CN111369494B - Winter wheat spike density detection method and device - Google Patents
Winter wheat spike density detection method and device Download PDFInfo
- Publication number
- CN111369494B CN111369494B CN202010082618.2A CN202010082618A CN111369494B CN 111369494 B CN111369494 B CN 111369494B CN 202010082618 A CN202010082618 A CN 202010082618A CN 111369494 B CN111369494 B CN 111369494B
- Authority
- CN
- China
- Prior art keywords
- density
- winter wheat
- spike
- neural network
- convolutional neural
- 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
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
-
- 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
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- 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)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the invention provides a method and a device for detecting the density of winter wheat ears, wherein the method comprises the following steps: acquiring a canopy image to be detected after the flowering period of winter wheat, and dividing the canopy image to be detected into a plurality of subgraphs to be detected; inputting the plurality of subgraphs to be detected into a convolutional neural network model, and outputting a subgraph density; determining winter wheat spike density according to the sub-density map; the convolutional neural network model is obtained after training based on a sample canopy image and a spike density map corresponding to the predetermined sample canopy image. The winter wheat spike density detection method and device provided by the embodiment of the invention utilize the artificial intelligence technology to detect, have high automation degree, can effectively reduce the manual intervention of spike density estimation, reduce the application cost and complexity, and effectively improve the accuracy and the instantaneity of winter wheat spike density estimation.
Description
Technical Field
The embodiment of the invention relates to the technical fields of system engineering and information, in particular to a method and a device for detecting winter wheat spike density.
Background
The winter wheat spike density is an important index for winter wheat yield estimation and phenotype analysis, and has important practical significance. The traditional spike density calculation method mainly depends on manual counting, and the method not only needs to consume a large amount of manpower and material resources, but also has lower efficiency and can not meet the requirements of large-area yield estimation and phenotype analysis. The spike density estimation method based on computer vision has the advantages of low cost and easiness in use, and is the most main spike density estimation means at present. The spike density estimation method based on computer vision mainly comprises the following steps: cutting wheat ears, performing binary pattern morphology optimization, and calculating the number of connected domains, so as to obtain the number of wheat ears in the image. Ear image segmentation is the core content of an ear density estimation method based on computer vision, and is usually performed on pixel-level or super-pixel-level ear extraction based on bottom features such as color and texture of ears.
Because winter wheat canopy images acquired in a field environment are easy to be interfered by illumination conditions and complex backgrounds, the accuracy of wheat ear image segmentation based on the bottom features is easy to be interfered by noise, the generalization capability is low, and good effects are difficult to obtain in practical application. In addition, the spike density in the winter wheat canopy image acquired in the field environment is higher, so that the mutual shielding of the wheat spikes is relatively serious, the size and shape of the wheat spikes are greatly changed, and the spike density estimation is challenged. How to overcome the influence of illumination conditions and complex backgrounds, the problems that the current spike density estimation method is easy to be interfered by noise, the spike overlap and the spike appearance change are large and the like are solved, and the accurate estimation of the spike density is realized, so that the method is a problem to be solved urgently.
Disclosure of Invention
The embodiment of the invention provides a method and a device for detecting the density of winter wheat ears, which are used for solving the technical problems in the prior art.
In order to solve the above technical problems, in one aspect, an embodiment of the present invention provides a method for detecting a winter wheat ear density, including:
acquiring a canopy image to be detected after the flowering period of winter wheat, and dividing the canopy image to be detected into a plurality of subgraphs to be detected;
inputting the plurality of subgraphs to be detected into a convolutional neural network model, and outputting a subgraph density;
determining winter wheat spike density according to the sub-density map;
the convolutional neural network model is obtained after training based on a sample canopy image and a spike density map corresponding to the predetermined sample canopy image.
Further, the determining the winter wheat ear density according to the sub-density map specifically includes:
calculating a sub-density map position index according to the number of pooling layers in the convolutional neural network model;
fusing the output sub-density maps according to the sub-density map position indexes to obtain a spike density map;
and determining the winter wheat spike density according to the spike density map.
Further, the training process of the convolutional neural network model comprises the following steps:
acquiring a sample canopy image of winter wheat after the flowering period;
recording the position information of the wheat ears by using point marks;
generating a spike density map of the sample according to the position information of the wheat spike;
dividing a sample canopy image and a spike density image corresponding to the sample canopy image into a plurality of sample subgraphs and corresponding subdensity images by adopting a sliding window;
and training the convolutional neural network by using the plurality of sample subgraphs and the corresponding subgraphs to obtain the convolutional neural network model.
