CN113313692A - Automatic banana young plant identification and counting method based on aerial visible light image - Google Patents
Automatic banana young plant identification and counting method based on aerial visible light image Download PDFInfo
- Publication number
- CN113313692A CN113313692A CN202110619392.XA CN202110619392A CN113313692A CN 113313692 A CN113313692 A CN 113313692A CN 202110619392 A CN202110619392 A CN 202110619392A CN 113313692 A CN113313692 A CN 113313692A
- Authority
- CN
- China
- Prior art keywords
- image
- sub
- banana
- images
- visible light
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 241000196324 Embryophyta Species 0.000 title claims abstract description 65
- 235000018290 Musa x paradisiaca Nutrition 0.000 title claims abstract description 64
- 238000000034 method Methods 0.000 title claims abstract description 28
- 240000005561 Musa balbisiana Species 0.000 title 1
- 241000234295 Musa Species 0.000 claims abstract description 64
- 238000002372 labelling Methods 0.000 claims abstract description 17
- 238000005520 cutting process Methods 0.000 claims abstract description 15
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims abstract description 6
- 238000001514 detection method Methods 0.000 claims description 9
- 238000012216 screening Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 240000000905 Nymphoides indica Species 0.000 description 9
- 235000017590 Nymphoides indica Nutrition 0.000 description 9
- 230000000694 effects Effects 0.000 description 6
- 244000157072 Hylocereus undatus Species 0.000 description 5
- 235000018481 Hylocereus undatus Nutrition 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 201000010099 disease Diseases 0.000 description 4
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 241000252254 Catostomidae Species 0.000 description 2
- 235000012055 fruits and vegetables Nutrition 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 235000013311 vegetables Nutrition 0.000 description 2
- 244000241257 Cucumis melo Species 0.000 description 1
- 235000015510 Cucumis melo subsp melo Nutrition 0.000 description 1
- 235000021015 bananas Nutrition 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000003337 fertilizer Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
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
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4038—Image mosaicing, e.g. composing plane images from plane sub-images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
-
- 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
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/68—Analysis of geometric attributes of symmetry
-
- 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
- G06T2207/20004—Adaptive image processing
-
- 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
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30242—Counting objects in image
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Geometry (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an automatic identification and counting method of banana young plants based on aerial visible light images, which comprises the following steps of S1, collecting the visible light images of the banana young plants as original images; s2, cutting the original image without overlapping degree; s3, preprocessing the image of the sub-image, and labeling the banana young plant; s4, constructing a fast-RCNN model, and training the fast-RCNN model based on the banana young plant labeled subgraph; s5, cutting the image to be detected into a sub-image, inputting the sub-image into a trained Faster-RCNN model, and obtaining a sub-image recognition image; and S6, splicing the sub-image recognition images into a complete image again, and performing de-duplication treatment on the boundary banana young plants to finish counting. The method has the advantages of high robustness under the variation of variables such as height, light conditions and the like, high efficiency, great reduction of manpower input and strong application and popularization potential.
Description
Technical Field
The invention relates to the field of image identification and counting of banana young plants, in particular to an automatic identification and counting method of banana young plants based on aerial visible light images.
Background
Because the banana plants are perennial herbaceous plants, the banana plants grow rapidly, the number of the banana plants in the next year is controlled by selecting the suckers, and the suckers grow more in the next year. In addition, most bananas are planted in subtropical and tropical regions, the weather is hot and humid, the disease is easy to propagate and spread, several banana diseases (especially blight) cause serious yield loss in the whole production field, and the diseased plants can only be removed in time and eliminated to slow down the spread of the disease because the diseased plants cannot be effectively cured. The method can be used for timely and rapidly investigating the number of plants in the banana garden, and has guiding significance for disease control of the banana garden and the aspects of water, fertilizer and medicine investment, yield prediction and the like of managers. The deep learning detection frame based on the fast RCNN has the advantages of high speed and high precision, and is widely applied to identification and counting of fruit and vegetable crops in recent years.
