CN112381662B - Pollen fertility rate assessment method and device - Google Patents
Pollen fertility rate assessment method and device Download PDFInfo
- Publication number
- CN112381662B CN112381662B CN202011193730.XA CN202011193730A CN112381662B CN 112381662 B CN112381662 B CN 112381662B CN 202011193730 A CN202011193730 A CN 202011193730A CN 112381662 B CN112381662 B CN 112381662B
- Authority
- CN
- China
- Prior art keywords
- pollen
- fertility
- image
- area
- assessment method
- 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.)
- Active
Links
- 230000035558 fertility Effects 0.000 title claims abstract description 122
- 238000000034 method Methods 0.000 title claims abstract description 44
- 230000003044 adaptive effect Effects 0.000 claims abstract description 25
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 22
- 238000012545 processing Methods 0.000 claims abstract description 17
- 239000002245 particle Substances 0.000 claims abstract description 15
- 238000005520 cutting process Methods 0.000 claims abstract description 11
- 238000009826 distribution Methods 0.000 claims abstract description 11
- 238000012549 training Methods 0.000 claims description 13
- 238000011156 evaluation Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 11
- 238000004590 computer program Methods 0.000 claims description 9
- 238000011176 pooling Methods 0.000 claims description 8
- 238000003860 storage Methods 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 4
- 230000000295 complement effect Effects 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 230000002222 downregulating effect Effects 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 230000001105 regulatory effect Effects 0.000 claims description 3
- 230000009977 dual effect Effects 0.000 abstract description 4
- 230000010287 polarization Effects 0.000 abstract description 4
- 238000004422 calculation algorithm Methods 0.000 description 6
- 239000000843 powder Substances 0.000 description 6
- 238000004891 communication Methods 0.000 description 5
- 210000004027 cell Anatomy 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 239000013598 vector Substances 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 235000010469 Glycine max Nutrition 0.000 description 2
- 244000068988 Glycine max Species 0.000 description 2
- 206010028980 Neoplasm Diseases 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000009395 breeding Methods 0.000 description 2
- 230000001488 breeding effect Effects 0.000 description 2
- 201000011510 cancer Diseases 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000013508 migration Methods 0.000 description 2
- 230000005012 migration Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 239000000523 sample Substances 0.000 description 2
- 241000196324 Embryophyta Species 0.000 description 1
- DKNPRRRKHAEUMW-UHFFFAOYSA-N Iodine aqueous Chemical compound [K+].I[I-]I DKNPRRRKHAEUMW-UHFFFAOYSA-N 0.000 description 1
- 206010041067 Small cell lung cancer Diseases 0.000 description 1
- 241000592344 Spermatophyta Species 0.000 description 1
- 210000004102 animal cell Anatomy 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 239000012804 pollen sample Substances 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 208000000649 small cell carcinoma Diseases 0.000 description 1
- 208000000587 small cell lung carcinoma Diseases 0.000 description 1
- 238000010186 staining Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Forestry; Mining
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
- G06T2207/30188—Vegetation; Agriculture
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biomedical Technology (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Business, Economics & Management (AREA)
- Artificial Intelligence (AREA)
- Agronomy & Crop Science (AREA)
- Mining & Mineral Resources (AREA)
- Marine Sciences & Fisheries (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Animal Husbandry (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the invention provides a pollen fertility rate assessment method and device, wherein the method comprises the following steps: acquiring a pollen image to be evaluated, obtaining a plurality of areas according to the pollen distribution state, extracting characteristics capable of distinguishing the fertile pollen and the non-fertile pollen from each pollen area, and carrying out adaptive contrast adjustment on the areas according to the characteristics; positioning and cutting single pollen by double-layer convolution operation on the adjusted image to obtain a standard image containing single pollen grains; inputting a standard image of a single pollen grain into a preset convolutional neural network model, and outputting a fertility or non-fertility classification result of the pollen grain; and obtaining the ratio of the number of the viable pollen to the number of the non-viable pollen in the original picture according to the classification result of each pollen particle, and determining the fertility rate of the pollen. The method can strengthen dual polarization of the fertility and the non-fertility state and improve the classification accuracy. The classification is obtained without manual statistics, so that the processing efficiency can be improved, and accuracy errors caused by subjective judgment are avoided.
Description
Technical Field
The invention relates to the field of crop breeding evaluation, in particular to a pollen fertility rate evaluation method and device.