Further, the position information of the wheat ears is coordinates of a center point of the wheat ears in the image.
Further, the generating the spike density map of the sample according to the position information of the wheat spike specifically includes:
and generating a spike density map of the sample by adopting a geometric self-adaption method according to the position information of the wheat spike.
Further, training the convolutional neural network by using the plurality of sample subgraphs and the corresponding subdensity graphs to obtain the convolutional neural network model, which specifically comprises:
taking the plurality of sample subgraphs and the corresponding subgraphs as input layers of a convolutional neural network model;
constructing a feature extractor of a convolutional neural network model;
and sequentially connecting the input layer, the feature extractor, an overlapping layer for fusing output results of the feature extractor and an output layer for fusing a plurality of channels in the overlapping layer, so as to complete the establishment of the convolutional neural network model.
Further, the feature extractor comprises four parallel processing modules, and each processing module comprises three processing units which are sequentially connected.
Further, the first processing unit of each processing module comprises a convolution layer, a modified linear unit ReLU layer and a pooling layer which are sequentially connected; the second processing unit comprises at least 3 convolution layers and modified linear unit ReLU layers which are sequentially connected, only the first modified linear unit ReLU layer is connected with a pooling layer, and the third processing unit comprises at least 3 convolution layers and modified linear unit ReLU layers which are sequentially connected;
the size of the convolution kernels in each convolution layer in each processing unit is kept unchanged, the number of the convolution kernels in each convolution layer in the second processing unit is sequentially increased by taking 2 as a multiplier, the number of the convolution kernels in each convolution layer in the third processing unit is sequentially decreased by taking 0.5 as a multiplier, the number of the convolution kernels in the second processing unit is equal to the number of the convolution kernels in the third processing unit, the size of the convolution kernels in the first processing unit is larger than the size of the convolution kernels in each convolution layer in the second processing unit, and the size of the convolution kernels in each convolution layer in the second processing unit is larger than the size of the convolution kernels in each convolution layer in the third processing unit.
Further, the size of the convolution kernel in the convolution layer in the first processing unit of each processing module decreases in sequence.
On the other hand, the embodiment of the invention provides a winter wheat spike density detection device, which is characterized by comprising:
the acquisition module is used for acquiring a canopy image to be detected after the winter wheat blooming period and dividing the canopy image to be detected into a plurality of sub-images to be detected;
the output module is used for inputting the plurality of sub-graphs to be detected into the convolutional neural network model and outputting a sub-density graph;
the detection module is used for determining the winter wheat spike density according to the sub-density map;
the convolutional neural network model is obtained after training based on a sample canopy image and a spike density map corresponding to the predetermined sample canopy image.
The winter wheat spike density detection method and device provided by the embodiment of the invention utilize the artificial intelligence technology to detect, have high automation degree, can effectively reduce the manual intervention of spike density estimation, reduce the application cost and complexity, and effectively improve the accuracy and the instantaneity of winter wheat spike density estimation.
Drawings
FIG. 1 is a schematic diagram of a method for detecting the density of winter wheat ears according to an embodiment of the present invention;
FIG. 2 is a logic flow diagram for detecting the density of winter wheat ears according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a winter wheat spike density detecting device according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make 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 schematic diagram of a method for detecting winter wheat spike density according to an embodiment of the present invention, and as shown in fig. 1, an embodiment of the present invention provides a method for detecting winter wheat spike density, where an execution main body is a device for detecting winter wheat spike density. The method comprises the following steps:
step S101, acquiring a canopy image to be detected after the winter wheat blooming period, and dividing the canopy image to be detected into a plurality of sub-images to be detected.
Specifically, fig. 2 is a logic flow chart for detecting the density of winter wheat ears, as shown in fig. 2, firstly, a canopy image to be detected after a blooming period of winter wheat is obtained, and the canopy image to be detected is divided into a plurality of sub-images to be detected.
For example, 2500×2500×3 pixels can be used to collect the canopy image to be detected, and the canopy image to be detected is divided into several sub-images to be detected by a sliding window with a size of 600×600 pixels and a 300-pixel step.
Dividing the canopy image to be detected into a plurality of sub-images to be detected, and recording the position index of the sub-images to be detected.