The fast-RCNN technology has been widely applied to the identification and application industries of melons, fruits and vegetables, for example, Nisar et al propose a dragon fruit counting and yield prediction method, which respectively uses an RGB model (R-G channel) and a YCbCr model (Cr channel) to segment the acquired dragon fruit image, then uses morphological processing to analyze the size threshold and the shape of the image, and judges whether the dragon fruit in the image is single or 2 adhered dragon fruits according to the range of roundness rate, thereby realizing the automatic counting of the dragon fruit. In the aspect of plant identification, plum vibration waves and the like are used for detecting hydroponic vegetable seedlings by using an improved Faster-RCNN detection model, aiming at the characteristics that hydroponic vegetable seedling images are tiny and dense, HRNet is used as a feature extraction network, information loss in the down-sampling process is reduced, and information of small target objects is well kept, so that good semantic information is provided for the regression and classification of subsequent candidate frames. However, for the identification of the target crop seedlings, the used images have the characteristics of short shooting distance and small target number, and the target crops cannot be counted in a large area. Therefore, the invention provides an automatic identification and counting method of banana young plants based on aerial visible light images, and aims to overcome the defects in the prior art.
Disclosure of Invention
Aiming at the problems, the invention provides an automatic identification and counting method of banana young plants based on aerial visible light images, which aims to solve the technical problems in the prior art, can cut original images into a plurality of sub-images without overlapping, and can count large-area target crops by combining with deduplication processing.
In order to achieve the purpose, the invention provides the following scheme: the invention provides an automatic identification and counting method of banana young plants based on aerial visible light images, which comprises the following steps:
s1, collecting a visible light image of a banana young plant as an original image;
s2, cutting the original image without overlapping degree to obtain a plurality of sub-images;
s3, carrying out image preprocessing on the subgraph, and labeling the banana young plant to obtain a labeled subgraph of the banana young plant;
s4, constructing a fast-RCNN model, and training the fast-RCNN model based on the banana young plant labeled subgraph;
s5, cutting the image to be detected into a sub-image, inputting the sub-image into a trained Faster-RCNN model, and obtaining a sub-image recognition image;
and S6, splicing the sub-image recognition images into a complete image again, and performing de-duplication treatment on the boundary banana young plants to finish counting.
Preferably, the specific steps of S2 non-overlapping image cropping are as follows:
s2.1, creating a plurality of uniform grid vertex coordinates according to the size of the original image;
s2.2, creating a five-dimensional tensor, wherein the five-dimensional tensor is used for storing the cut sub-images, the front two dimensions in the five-dimensional tensor represent the coordinate positions of the sub-images in the original image, and the rear three dimensions represent the information of each sub-image;
and S2.3, cutting the original image, and storing sub-image data determined by the coordinate values of the upper left corner and the lower right corner of each grid into the five-dimensional tensor.
Preferably, the S3 preprocessing applies a contrast-limited CLAHE algorithm to equalize the luminance values of the images in HSV color space.
Preferably, the specific calculation step of the deduplication processing in S6 is:
s6.1, calculating the coordinates of the central point of the sub-graph;
s6.2, traversing and calculating the distance between the box center point marked by each banana young plant and the sub-image edge, setting a first preset distance threshold and a second preset distance threshold, and screening out the sub-image edge marking box based on the first preset distance threshold and the second preset distance threshold;
s6.3, setting an area threshold value by calculating the area of the labeling square frame at the edge of the subgraph, and screening out the labeling square frames of incomplete banana young plants based on the area threshold value;
s6.4, calculating the distance between the central points of two adjacent boxes in the marked boxes of the incomplete banana young plants, and identifying the repeated detection condition through a third preset distance threshold value.
Preferably, the first preset distance threshold and the second preset distance threshold are both variable thresholds.
Preferably, the setting of the variable threshold is specifically performed according to the original visible light image of the aerial photography height.