Background
Pollen is a microspore heap of seed plants, mature pollen grains are small gametophytes of which male gametes can be produced. Pollen is produced from anthers in stamens, reaches pistils by various methods, pollinates ovules. Therefore, the breeding rate of crops is often closely related to the fertility rate of pollen.
The current pollen fertility rate evaluation is usually carried out according to that pollen is shaken off from anthers, and then the pollen is prepared into tablets by iodine-potassium iodide staining, and then is observed under an electron microscope and subjected to statistical analysis. The existing pollen fertility rate assessment method consumes a large amount of manpower and material resources, and the accuracy of results is low due to manual statistics errors caused by complicated calculation amount.
The data show that the deep learning-convolution network is endlessly applied to plant or animal cell processing, but the processing of soybean pollen is still in a relatively original stage, most of the existing algorithms are based on the traditional digital image processing technology to perform gray level conversion, edge detection, morphological operation and the like on images so as to make preliminary detection estimation judgment, the accuracy and the robustness are poor, and large-scale effective utilization cannot be performed. The application of deep learning and convolutional networks has been widely used and advanced to deal with the relatively similar problem, namely the identification and monitoring of cancer cells. Due to the similarity of identification and classification of cancer cells and soybean pollen cells, certain migration learning can be performed, and a related algorithm is reasonably utilized and improved to obtain higher identification accuracy. Some of the algorithms have strong reference utility value for processing pollen images. For example, the method for assisting to detect the small-cell lung cancer by adopting the convolution network algorithm can be utilized to identify and detect the irregular small-cell pollen sample, the convolution network well combines the advantages of the prior detection algorithm, the accuracy can be considered, the misjudgment rate can be better reduced, and the learning efficiency can be improved. In addition, by adopting an improved algorithm based on a deep convolution network, ZCA is utilized to preprocess cells so as to lighten the correlation of image data characteristics, and the method can be applied to preprocessing of pollen images in a migration manner, lighten the correlations between pollen and pollen, between pollen and background images and between pollen and impurities, and reduce the misjudgment rate.
Disclosure of Invention
The embodiment of the invention provides a pollen fertility rate assessment method and device, which are used for solving the defects in the prior art.
The embodiment of the invention provides a pollen fertility rate assessment method, which comprises the following steps: acquiring pollen images to be evaluated, obtaining a plurality of pollen areas according to the pollen distribution state, extracting characteristics capable of distinguishing fertile pollen and non-fertile pollen from each pollen area, and respectively performing adaptive contrast adjustment according to the distinguishing result of the pollen areas; positioning and cutting pollen areas through double-layer convolution operation on the adjusted images to obtain standard images containing single pollen grains; inputting a standard image of a single pollen grain into a preset convolutional neural network model, and outputting a fertility or non-fertility classification result of the pollen grain; obtaining the ratio of the number of the viable pollen to the number of the non-viable pollen in the pollen image to be evaluated according to the classification result of each pollen particle, and determining the fertility rate of the pollen; the preset convolutional neural network model is obtained after training according to standard images of single Zhang Hua powder particles with labels.
According to one embodiment of the invention, the method for evaluating pollen fertility rate comprises the following steps of: if the pollen area is the area with the fertile pollen, the brightness of the whole area is adjusted downwards; if the pollen area is a fertility pollen-free area, the brightness of the whole area is up-regulated.
According to the pollen fertility rate assessment method of one embodiment of the present invention, if the pollen region is both viable and sterile pollen, the extracted features further include: BG max、BRmax and BGR; correspondingly, performing adaptive contrast adjustment according to BGR;
Wherein:
BGR＝[BGmax,BRmax]|min；
r, G, B are the corresponding values of the RGB color space, respectively.
According to one embodiment of the invention, the pollen fertility rate assessment method performs adaptive contrast adjustment according to BGR, and comprises the following steps: when the brightness of the whole region is adjusted downwards, if the gray level is smaller than the region adjustment threshold value, the gray level value is set to be a fixed value and is not adjusted downwards any more; the region adjustment threshold is determined as follows:
Wherein bblow and dd are experience parameters, a is a brightness down-regulating factor and also is an experience parameter; gray min is the minimum value of the regional Gray scale before adjustment, and low in is the regional adjustment threshold.
According to the pollen fertility rate assessment method of one embodiment of the invention, after respectively performing adaptive contrast adjustment according to the distinguishing result of the pollen areas, the method further comprises: and (3) filling holes in the binarized image of the adjusted image to complement the hollow fertile pollen.