The position index of the sub-graph to be detected comprises: the coordinates of the pixel point at the upper left corner of the sub-graph to be detected in the image to be detected can be extracted by using a first formula:
wherein, the first formula is:
Index (i,j) ＝I (i,j)
Index (i,j) index of jth sub-picture of ith image to be detected, I (i,j) The coordinates of the pixel point at the upper left corner of the jth sub-graph of the ith image to be detected in the image to be detected.
Step S102, inputting the plurality of sub-graphs to be detected into a convolutional neural network model, and outputting a sub-density graph.
Specifically, after a plurality of sub-graphs to be detected are obtained, the plurality of sub-graphs to be detected are input into a convolutional neural network model, and a sub-density graph is output.
The convolutional neural network model is obtained after training based on a sample canopy image and a spike density map corresponding to the predetermined sample canopy image.
And step S103, determining the winter wheat spike density according to the sub-density map.
Specifically, after the sub-density map is obtained, the winter wheat head density is determined from the sub-density map.
The winter wheat spike density detection method provided by the embodiment of the invention utilizes the artificial intelligence technology to detect, has high automation degree, can effectively reduce the manual intervention of spike density estimation, reduces the application cost and complexity, and effectively improves the accuracy and the real-time performance of winter wheat spike density estimation.
Based on any of the above embodiments, further, the determining the winter wheat ear density according to the sub-density map specifically includes:
calculating a sub-density map position index according to the number of pooling layers in the convolutional neural network model;
fusing the output sub-density maps according to the sub-density map position indexes to obtain a spike density map;
and determining the winter wheat spike density according to the spike density map.
Specifically, in the embodiment of the invention, the winter wheat spike density is determined according to the sub-density map, and the method specifically comprises the following steps of:
first, a sub-density map location index is calculated based on the number of pooling layers in the convolutional neural network model.
Calculating the position index of the sub-density map to be detected, firstly extracting a correction factor by using a second formula, and then extracting the position index of the sub-density map to be detected by using a third formula;
wherein the second formula is:
f is a correction factor, and N is the number of pooling layers in the convolutional neural network.
The third formula is:
Index_m (i,j) ＝f*Index (i,j)
Index (i,j) index_m for the j-th sub-picture of the i-th image to be detected (i,j) Index of the jth sub-density map for the ith image to be detected.
And then, fusing the output sub-density maps according to the sub-density map position index to obtain an ear density map.
Finally, determining the winter wheat spike density according to the spike density map.
The winter wheat spike density detection method provided by the embodiment of the invention utilizes the artificial intelligence technology to detect, has high automation degree, can effectively reduce the manual intervention of spike density estimation, reduces the application cost and complexity, and effectively improves the accuracy and the real-time performance of winter wheat spike density estimation.
Based on any one of the above embodiments, further, the training process of the convolutional neural network model includes the following steps:
acquiring a sample canopy image of winter wheat after the flowering period;
recording the position information of the wheat ears by using point marks;
generating a spike density map of the sample according to the position information of the wheat spike;
dividing a sample canopy image and a spike density image corresponding to the sample canopy image into a plurality of sample subgraphs and corresponding subdensity images by adopting a sliding window;
and training the convolutional neural network by using the plurality of sample subgraphs and the corresponding subgraphs to obtain the convolutional neural network model.
Specifically, firstly, a canopy image after the flowering period of winter wheat collected in a field environment is obtained as a sample.
After the sample canopy image is acquired, a pre-process may also be performed, where the pre-process is to adjust the size of the sample canopy image, for example, to 2500×2500×3 pixels.
Then, the position information of the wheat ears is recorded by using the dot mark.
In a specific application, preferably, the position information of the wheat ears includes: coordinates of the center point of the wheat ear in the image.
Then, a sample spike density map is generated from the position information of the wheat spike.
The spike density map of the sample is generated by adopting a geometric self-adaption method.
Then, the sample canopy image and the corresponding spike density map thereof are divided into a plurality of sample subgraphs and corresponding subgraphs by adopting a sliding window.
A sliding window sample image of 600 x 600 pixel size, 300 pixel step size, and its corresponding ear density map may be divided into sub-images and corresponding sub-density maps.
And finally, training the convolutional neural network by utilizing the plurality of sample subgraphs and the corresponding subgraphs to obtain the convolutional neural network model.
In a particular application, the convolutional neural network structure preferably comprises 1 input layer, 1 feature extractor, 1 superimposed layer, and 1 output layer. The feature extractor comprises 4 parallel processing modules, and each processing module comprises 3 processing units which are connected in sequence.