Preferably, the variable threshold is calculated as:
wherein, threshold0 is the distance of the current aerial photography height, threshold1 is the distance threshold of aerial photography image of banana young plants to be counted, Δ d0 is the distance of every two banana young plants in the sub-image of the current aerial photography image, and Δ d1 is the distance of every two banana young plants in the sub-image.
The invention discloses the following technical effects:
according to the method, the images of banana young plants at different light conditions and different shooting heights are obtained on the spot by the unmanned aerial vehicle to construct the original images, aiming at the problem that the number of candidate images generated by the Faster-RCNN framework on the input image is limited, the target crops of the target images are reduced by using a processing method of image non-overlapping degree cutting, and then the large-area detection of the number of banana young plants is realized by adopting a mode of splicing back to a large image. In the subgraph identification result, the boundary generated by the subgraph during interception divides a large number of banana plants into two parts, even one part into four parts. Some trans-border banana plants are respectively detected in adjacent subgraphs (repeated boundary identification), so that the statistical quantity is increased, and a large error is occupied. The counting method for automatically identifying the banana young plants, which is provided by the target detection model based on the Faster-RCNN, has high robustness under the variation of variables such as height, light conditions and the like, is efficient and portable, greatly reduces the manpower input, can provide related technical support for users in the aspects of farm management, yield prediction and the like, and has strong application and popularization potentials.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is an original image of an 80m aerial photograph taken by an unmanned aerial vehicle in an embodiment of the present invention;
fig. 3 is a comparison diagram of enhancement effects of equalizing luminance values of sub-images in HSV color space by using a CLAHE algorithm in the embodiment of the present invention, where fig. 3(a) is a sub-image original image, and fig. 3(b) is an image after sub-image enhancement;
FIG. 4 is a comparison of labeling effects of the present invention using labelling;
FIG. 5 is a graph showing the effect of error in repeated identification of banana plantlets detected by the present invention;
FIG. 6 is a schematic diagram of the deduplication process of the present invention;
FIG. 7 is a flow chart of the deduplication algorithm of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Referring to fig. 1 to 7, the present embodiment provides an automatic identification and counting method for banana young plants based on aerial visible light images, including the following steps:
s1, acquiring visible light images of field banana young plants as original images in an unmanned aerial vehicle aerial photography mode, wherein the unmanned aerial vehicle adopts Xintom 4, and the visible light images of the large-area banana young plants are shown in a figure 2;
s2, cutting the original image without overlapping degree to obtain a plurality of sub-images, and specifically comprising the following steps:
s2.1, creating (m +1) × (n +1) uniform grid vertex coordinates according to the size of the original image, and performing pixel-level fine adjustment on the original image by adopting an image scaling method to enable the length and the width of the original image to be evenly divided by m and n;
s2.2, storing the image by adopting a method for increasing dimensionality, creating a five-dimensional tensor, wherein the five-dimensional tensor is used for storing the cut sub-images, the front two-dimensional representation in the five-dimensional tensor represents the coordinate position of the sub-images in the original image (such as the block of the ith row and the jth column), and the rear three-dimensional representation represents the information (such as the length, the width and the channel number) of each sub-image;
and S2.3, cutting the original image by adopting a numpy slicing function, and storing sub-image pixel data determined by the coordinate values of the upper left corner and the lower right corner of each grid into the five-dimensional tensor.
S3, carrying out image preprocessing on the sub-graph, and labeling the banana young plants to obtain labeled sub-graphs of the banana young plants;
the contrast difference between partial subgraphs is obvious, image preprocessing is carried out in order to eliminate the difference between the subgraphs, and the brightness value of the image in the HSV color space is equalized by applying a self-adaptive histogram equalization algorithm (CLAHE) with limited contrast, so that the image contrast is enhanced, the image difference caused by the change of ambient light is reduced, and as shown in figure 3, the enhancement algorithm has the effect of obviously improving the brightness of dim subgraphs, and meanwhile, highlight subgraphs cannot be overexposed. And manually labeling the processed sub-graph by using a labeling tool labellImg, wherein incomplete banana plants appear at the edge part of the sub-graph due to grid cutting, labeling plants with display pictures exceeding half plants mainly during labeling, and a data set image is shown in fig. 4.