According to one embodiment of the invention, the pollen fertility rate assessment method inputs a standard image of a single pollen grain into a preset convolutional neural network model, and comprises the following steps: inputting a three-channel color map of a single pollen grain into two convolution layers of a convolution neural network, wherein each convolution layer is next to a batch normalization processing layer, and outputting by adopting LeakyReLU activation functions; and inputting a pooling layer adopting Max-pooling, discarding 50% after passing through two full-connection layers, and finally obtaining the classification probability through a softmax function.
The embodiment of the invention also provides a pollen fertility rate assessment device, which comprises a feature extraction module, a characteristic analysis module and a pollen fertility rate assessment module, wherein the feature extraction module is used for acquiring a pollen image to be assessed, obtaining a plurality of pollen areas according to the pollen distribution state, extracting features capable of distinguishing fertility pollen and non-fertility pollen from each pollen area, and respectively carrying out adaptive contrast adjustment according to the distinguishing result of the pollen areas; the dividing and cutting module is used for positioning and cutting the pollen area through double-layer convolution operation on the adjusted image to obtain a standard image containing single pollen grains; the convolution classification module is used for inputting a standard image of a single pollen grain into a preset convolution neural network model and outputting a fertility or non-fertility classification result of the pollen grain; the fertility evaluation module is used for obtaining the ratio of the number of the fertility pollen to the number of the non-fertility pollen in the pollen image to be evaluated according to the classification result of each pollen particle, and determining the fertility rate of the pollen; the preset convolutional neural network model is obtained after training according to standard images of single Zhang Hua powder particles with labels.
The embodiment of the invention also provides electronic equipment, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor realizes the steps of any pollen fertility rate assessment method when executing the program.
The embodiments of the present invention also provide a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of a pollen fertility rate assessment method as described in any one of the above.
According to the pollen fertility rate assessment method and device provided by the embodiment of the invention, the adaptive contrast adjustment is carried out on each pollen area according to the extracted characteristics, so that the dual polarization of fertility and non-fertility states can be enhanced, and the classification accuracy is improved. The characteristic data of each pollen area is input into a preset convolutional neural network model to obtain a classification result, manual statistics is not needed, the processing efficiency can be improved, and accuracy errors caused by subjective judgment in the manual processing process are avoided.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a pollen fertility rate assessment method provided by an embodiment of the invention;
FIG. 2 is a schematic diagram of a pollen fertility rate assessment device according to an embodiment of the present invention;
Fig. 3 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Fig. 1 is a flow chart of a pollen fertility rate evaluation method according to an embodiment of the present invention, and as shown in fig. 1, the embodiment of the present invention provides a pollen fertility rate evaluation method, including:
101. Obtaining a pollen image to be evaluated, obtaining a plurality of pollen areas according to the pollen distribution state, extracting feature vectors capable of distinguishing the fertile pollen and the non-fertile pollen from each pollen area, and respectively carrying out adaptive contrast adjustment according to the distinguishing result of the pollen areas.
To evaluate fertility of pollen of a crop, a pollen image including a plurality of pollens may be acquired first. For example, after a sample to be evaluated is collected, a tablet is made, and an image is taken under a microscope to obtain a pollen image to be evaluated. The pollen image comprises a plurality of pollens, and the outline of the pollens adjacent to the pollens can obtain a pollen area. And extracting the characteristics of each pollen area to judge the fertility of the pollen in the area. For example, the extracted features are RGB values of the three-channel image, RGB-related combined values, gray maximum and minimum values, and the like.
The areas with and without viable pollen can be defined according to the extracted characteristics. For example, the threshold range in the gray space is determined according to each pollen region. Extracting the characteristics which can distinguish the viable pollen from the non-viable pollen, and carrying out corresponding judgment.
For the pollen area image with obvious color distinction, the converted gray level image is binarized, and whether pollen at the corresponding position is fertile or not can be determined according to the corresponding threshold value. However, the regional dataset has a plurality of images with unobvious color distinction, and the basic characteristic is that the fertility pollen and the non-fertility pollen are in a critical state, which causes great difficulty in judging the determination state (fertility and non-fertility) of the regional dataset. In the second case, the viable pollen and the non-viable pollen are in a hollow state, so that the middle part is easy to filter out after the subsequent binarization, only one aperture is left, the convolution count cannot reach a specific threshold value and is ignored, and the result is inaccurate. Therefore, before the original image is subjected to binarization treatment to obtain the fertility-promoting and non-fertility-promoting pollen, the image must be subjected to adaptive contrast adjustment, so that the transition state is reduced and the dual polarization of the state is enhanced by the treatment. The purpose of filtering background noise and enhancing the contrast of the viable pollen and the non-viable pollen in the original picture is achieved by carrying out adaptive contrast adjustment on the original picture.