In a specific application, preferably, the 1 st processing unit of the 1 st processing module comprises 1 convolution layer, 1 modified linear unit ReLU layer and 1 pooling layer. Wherein, 8 convolution kernels of 11×11 are adopted in the convolution layer, and a maximum pooling function of 2×2 is adopted in the pooling layer; the 2 nd processing unit includes 3 convolutional layers, 3 modified linear unit ReLU layers and 1 pooling layer. Wherein, the convolution layers adopt 9×9 convolution kernels, the number is 16, 32 and 64 respectively, and the pooling layer is positioned at the back of the first modified linear unit ReLU layer and adopts a maximum pooling function of 2×2; the 3 rd processing unit includes 3 convolutional layers. Wherein, the convolution layers adopt 1×1 convolution kernels, and the numbers are respectively 32, 16 and 8.
In a specific application, preferably, the 1 st processing unit of the 2 nd processing module comprises 1 convolution layer, 1 modified linear unit ReLU layer and 1 pooling layer. Wherein, 16 convolution kernels of 9×9 are adopted in the convolution layer, and a maximum pooling function of 2×2 is adopted in the pooling layer; the 2 nd processing unit includes 3 convolutional layers, 3 modified linear unit ReLU layers and 1 pooling layer. Wherein, the convolution layers adopt 7×7 convolution kernels, the number is 32, 64 and 128 respectively, and the pooling layer is positioned at the back of the first modified linear unit ReLU layer and adopts a maximum pooling function of 2×2; the 3 rd processing unit includes 3 convolutional layers. Wherein, the convolution layers adopt 1×1 convolution kernels, and the numbers are 64, 32 and 16 respectively.
In a specific application, preferably, the 1 st processing unit of the 3 rd processing module comprises 1 convolution layer, 1 modified linear unit ReLU layer and 1 pooling layer. Wherein, the convolution layer adopts 32 convolution kernels of 7×7, and the pooling layer adopts a maximum pooling function of 2×2; the 2 nd processing unit includes 3 convolutional layers, 3 modified linear unit ReLU layers and 1 pooling layer. Wherein, the convolution layers all adopt convolution kernels of 5×5, the number is 64, 128 and 256 respectively, and the pooling layer is positioned at the back of the first modified linear unit ReLU layer and adopts a maximum pooling function of 2×2; the 3 rd processing unit includes 3 convolutional layers. Wherein, the convolution layers adopt 1×1 convolution kernels, and the numbers are 128, 64 and 32 respectively.
In a specific application, preferably, the 1 st processing unit of the 4 th processing module comprises 1 convolution layer, 1 modified linear unit ReLU layer and 1 pooling layer. Wherein, the convolution layer adopts 64 convolution kernels of 5×5, and the pooling layer adopts a maximum pooling function of 2×2; the 2 nd processing unit includes 3 convolutional layers, 3 modified linear unit ReLU layers and 1 pooling layer. Wherein, the convolution layers all adopt 3×3 convolution kernels, the number is 128, 256 and 512, and the pooling layer is positioned at the back of the first modified linear unit ReLU layer and adopts a maximum pooling function of 2×2; the 3 rd processing unit includes 3 convolutional layers. Wherein, the convolution layers adopt 1×1 convolution kernels, and the numbers are 256, 128 and 64 respectively.
In a specific application, preferably, the overlay layer overlays the feature map of the 4 processing module results; the output layer comprises 1 convolution layer, and the characteristic images output by the superimposed layers are fused by adopting 1 convolution kernel of 1 multiplied by 1, so that the spike density image is output.
According to the winter wheat spike density detection method provided by the embodiment of the invention, the canopy image acquired in a field environment after the blooming period of winter wheat is preprocessed, the preprocessed image is divided into a sample set and a set to be detected, and the position information of the wheat spike is recorded by adopting point marks; generating a spike density map of the sample set image according to the wheat spike position information; dividing a winter wheat canopy image and a spike density image corresponding to the winter wheat canopy image into a subgraph and a subdensity image corresponding to the subgraph by adopting a sliding window, dividing an image of a set to be detected into subgraphs, and recording the position index of the subgraph of the set to be detected; taking a sample set subgraph and corresponding subdensity chart data as the input of a model, constructing a convolutional neural network model for estimating the spike density, and carrying out model training and verification by adopting an adaptive moment estimation algorithm; testing by adopting a to-be-detected sub-graph to obtain a to-be-detected sub-graph; calculating the position index of the density map of the set to be detected according to the number of pooling layers in the convolutional neural network model; and fusing the generated to-be-detected set density map according to the to-be-detected set density map position index to obtain a to-be-detected set image spike density map, and further obtaining the corresponding spike density. The artificial intervention of spike density estimation can be effectively reduced, the application cost and the complexity are reduced, and the accuracy and the instantaneity of winter wheat spike density estimation are effectively improved. The invention can greatly contribute to winter wheat spike density estimation on the basis of combining with counting such as image processing, deep learning and the like.