S4, constructing a fast-RCNN model, and training the model based on the banana young plant labeled subgraph;
dividing a data set from the labeling subgraph of the banana young plants, and according to the training set: and (3) test set: the verification set is 8: 1: 1, randomly splitting an original data set, and setting parameters of a model as shown in a table 1;
TABLE 1 model training parameters
S5, cutting the image to be detected into a sub-image, inputting the sub-image into a trained Faster-RCNN model, and detecting to obtain a sub-image recognition image;
s6, splicing the sub-image recognition images into a complete image again, wherein the splicing principle is that an empty original image is created firstly, and then corresponding sub-images are filled in corresponding positions, the sub-images containing the recognition results are spliced to enable a user to clearly see the complete recognition results, but if the number is simply counted, only the recognized frames are spliced; removing the weight of the banana plants which are divided into two or four in the cutting process in the large graph by repeating the frame selection; the repeated recognition phenomenon is shown in fig. 5, the deduplication principle diagram is shown in fig. 6, the deduplication algorithm design block diagram is shown in fig. 7, counting is finally completed, and deduplication processing specifically includes:
s6.1, calculating the coordinates of the center point of the sub-graph based on the grid vertex coordinates (midx1, midy 1):
s6.2, traversing and calculating distances d1 and d2 between the marked box center point and the sub-graph edge of each banana young plant, setting a first preset distance threshold T1 and a second preset distance threshold T2, and screening out the sub-graph edge marked boxes based on the first preset distance threshold T1 and the second preset distance threshold T2, wherein the formula is as follows:
d2＝|midx-2m|；
d1＝|midy-h|；
d1＜T1 and d2＜T2；
wherein m is the width of the subgraph and h is the height of the subgraph.
The first preset distance threshold and the second preset distance threshold are both variable thresholds, and the setting of the variable thresholds is specifically to set according to a visible light original image of an aerial photography height, and the threshold selected under the current aerial photography height of 80m cannot be compatible with aerial photography images at other heights, so the variable thresholds are designed to meet the accuracy of a sub-image guarantee algorithm under different flight heights, the distance threshold of the current aerial photography height of 80m is set to be threshold0, the distance threshold of the aerial photography image needing to be counted is input to be threshold1, the distance Δ d0 between every two banana young plants in the sub-image of 80m aerial photography image is calculated, the distance Δ d1 between every two banana young plants in the sub-image is input, and then the calculation formula of the new thresholds is as follows:
wherein, threshold0 is the distance of the current aerial height, and threshold1 is the distance threshold of aerial images of banana young plants to be counted.
S6.3, calculating the area of the labeling square frame at the edge of the subgraph, setting an area threshold, and screening out the labeling square frame of the incomplete banana young plant based on the area threshold, wherein the formula of the area threshold S is as follows:
s＝(x2-x1)(y2-y1)。
s6.4, calculating the distance between the central points of two adjacent boxes in the marked boxes of the incomplete banana young plants, and identifying the repeated detection condition through a third preset distance threshold value:
the invention has the following technical effects:
according to the method, the images of banana young plants at different light conditions and different shooting heights are obtained on the spot by the unmanned aerial vehicle to construct the original images, aiming at the problem that the number of candidate images generated by the Faster-RCNN framework on the input image is limited, the target crops of the target images are reduced by using a processing method of image non-overlapping degree cutting, and then the large-area detection of the number of banana young plants is realized by adopting a mode of splicing back to a large image. In the subgraph identification result, the boundary generated by the subgraph during interception divides a large number of banana plants into two parts, even one part into four parts. Some trans-border banana plants are respectively detected in adjacent subgraphs (repeated boundary identification), so that the statistical quantity is increased, and a large error is occupied. The counting method for automatically identifying the banana young plants, which is provided by the target detection model based on the Faster-RCNN, has high robustness under the variation of variables such as height, light conditions and the like, is efficient and portable, greatly reduces the manpower input, can provide related technical support for users in the aspects of farm management, yield prediction and the like, and has strong application and popularization potentials.