102. And (3) positioning and cutting the pollen area through double-layer convolution operation on the adjusted image to obtain a standard image containing single pollen grains.
For the problems that the adhesion of pollen is serious and the counting is seriously affected by clustered pollen clusters caused by improper sample preparation, the adhesion-shaped pollen clusters occupy too large area (usually consisting of 2-6 pollens), are odd-shaped, and cannot be separated after one convolution. To solve this problem, a multi-layer convolution network was constructed with the aim of counting the original cluster of make-up pollen clusters multiple times by multiple convolutions. Through simulation experiments, the double-layer convolution structure can obtain an ideal processing result under low consumption. The adjusted standardized image data is subjected to a double-layer convolution operation to initially extract the possible pixel positions of pollen and to separate and count overlapping pollen. Then, the square area is truncated with the determined position as the center and the pollen diameter as the side length to obtain image data of single pollen, that is, standard image of single pollen grains. In training, the standard image is labeled with categories (fertile and non-fertile) and then used as a training and testing set for the classifier. After training is completed, it is used for classification in 103.
103. And inputting the standard image of the single pollen grain into a preset convolutional neural network model, and outputting a fertility or non-fertility classification result of the pollen grain.
And (3) a preset convolutional neural network model is obtained after training by taking a fertility result (fertility and non-fertility) as a label according to a pollen grain standard image of the fertility result as input. After model training is completed, standard images of single pollen grains are input into the model, and then fertility classification results of each pollen grain can be obtained respectively.
104. Determining the fertility rate of pollen according to the number of the fertility pollen and the number of the non-fertility pollen obtained by the classification result,
The pollen distribution binary image of the fertility part and the non-fertility part is extracted through complex convolution network processing. In order to more intuitively observe the fertility and the non-fertility identification accuracy, the positions of the fertility part and the non-fertility part are marked on the original image of the pollen image to be evaluated, and the results can be identified with high accuracy and efficiency and marked at corresponding positions by using outline detection functions cv2.findContours () and cv2.drawContours () based on an OpenCV library. The pollen fertility rate represented by the data can be determined by the following formula:
Wherein n1 is the number of fertile pollen grains and n2 is the number of non-fertile pollen grains.
According to the pollen fertility rate assessment method, the adaptive contrast adjustment is carried out on each pollen area according to the extracted characteristics, so that dual polarization of fertility and non-fertility states can be enhanced, and classification accuracy is improved. The characteristic data of each pollen area is input into a preset convolutional neural network model to obtain a classification result, manual statistics is not needed, the processing efficiency can be improved, and accuracy errors caused by subjective judgment in the manual processing process are avoided.
Based on the foregoing embodiment, as an alternative embodiment, performing adaptive contrast adjustment on each pollen region according to the extracted feature vector includes: if the pollen area is the area with the fertile pollen, the brightness of the whole area is adjusted downwards; if the pollen area is a fertility pollen-free area, the brightness of the whole area is up-regulated.
In the image processing, a bleached (camera overexposed) picture or an excessively dark (underexposed) picture is corrected. And carrying out linear or nonlinear transformation on the normalized picture data within a specific threshold range so as to correct the overexposed or excessively darkened area of the original picture, thereby achieving the purpose of balancing the data. The basic form of gamma conversion is as follows:
S＝C*Rγ；
Wherein R is a pollen image to be evaluated, S is an adjusted image, gamma control is adjusted in nonlinearity, and when the gamma value is smaller than 1, a region with lower gray level in the image is stretched, and a part with higher gray level is compressed; when the gamma value is greater than 1, the region of the image with higher gray level is stretched, and the portion with lower gray level is compressed.
The specific regulation rule is that the fertility pollen is more fertility (deepening) on the pixels, and the non-fertility pollen is more non-fertility (making shallower). In combination with the analysis of a large number of marked data, the areas with and without viable pollen and the possible viable pollen part in the gray space are defined, so that the aim of distinguishing the viable pollen and the non-viable pollen is fulfilled. For areas with viable pollen, the adjustment strategy is to adjust its overall brightness down, e.g., to 80% of the original, and for areas without viable pollen, the adjustment strategy is to adjust its brightness up, e.g., to 20% of the original.