Based on any of the above embodiments, fig. 3 is a schematic diagram of a winter wheat spike density detection apparatus according to an embodiment of the present invention, and as shown in fig. 3, the embodiment of the present invention provides a winter wheat spike density detection apparatus, which includes an obtaining module 301, an output module 302, and a detection module 303, wherein:
the acquisition module 301 is configured to acquire a canopy image to be detected after a flowering period of winter wheat, and divide the canopy image to be detected into a plurality of sub-images to be detected; the output module 302 is configured to input the plurality of sub-graphs to be detected to a convolutional neural network model, and output a sub-density map; the detection module 303 is used for determining the winter wheat spike density according to the sub-density map; the convolutional neural network model is obtained after training based on a sample canopy image and a spike density map corresponding to the predetermined sample canopy image.
An embodiment of the present invention provides a winter wheat spike density detection apparatus, which is configured to perform the method described in any one of the embodiments, and specific steps of performing the method described in the one embodiment by using the apparatus provided in the present embodiment are the same as those of the corresponding embodiment, and are not repeated herein.
The winter wheat spike density detection device provided by the embodiment of the invention utilizes the artificial intelligence technology to detect, has high automation degree, can effectively reduce the manual intervention of spike density estimation, reduces the application cost and complexity, and effectively improves the accuracy and the instantaneity of winter wheat spike density estimation.
Fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention, as shown in fig. 4, where the electronic device includes: a processor (processor) 401, a communication interface (Communications Interface) 402, a memory (memory) 403 and a communication bus 404, wherein the processor 401, the communication interface 402 and the memory 403 complete communication with each other through the communication bus 404. Processor 401 and memory 402 communicate with each other via bus 403. The processor 401 may call logic instructions in the memory 403 to perform the following method:
acquiring a canopy image to be detected after the flowering period of winter wheat, and dividing the canopy image to be detected into a plurality of subgraphs to be detected;
inputting the plurality of subgraphs to be detected into a convolutional neural network model, and outputting a subgraph density;
determining winter wheat spike density according to the sub-density map;
the convolutional neural network model is obtained after training based on a sample canopy image and a spike density map corresponding to the predetermined sample canopy image.
Further, the logic instructions in the memory described above 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 U-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.
Further, embodiments of the present invention 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, are capable of performing the steps of the method embodiments described above, for example comprising:
acquiring a canopy image to be detected after the flowering period of winter wheat, and dividing the canopy image to be detected into a plurality of subgraphs to be detected;
inputting the plurality of subgraphs to be detected into a convolutional neural network model, and outputting a subgraph density;
determining winter wheat spike density according to the sub-density map;
the convolutional neural network model is obtained after training based on a sample canopy image and a spike density map corresponding to the predetermined sample canopy image.
Further, embodiments of the present invention provide a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the method embodiments described above, for example, including:
acquiring a canopy image to be detected after the flowering period of winter wheat, and dividing the canopy image to be detected into a plurality of subgraphs to be detected;
inputting the plurality of subgraphs to be detected into a convolutional neural network model, and outputting a subgraph density;
determining winter wheat spike density according to the sub-density map;
the convolutional neural network model is obtained after training based on a sample canopy image and a spike density map corresponding to the predetermined sample canopy image.
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 method for detecting the density of winter wheat ears, which is characterized by comprising the following steps:
acquiring a canopy image to be detected after the flowering period of winter wheat, and dividing the canopy image to be detected into a plurality of subgraphs to be detected;
inputting the plurality of subgraphs to be detected into a convolutional neural network model, and outputting a subgraph density;
determining winter wheat spike density according to the sub-density map;
the convolutional neural network model is obtained after training based on a sample canopy image and a spike density map corresponding to the predetermined sample canopy image;
the determining the winter wheat spike density according to the sub-density map specifically comprises the following steps:
calculating a sub-density map position index according to the number of pooling layers in the convolutional neural network model;
fusing the output sub-density maps according to the sub-density map position indexes to obtain a spike density map;
and determining the winter wheat spike density according to the spike density map.
