interval is [39.87%, 64.53%), the second aspect ratio interval is [1.07,1.35], the second floury endosperm fraction interval is [20.3%,31.82% ], the third aspect ratio interval is (0,0.7 ], the third floury endosperm is in the ratio interval of [72%, + -infinity ], the fourth aspect ratio interval is [0.75, + -infinity), and the fourth floury endosperm is in the ratio interval of (0, 65% ].
8. The method of any one of claims 1 to 7, wherein after determining the flower shape of each of the identified popped corn kernels based on the relative aspect ratio and the floury endosperm fraction of each of the identified popped corn kernels, the method further comprises: generating a structured Excel data report of the image of the popped corn kernels to be identified according to the total endosperm area, the floury endosperm area and the floury endosperm ratio of each of the image of the popped corn kernels to be identified; and generating a result image with visual labels according to the image of the pop corn kernels to be identified, wherein the visual labels comprise a minimum circumscribed rectangle of the kernels, endosperm outline and number.
9. A method for identifying the pattern of popped corn, comprising: Obtaining the relative length-width ratio and the floury endosperm ratio of each popped corn kernel to be identified; Determining that the flower type of the popcorn kernel to be identified is a spherical flower type under the conditions that the popcorn kernel to be identified belongs to an inbred line population, the relative length-width ratio of the popcorn kernel to be identified is within [0.96,1.04] and the floury endosperm ratio of the popcorn kernel to be identified is within [39.87% and 64.53%; Determining that the flower type of the pop corn kernel to be identified is a butterfly flower type under the conditions that the pop corn kernel to be identified belongs to the inbred line population, the relative length-width ratio of the pop corn kernel to be identified is within [1.07,1.35] and the floury endosperm ratio of the pop corn kernel to be identified is within [20.3%,31.82% ]; in the case where the popcorn kernel to be identified belongs to an isolated population, the relative aspect ratio of the popcorn kernel to be identified is within (0,0.7), and the floury endosperm fraction of the popcorn kernel to be identified is at 72%, in +++). In the case of a situation in which the number of the elements, determining that the flower type of the burst corn kernels to be identified is a spherical flower type; And determining that the flower type of the burst corn kernel to be identified is a butterfly flower type under the conditions that the burst corn kernel to be identified belongs to the separation group, the relative length-width ratio of the burst corn kernel to be identified is within [0.75, ++ ] and the floury endosperm ratio of the burst corn kernel to be identified is within (0, 65% ].
10. A popping popcorn type identification apparatus, comprising: the first acquisition unit is used for acquiring an image of the popped corn seeds to be identified, wherein the image of the popped corn seeds to be identified comprises at least one image of the popped corn seeds to be identified; the extraction unit is used for extracting the corn kernel outline of the burst corn kernel image to be identified to obtain at least one kernel closed outline; A first calculation unit for calculating the relative aspect ratio of each of the popcorn kernels to be identified based on at least one of the kernel closure profiles; The second acquisition unit is used for acquiring an anatomic seed grain image of each burst corn seed grain to be identified, wherein the anatomic seed grain image comprises an endosperm structure image of the burst corn seed grains to be identified; a second calculation unit for calculating the floury endosperm fraction of each of the popcorn kernels to be identified from the anatomical kernel images of each of the popcorn kernels to be identified; And the determining unit is used for determining the flower type of each burst corn seed to be identified according to the relative length-width ratio and the floury endosperm ratio of each burst corn seed to be identified.

Description

Popcorn pattern identification method, apparatus and computer program product Technical Field The application relates to the technical field of plant breeding, in particular to a pop corn pattern identification method, a pop corn pattern identification device, a computer readable storage medium and a computer program product. Background Puffed popcorn of popped corn mainly includes three major categories, typically spherical, typically butterfly, and mixed, with spherical and butterfly being the two most representative core types in production and breeding. The two have obvious differences in the expansion volume, the texture structure and the commodity processing suitability, and are key phenotypic characters which are important to be focused in breeding improvement. At present, the molecular mechanism of the formation of the popcorn type is not completely elucidated, the traditional identification depends on a popping test, has irreversibility, causes the loss of germination capacity of seeds, and cannot meet the requirements of resource recycling and early large-scale screening in breeding. The existing burst corn pattern identification technology mainly comprises the following two major directions of core surrounding 'phenotype association' and 'direct verification', wherein one type is based on a physiological mechanism formed by the pattern, the pattern of the popped seeds is determined by the endosperm structure, the seed pattern, the seed coat structure and other phenotypes, the pattern is deduced indirectly through the association index, the other type is directly dependent on the popping reaction, and identification is completed according to the visual pattern of the popped pattern, which is also the mainstream verification logic in the industry at present. The existing flower type identification method for burst corn and the corresponding defects and pain points are as follows: Firstly, the traditional puffing test identification method is the most basic and widely applied method, the principle is that the processing conditions are simulated by a popcorn machine, the patterns (spheres/butterflies) are directly observed to finish the judgment after the grains are exploded, but the core pain point is that the identification process is irreversible and destructive, the exploded grains completely lose germination capacity and cannot be used as breeding materials to continue seed reserving, hybridization or subsequent character observation, and for the separation groups with complex genetic background and scarce materials, a large amount of precious breeding materials are lost, the generation breeding and resource retention of the separation groups are severely restricted, meanwhile, the method is complex in operation and long in time consumption, the single-batch processing sample size is limited, the large-scale and high-flux identification requirements of the separation groups cannot be met, and the patterns judgment deviation is easy to occur due to the influence of parameters such as explosion temperature, time and the like. Secondly, part of research attempts to observe endosperm structures through dissected grains and judge flower patterns by combining experience, and the principle is based on potential correlation of endosperm textures, distribution and flower patterns, but the method has multiple defects that firstly, the correlation judgment of endosperm structures and flower patterns is strong in subjectivity and high in error rate depending on experience of operators, under the scene of complex phenotype difference of separation groups, identification consistency is difficult to guarantee, secondly, the operation flow is complicated, manual observation and counting are needed by microscopic equipment, the efficiency is extremely low, and the method cannot adapt to the breeding requirement of large-scale sample screening of the separation groups. The method comprises the following steps of determining the grain shape index by using an image recognition technology, deducing the flower type indirectly, wherein the principle is based on the correlation of the grain shape and part of the phenotype of the flower type, but the technology is not mature, and obvious short plates exist, wherein an algorithm is not optimized for the genetic characteristics of a separation group, the single index is easily influenced by genetic background and growth environment, is not suitable for a lossless operation flow, and is still difficult to consider the core requirements of identification accuracy and germplasm resource protection, and the method cannot be used as a reliable means for identifying the flower type of pop corn. Disclosure of Invention The main object of the present application is to provide a method, a device, a computer readable storage medium and a computer program product for identifying the pattern of popped corn kernels, which at least solve the problem that in the prior art, the pattern o