Based on the foregoing embodiment, as an alternative embodiment, if the pollen area is both viable and sterile pollen, the extracted feature vector further includes:
BG max、BRmax and BGR;
Correspondingly, performing adaptive contrast adjustment according to BGR;
Wherein:
BGR＝[BGmax,BRmax]|min；
r, G, B are the corresponding values of the RGB color space, respectively.
Based on the foregoing embodiment, as an alternative embodiment, performing adaptive contrast adjustment according to BGR includes: when the brightness of the whole region is adjusted downwards, if the gray level is smaller than the region adjustment threshold value, the gray level value is set to be a fixed value and is not adjusted downwards any more;
The region adjustment threshold is determined as follows:
Wherein bblow is an empirical value, which can be obtained from data analysis, 10 in this example, dd is an empirical value, which can be obtained from data analysis, 1.255 in this example. a is a brightness down-regulating factor, which is 20 in this embodiment. Gray min is the minimum value of Gray before adjustment, low in is the set region adjustment threshold, is the minimum value of overall adjustment, and when the Gray is smaller than the minimum value, the Gray is not changed any more and is set to be a fixed value.
Based on the foregoing embodiment, as an alternative embodiment, after performing adaptive contrast adjustment on each pollen area according to the extracted feature vector, the method further includes: and (3) filling holes in the binarized image of the adjusted image to complement the hollow fertile pollen.
After preliminary brightness adjustment, the problem of hollow of the viable pollen can not be completely solved due to the allowable range of large-range brightness down-adjustment, and especially the middle of the viable pollen becomes white after binarization. When the next step of convolution is performed, the pollen is easily filtered out by errors due to the fact that the threshold value is not reached, and huge interference is caused to the counting of the later viable pollen. To solve this problem, the binarized image is hole-filled to complement the hollow viable pollen. The function for hole filling may be chosen imfill to fill the hole region in the binary image. For example, there may be a white circle on a black background.
Based on the foregoing embodiment, as an alternative embodiment, inputting a standard image of a single pollen grain into a preset convolutional neural network model includes: inputting a three-channel color map of a single pollen grain into two convolution layers of a convolution neural network, wherein each convolution layer is next to a batch normalization processing layer, and outputting by adopting LeakyReLU activation functions; and inputting a pooling layer adopting Max-pooling, discarding 50% after passing through two full-connection layers, and finally obtaining the classification probability through a softmax function.
Specifically, the input picture is a three-channel color picture containing single pollen grains, the size can be 32 x 3, two layers of convolution layers are shared, the convolution kernel size is 5*5, the convolution layers are subjected to Batch Normalization (BN), the activation function adopts LeakyReLU functions, the pooling layer adopts a Max-pooling mode, two layers of full-connection layers are adopted, then 50% of the full-connection layers are discarded, and finally the classification probability is calculated through a softmax function.
The pollen fertility rate evaluation device provided by the embodiment of the invention is described below, and the pollen fertility rate evaluation device described below and the pollen fertility rate evaluation method described above can be referred to correspondingly.
Fig. 2 is a schematic structural diagram of a pollen fertility rate assessment device according to an embodiment of the invention, as shown in fig. 2, the pollen fertility rate assessment device includes: a feature extraction module 201, a partition clipping module 202, a convolution classification module 203, and a fertility assessment module 204. The feature extraction module 201 is configured to obtain a pollen image to be evaluated, obtain a plurality of pollen areas according to a pollen distribution state, extract features capable of distinguishing viable pollen and non-viable pollen from each pollen area, and respectively perform adaptive contrast adjustment according to a distinguishing result of the pollen areas; the division clipping module 202 is used for positioning and clipping the pollen area through double-layer convolution operation on the adjusted image to obtain a standard image containing single pollen grains; the convolution classification module 203 is configured to input a standard image of a single pollen grain into a preset convolution neural network model, and output a fertility or non-fertility classification result of the pollen grain; the fertility evaluation module 204 is used for obtaining the ratio of the number of the fertility pollen to the number of the non-fertility pollen in the pollen image to be evaluated according to the classification result of each pollen particle, and determining the fertility rate of the pollen; the preset convolutional neural network model is obtained after training according to standard images of single Zhang Hua powder particles with labels.
The embodiment of the device provided by the embodiment of the present invention is for implementing the above embodiments of the method, and specific flow and details refer to the above embodiments of the method, which are not repeated herein.