2. The method for detecting the density of winter wheat ears according to claim 1, wherein the training process of the convolutional neural network model comprises the following steps:
acquiring a sample canopy image of winter wheat after the flowering period;
recording the position information of the wheat ears by using point marks;
generating a spike density map of the sample according to the position information of the wheat spike;
dividing a sample canopy image and a spike density image corresponding to the sample canopy image into a plurality of sample subgraphs and corresponding subdensity images by adopting a sliding window;
and training the convolutional neural network by using the plurality of sample subgraphs and the corresponding subgraphs to obtain the convolutional neural network model.
3. The method of claim 2, wherein the position information of the wheat head is coordinates of a center point of the wheat head in an image.
4. The method for detecting winter wheat head density according to claim 2, wherein the generating the sample head density map according to the position information of the wheat head specifically comprises:
and generating a spike density map of the sample by adopting a geometric self-adaption method according to the position information of the wheat spike.
5. The method for detecting winter wheat head density according to claim 2, wherein the training the convolutional neural network by using the plurality of sample subgraphs and the corresponding subgraphs to obtain the convolutional neural network model specifically comprises:
taking the plurality of sample subgraphs and the corresponding subgraphs as input layers of a convolutional neural network model;
constructing a feature extractor of a convolutional neural network model;
and sequentially connecting the input layer, the feature extractor, an overlapping layer for fusing output results of the feature extractor and an output layer for fusing a plurality of channels in the overlapping layer, so as to complete the establishment of the convolutional neural network model.
6. The method for detecting the density of winter wheat ears according to claim 5, wherein the feature extractor comprises four parallel processing modules, and each processing module comprises three processing units connected in sequence.
7. The method of claim 6, wherein the first processing unit of each processing module comprises a convolution layer, a modified linear unit ReLU layer and a pooling layer connected in sequence; the second processing unit comprises at least 3 convolution layers and modified linear unit ReLU layers which are sequentially connected, only the first modified linear unit ReLU layer is connected with a pooling layer, and the third processing unit comprises at least 3 convolution layers and modified linear unit ReLU layers which are sequentially connected;
the size of the convolution kernels in each convolution layer in each processing unit is kept unchanged, the number of the convolution kernels in each convolution layer in the second processing unit is sequentially increased by taking 2 as a multiplier, the number of the convolution kernels in each convolution layer in the third processing unit is sequentially decreased by taking 0.5 as a multiplier, the number of the convolution kernels in the second processing unit is equal to the number of the convolution kernels in the third processing unit, the size of the convolution kernels in the first processing unit is larger than the size of the convolution kernels in each convolution layer in the second processing unit, and the size of the convolution kernels in each convolution layer in the second processing unit is larger than the size of the convolution kernels in each convolution layer in the third processing unit.
8. The method of claim 7, wherein the size of the convolution kernel in the convolution layer in the first processing unit of each processing module decreases in sequence.