The above-described embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the scope of the present invention, and various modifications and improvements of the technical solutions of the present invention can be made by those skilled in the art without departing from the spirit of the present invention, and the technical solutions of the present invention are within the scope of the present invention defined by the claims.
Claims (7)
1. An aerial visible light image-based banana young plant automatic identification and counting method is characterized by comprising the following steps:
s1, collecting a visible light image of a banana young plant as an original image;
s2, cutting the original image without overlapping degree to obtain a plurality of sub-images;
s3, carrying out image preprocessing on the subgraph, and labeling the banana young plant to obtain a labeled subgraph of the banana young plant;
s4, constructing a fast-RCNN model, and training the fast-RCNN model based on the banana young plant labeled subgraph;
s5, cutting the image to be detected into a sub-image, inputting the sub-image into a trained Faster-RCNN model, and obtaining a sub-image recognition image;
and S6, splicing the sub-image recognition images into a complete image again, and performing de-duplication treatment on the boundary banana young plants to finish counting.
2. The method for automatically identifying and counting banana seedlings based on aerial visible light images as claimed in claim 1, wherein the specific steps of S2 non-overlapping image cropping are as follows:
s2.1, creating a plurality of uniform grid vertex coordinates according to the size of the original image;
s2.2, creating a five-dimensional tensor, wherein the five-dimensional tensor is used for storing the cut sub-images, the front two dimensions in the five-dimensional tensor represent the coordinate positions of the sub-images in the original image, and the rear three dimensions represent the information of each sub-image;
and S2.3, cutting the original image, and storing sub-image data determined by the coordinate values of the upper left corner and the lower right corner of each grid into the five-dimensional tensor.
3. The method for automatically identifying and counting banana young plants based on aerial visible light images as claimed in claim 1, wherein the S3 preprocessing is implemented by applying a contrast-limited CLAHE algorithm to balance brightness values of images in HSV color space.
4. The method for automatically identifying and counting banana seedlings based on aerial visible light images as claimed in claim 2, wherein the specific calculation steps of the de-duplication process in S6 are as follows:
s6.1, calculating the coordinates of the central point of the sub-graph;
s6.2, traversing and calculating the distance between the box center point marked by each banana young plant and the sub-image edge, setting a first preset distance threshold and a second preset distance threshold, and screening out the sub-image edge marking box based on the first preset distance threshold and the second preset distance threshold;
s6.3, setting an area threshold value by calculating the area of the labeling square frame at the edge of the subgraph, and screening out the labeling square frames of incomplete banana young plants based on the area threshold value;
s6.4, calculating the distance between the central points of two adjacent boxes in the marked boxes of the incomplete banana young plants, and identifying the repeated detection condition through a third preset distance threshold value.
5. The method for automatically identifying and counting banana young plants based on aerial visible light images as claimed in claim 4, wherein the first preset distance threshold and the second preset distance threshold are both variable thresholds.
6. The method for automatically identifying and counting banana young plants based on aerial visible light images as claimed in claim 5, wherein the setting of the variable threshold is specifically based on an aerial height visible light original image.