According to the pollen fertility rate assessment device provided by the embodiment of the invention, the characteristic data of each pollen area is input into the preset convolutional neural network model, and the classification result of the pollen in each area as the fertility pollen and the non-fertility pollen is output, so that the pollen fertility rate assessment device is obtained without manual statistics, the processing efficiency can be improved, and the accuracy error caused by subjective judgment in the manual processing process is avoided.
Fig. 3 is a schematic structural diagram of an electronic device according to an embodiment of the present invention, as shown in fig. 3, the electronic device may include: processor 301, communication interface (Communications Interface) 302, memory 303, and communication bus 304, wherein processor 301, communication interface 302, and memory 303 communicate with each other via communication bus 304. Processor 301 may invoke logic instructions in memory 303 to perform a pollen fertility assessment method comprising: acquiring pollen images to be evaluated, obtaining a plurality of pollen areas according to the pollen distribution state, extracting characteristics capable of distinguishing fertile pollen and non-fertile pollen from each pollen area, and respectively performing adaptive contrast adjustment according to the distinguishing result of the pollen areas; positioning and cutting pollen areas through double-layer convolution operation on the adjusted images to obtain standard images containing single pollen grains; inputting a standard image of a single pollen grain into a preset convolutional neural network model, and outputting a fertility or non-fertility classification result of the pollen grain; obtaining the ratio of the number of the viable pollen to the number of the non-viable pollen in the pollen image to be evaluated according to the classification result of each pollen particle, and determining the fertility rate of the pollen; the preset convolutional neural network model is obtained after training according to standard images of single Zhang Hua powder particles with labels.
Further, the logic instructions in the memory 303 may be implemented in the form of software functional units and stored in a computer readable storage medium when sold or used as a stand alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, embodiments of the present invention also provide a computer program product comprising a computer program stored on a non-transitory computer readable storage medium, the computer program comprising program instructions which, when executed by a computer, enable the computer to perform the pollen fertility assessment method provided by the above-described method embodiments, the method comprising: acquiring pollen images to be evaluated, obtaining a plurality of pollen areas according to the pollen distribution state, extracting characteristics capable of distinguishing fertile pollen and non-fertile pollen from each pollen area, and respectively performing adaptive contrast adjustment according to the distinguishing result of the pollen areas; positioning and cutting pollen areas through double-layer convolution operation on the adjusted images to obtain standard images containing single pollen grains; inputting a standard image of a single pollen grain into a preset convolutional neural network model, and outputting a fertility or non-fertility classification result of the pollen grain; obtaining the ratio of the number of the viable pollen to the number of the non-viable pollen in the pollen image to be evaluated according to the classification result of each pollen particle, and determining the fertility rate of the pollen; the preset convolutional neural network model is obtained after training according to standard images of single Zhang Hua powder particles with labels.
In yet another aspect, embodiments of the present invention also provide a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the pollen fertility rate assessment method provided by the above embodiments, the method comprising: acquiring pollen images to be evaluated, obtaining a plurality of pollen areas according to the pollen distribution state, extracting characteristics capable of distinguishing fertile pollen and non-fertile pollen from each pollen area, and respectively performing adaptive contrast adjustment according to the distinguishing result of the pollen areas; positioning and cutting pollen areas through double-layer convolution operation on the adjusted images to obtain standard images containing single pollen grains; inputting a standard image of a single pollen grain into a preset convolutional neural network model, and outputting a fertility or non-fertility classification result of the pollen grain; obtaining the ratio of the number of the viable pollen to the number of the non-viable pollen in the pollen image to be evaluated according to the classification result of each pollen particle, and determining the fertility rate of the pollen; the preset convolutional neural network model is obtained after training according to standard images of single Zhang Hua powder particles with labels.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (9)
1. A pollen fertility assessment method, comprising:
acquiring pollen images to be evaluated, obtaining a plurality of pollen areas according to the pollen distribution state, extracting characteristics capable of distinguishing fertile pollen and non-fertile pollen from each pollen area, and respectively performing adaptive contrast adjustment according to the distinguishing result of the pollen areas;
positioning and cutting pollen areas through double-layer convolution operation on the adjusted images to obtain standard images containing single pollen grains;
Inputting a standard image of a single pollen grain into a preset convolutional neural network model, and outputting a fertility or non-fertility classification result of the pollen grain;
Obtaining the ratio of the number of the viable pollen to the number of the non-viable pollen in the pollen image to be evaluated according to the classification result of each pollen particle, and determining the fertility rate of the pollen;
The preset convolutional neural network model is obtained by training based on a standard image of sample pollen grains and fertility result labels of the sample pollen grains.