9. A winter wheat head density detection device, comprising:
the acquisition module is used for acquiring a canopy image to be detected after the winter wheat blooming period and dividing the canopy image to be detected into a plurality of sub-images to be detected;
the output module is used for inputting the plurality of sub-graphs to be detected into the convolutional neural network model and outputting a sub-density graph;
the detection module is used for determining the winter wheat spike density according to the sub-density map;
the convolutional neural network model is obtained after training based on a sample canopy image and a spike density map corresponding to the predetermined sample canopy image;
the detection module is specifically used for:
calculating a sub-density map position index according to the number of pooling layers in the convolutional neural network model;
fusing the output sub-density maps according to the sub-density map position indexes to obtain a spike density map;
and determining the winter wheat spike density according to the spike density map.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010082618.2A CN111369494B (en) | 2020-02-07 | 2020-02-07 | Winter wheat spike density detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010082618.2A CN111369494B (en) | 2020-02-07 | 2020-02-07 | Winter wheat spike density detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111369494A CN111369494A (en) | 2020-07-03 |
| CN111369494B true CN111369494B (en) | 2023-05-02 |
Family
ID=71211420
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010082618.2A Active CN111369494B (en) | 2020-02-07 | 2020-02-07 | Winter wheat spike density detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111369494B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112115988B (en) * | 2020-09-03 | 2024-02-02 | 中国农业大学 | Wheat ear counting method and device and self-walking trolley |
| CN112017986A (en) * | 2020-10-21 | 2020-12-01 | 季华实验室 | Semiconductor product defect detection method and device, electronic equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105842164A (en) * | 2016-05-06 | 2016-08-10 | 北京农业信息技术研究中心 | Wheat spike number measuring device and use method thereof |
| WO2018234733A1 (en) * | 2017-06-19 | 2018-12-27 | Earlham Institute | Data processing of images of a crop |
| CN109145848A (en) * | 2018-08-30 | 2019-01-04 | 西京学院 | A kind of wheat head method of counting |
| CN109740721A (en) * | 2018-12-19 | 2019-05-10 | 中国农业大学 | Wheat head method of counting and device |
-
2020
- 2020-02-07 CN CN202010082618.2A patent/CN111369494B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105842164A (en) * | 2016-05-06 | 2016-08-10 | 北京农业信息技术研究中心 | Wheat spike number measuring device and use method thereof |
| WO2018234733A1 (en) * | 2017-06-19 | 2018-12-27 | Earlham Institute | Data processing of images of a crop |
| CN109145848A (en) * | 2018-08-30 | 2019-01-04 | 西京学院 | A kind of wheat head method of counting |
| CN109740721A (en) * | 2018-12-19 | 2019-05-10 | 中国农业大学 | Wheat head method of counting and device |
Non-Patent Citations (3)
| Title |
|---|
| Daoyong Wang et.al.Combined Use of FCN and Harris Corner Detection for Counting Wheat Ears in Field Conditions.《IEEE Access》.2019,第7卷第178930-178941页. * |
| Juncheng Ma et.al.Segmenting ears of winter wheat at flowering stage using digital images and deep learning.《Computers and Electronics in Agriculture》.2019,第1-16页. * |
| 张领先等.基于卷积神经网络的冬小麦麦穗检测计数系统.《农业机械学报》.2019,第50卷(第50期),第144-150页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111369494A (en) | 2020-07-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | Saliency detection based on self-adaptive multiple feature fusion for remote sensing images | |
| CN112287940A (en) | Semantic segmentation method of attention mechanism based on deep learning | |
| CN112418216B (en) | Text detection method in complex natural scene image | |
| CN112861690B (en) | Multi-method fused remote sensing image change detection method and system | |
| CN109978037A (en) | Image processing method, model training method, device and storage medium | |
| CN113240691A (en) | Medical image segmentation method based on U-shaped network | |
| CN112837344B (en) | Target tracking method for generating twin network based on condition countermeasure | |
| CN109377441B (en) | Tongue image acquisition method and system with privacy protection function | |
| CN107871316B (en) | Automatic X-ray film hand bone interest area extraction method based on deep neural network | |
| CN111369494B (en) | Winter wheat spike density detection method and device | |
| CN110363072B (en) | Tongue picture identification method, tongue picture identification device, computer equipment and computer readable storage medium | |
| CN111091101B (en) | High-precision pedestrian detection method, system and device based on one-step method | |
| CN110222760A (en) | A kind of fast image processing method based on winograd algorithm | |
| CN108830856B (en) | GA automatic segmentation method based on time series SD-OCT retina image | |
| WO2023151237A1 (en) | Face pose estimation method and apparatus, electronic device, and storage medium | |
| CN112037146A (en) | Medical image artifact automatic correction method and device and computer equipment | |
| CN110807775A (en) | Traditional Chinese medicine tongue image segmentation device and method based on artificial intelligence and storage medium | |
| CN112766123B (en) | Crowd counting method and system based on criss-cross attention network | |
| CN112800955A (en) | Remote sensing image rotating target detection method and system based on weighted bidirectional feature pyramid | |
| CN111862040B (en) | Portrait picture quality evaluation method, device, equipment and storage medium | |
| CN114399480A (en) | Method and device for detecting severity of vegetable leaf disease | |
| Yu et al. | Segmentation and density statistics of mariculture cages from remote sensing images using mask R-CNN | |
| CN115482523A (en) | Small object target detection method and system of lightweight multi-scale attention mechanism | |
| CN108597589B (en) | Model generation method, target detection method and medical imaging system | |
| CN110007764A (en) | A kind of gesture skeleton recognition methods, device, system and storage medium |
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 |