2. The pollen fertility assessment method according to claim 1, wherein the adaptive contrast adjustment is performed according to the discrimination result of pollen areas, respectively, comprising:
If the pollen area is the area with the fertile pollen, the brightness of the whole area is adjusted downwards;
if the pollen area is a fertility pollen-free area, the brightness of the whole area is up-regulated.
3. The pollen fertility assessment method of claim 2, wherein if the pollen area is both viable and sterile pollen, the extracted features further comprise:
BG max、BRmax and BGR;
Correspondingly, performing adaptive contrast adjustment according to BGR;
Wherein:
BGR＝[BGmax,BRmax]|min；
r, G, B are the corresponding values of the RGB color space, respectively.
4. A pollen fertility assessment method according to claim 3, wherein adaptive contrast adjustment according to BGR comprises:
When the brightness of the whole region is adjusted downwards, if the gray level is smaller than the region adjustment threshold value, the gray level value is set to be a fixed value and is not adjusted downwards any more;
The region adjustment threshold is determined as follows:
Wherein bblow and dd are experience parameters, a is a brightness down-regulating factor and also is an experience parameter; gray min is the minimum value of the regional Gray scale before adjustment, and low in is the regional adjustment threshold.
5. The pollen fertility assessment method according to claim 1, wherein after performing adaptive contrast adjustment according to the discrimination results of pollen areas, respectively, further comprising:
and (3) filling holes in the binarized image of the adjusted image to complement the hollow fertile pollen.
6. The pollen fertility assessment method according to claim 1, wherein inputting the standard image of individual pollen grains into a predetermined convolutional neural network model comprises:
Inputting a three-channel color map of a single pollen grain into two convolution layers of a convolution neural network, wherein each convolution layer is next to a batch normalization processing layer, and outputting by adopting LeakyReLU activation functions;
and inputting a pooling layer adopting Max-pooling, discarding 50% after passing through two full-connection layers, and finally obtaining the classification probability through a softmax function.
7. A pollen fertility assessment device, comprising:
The feature extraction module is used for acquiring a pollen image to be evaluated, obtaining a plurality of pollen areas according to the pollen distribution state, extracting features capable of distinguishing the fertile pollen from the non-fertile pollen from each pollen area, and respectively carrying out adaptive contrast adjustment according to the distinguishing result of the pollen areas;
the dividing and cutting module is used for positioning and cutting the pollen area through double-layer convolution operation on the adjusted image to obtain a standard image containing single pollen grains;
The convolution classification module is used for inputting a standard image of a single pollen grain into a preset convolution neural network model and outputting a fertility or non-fertility classification result of the pollen grain;
The fertility evaluation module is used for obtaining the ratio of the number of the fertility pollen to the number of the non-fertility pollen in the pollen image to be evaluated according to the classification result of each pollen particle, and determining the fertility rate of the pollen;
The preset convolutional neural network model is obtained by training based on a standard image of sample pollen grains and fertility result labels of the sample pollen grains.
8. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor performs the steps of the pollen fertility assessment method according to any one of claims 1 to 6 when the program is executed by the processor.
9. A non-transitory computer readable storage medium having stored thereon a computer program, characterized in that the computer program when executed by a processor implements the steps of the pollen fertility assessment method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011193730.XA CN112381662B (en) | 2020-10-30 | 2020-10-30 | Pollen fertility rate assessment method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011193730.XA CN112381662B (en) | 2020-10-30 | 2020-10-30 | Pollen fertility rate assessment method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112381662A CN112381662A (en) | 2021-02-19 |
| CN112381662B true CN112381662B (en) | 2024-04-26 |
Family
ID=74577092
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011193730.XA Active CN112381662B (en) | 2020-10-30 | 2020-10-30 | Pollen fertility rate assessment method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112381662B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106339591A (en) * | 2016-08-25 | 2017-01-18 | 汤平 | Breast cancer prevention self-service health cloud service system based on deep convolutional neural network |
| CN107491733A (en) * | 2017-07-19 | 2017-12-19 | 南京农业大学 | A kind of chrysanthemum recognition methods based on deep neural network |
| CN107907541A (en) * | 2017-12-26 | 2018-04-13 | 青海省农林科学院 | The identification method of heterozygote in a kind of cabbage type rape male and female sterile system fertile plant |
| WO2018140014A1 (en) * | 2017-01-25 | 2018-08-02 | Athelas, Inc. | Classifying biological samples using automated image analysis |
| WO2019237203A1 (en) * | 2018-06-12 | 2019-12-19 | Paige Growth Technologies Inc. | Devices, systems and methods of identifying plants, plant material and plant state |
| CN111369543A (en) * | 2020-03-07 | 2020-07-03 | 北京工业大学 | Rapid pollen particle detection algorithm based on dual self-attention module |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050251347A1 (en) * | 2004-05-05 | 2005-11-10 | Pietro Perona | Automatic visual recognition of biological particles |
| CN105303531B (en) * | 2015-10-15 | 2018-01-26 | 深圳市华星光电技术有限公司 | A kind of setting contrast method and device |
| US10552663B2 (en) * | 2017-05-02 | 2020-02-04 | Techcyte, Inc. | Machine learning classification and training for digital microscopy cytology images |
| US11504748B2 (en) * | 2017-12-03 | 2022-11-22 | Seedx Technologies Inc. | Systems and methods for sorting of seeds |
-
2020
- 2020-10-30 CN CN202011193730.XA patent/CN112381662B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106339591A (en) * | 2016-08-25 | 2017-01-18 | 汤平 | Breast cancer prevention self-service health cloud service system based on deep convolutional neural network |
| WO2018140014A1 (en) * | 2017-01-25 | 2018-08-02 | Athelas, Inc. | Classifying biological samples using automated image analysis |
| CN107491733A (en) * | 2017-07-19 | 2017-12-19 | 南京农业大学 | A kind of chrysanthemum recognition methods based on deep neural network |
| CN107907541A (en) * | 2017-12-26 | 2018-04-13 | 青海省农林科学院 | The identification method of heterozygote in a kind of cabbage type rape male and female sterile system fertile plant |
| WO2019237203A1 (en) * | 2018-06-12 | 2019-12-19 | Paige Growth Technologies Inc. | Devices, systems and methods of identifying plants, plant material and plant state |
| CN111369543A (en) * | 2020-03-07 | 2020-07-03 | 北京工业大学 | Rapid pollen particle detection algorithm based on dual self-attention module |
Non-Patent Citations (1)
| Title |
|---|
| 基于卷积神经网络的玉米籽粒精选系统研制;王建宇;《中国优秀硕士论文全文数据库 农业科技辑》;正文第1-4章 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112381662A (en) | 2021-02-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | Robust saliency detection via regularized random walks ranking | |
| Waghmare et al. | Detection and classification of diseases of grape plant using opposite colour local binary pattern feature and machine learning for automated decision support system | |
| Sannakki et al. | Diagnosis and classification of grape leaf diseases using neural networks | |
| CN111860533A (en) | Image recognition method and device, storage medium and electronic device | |
| CN109740721B (en) | Wheat ear counting method and device | |
| US10388009B2 (en) | Machine-learning measurements of quantitative feature attributes | |
| CN106845497B (en) | Corn early-stage image drought identification method based on multi-feature fusion | |
| CN112949704B (en) | Tobacco leaf maturity state identification method and device based on image analysis | |
| CN113191222B (en) | Underwater fish target detection method and device | |
| Ji et al. | In-field automatic detection of maize tassels using computer vision | |
| CN112183448B (en) | Method for dividing pod-removed soybean image based on three-level classification and multi-scale FCN | |
| CN114240978A (en) | Cell edge segmentation method and device based on adaptive morphology | |
| CN115546187A (en) | Agricultural pest and disease detection method and device based on YOLO v5 | |
| CN116524224A (en) | Machine vision-based method and system for detecting type of cured tobacco leaves | |
| CN118053154A (en) | Oyster mushroom growth monitoring method and device, electronic equipment and storage medium | |
| CN114596274A (en) | Natural background citrus greening disease detection method based on improved Cascade RCNN network | |
| CN113077452B (en) | Apple tree pest and disease detection method based on DNN network and spot detection algorithm | |
| CN112381662B (en) | Pollen fertility rate assessment method and device | |
| CN117253192A (en) | Intelligent system and method for silkworm breeding | |
| CN117372881A (en) | Intelligent identification method, medium and system for tobacco plant diseases and insect pests | |
| CN107368847A (en) | A kind of crop leaf diseases recognition methods and system | |
| CN107133634B (en) | Method and device for acquiring plant water shortage degree | |
| Feng et al. | Research of image recognition of plant diseases and pests based on deep learning | |
| CN112613521B (en) | Multilevel data analysis system and method based on data conversion | |
| CN113627329A (en) | Wheat seed hyperspectral image classification method and system based on hybrid convolutional network |
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 |