7. The method for automatically identifying and counting banana young plants based on aerial visible light images as claimed in claim 6, wherein the variable threshold is calculated as:
wherein, threshold0 is the distance of the current aerial photography height, threshold1 is the distance threshold of aerial photography image of banana young plants to be counted, Δ d0 is the distance of every two banana young plants in the sub-image of the current aerial photography image, and Δ d1 is the distance of every two banana young plants in the sub-image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110619392.XA CN113313692B (en) | 2021-06-03 | 2021-06-03 | Automatic banana young plant identification and counting method based on aerial visible light image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110619392.XA CN113313692B (en) | 2021-06-03 | 2021-06-03 | Automatic banana young plant identification and counting method based on aerial visible light image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113313692A true CN113313692A (en) | 2021-08-27 |
| CN113313692B CN113313692B (en) | 2023-04-25 |
Family
ID=77377275
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110619392.XA Active CN113313692B (en) | 2021-06-03 | 2021-06-03 | Automatic banana young plant identification and counting method based on aerial visible light image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113313692B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113888397A (en) * | 2021-10-08 | 2022-01-04 | 云南省烟草公司昆明市公司 | Tobacco pond cleaning and plant counting method based on unmanned aerial vehicle remote sensing and image processing technology |
| CN114742822A (en) * | 2022-05-20 | 2022-07-12 | 青岛农业大学 | Construction method and application of strawberry identification and counting model |
| CN118097121A (en) * | 2024-04-18 | 2024-05-28 | 浙江双元科技股份有限公司 | Target recognition and counting method and device based on image segmentation and deep learning |
| CN118097121B (en) * | 2024-04-18 | 2024-06-28 | 浙江双元科技股份有限公司 | Target recognition and counting method and device based on image segmentation and deep learning |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106327426A (en) * | 2016-08-19 | 2017-01-11 | 携程计算机技术（上海）有限公司 | Image replication removing method and image replication removing system |
| CN107818343A (en) * | 2017-10-30 | 2018-03-20 | 中国科学院计算技术研究所 | Method of counting and device |
| CN109636824A (en) * | 2018-12-20 | 2019-04-16 | 巢湖学院 | A kind of multiple target method of counting based on image recognition technology |
| CN110569786A (en) * | 2019-09-06 | 2019-12-13 | 中国农业科学院农业资源与农业区划研究所 | fruit tree identification and quantity monitoring method and system based on unmanned aerial vehicle data acquisition |
| CN111080529A (en) * | 2019-12-23 | 2020-04-28 | 大连理工大学 | Unmanned aerial vehicle aerial image splicing method for enhancing robustness |
| CN111537515A (en) * | 2020-03-31 | 2020-08-14 | 国网辽宁省电力有限公司朝阳供电公司 | Iron tower bolt defect display method and system based on three-dimensional live-action model |
| CN111738159A (en) * | 2020-06-23 | 2020-10-02 | 桂林市思奇通信设备有限公司 | Cotton terminal bud positioning method based on vector calibration |
| CN112017231A (en) * | 2020-08-27 | 2020-12-01 | 中国平安财产保险股份有限公司 | Human body weight identification method and device based on monocular camera and storage medium |
-
2021
- 2021-06-03 CN CN202110619392.XA patent/CN113313692B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106327426A (en) * | 2016-08-19 | 2017-01-11 | 携程计算机技术（上海）有限公司 | Image replication removing method and image replication removing system |
| CN107818343A (en) * | 2017-10-30 | 2018-03-20 | 中国科学院计算技术研究所 | Method of counting and device |
| CN109636824A (en) * | 2018-12-20 | 2019-04-16 | 巢湖学院 | A kind of multiple target method of counting based on image recognition technology |
| CN110569786A (en) * | 2019-09-06 | 2019-12-13 | 中国农业科学院农业资源与农业区划研究所 | fruit tree identification and quantity monitoring method and system based on unmanned aerial vehicle data acquisition |
| CN111080529A (en) * | 2019-12-23 | 2020-04-28 | 大连理工大学 | Unmanned aerial vehicle aerial image splicing method for enhancing robustness |
| CN111537515A (en) * | 2020-03-31 | 2020-08-14 | 国网辽宁省电力有限公司朝阳供电公司 | Iron tower bolt defect display method and system based on three-dimensional live-action model |
| CN111738159A (en) * | 2020-06-23 | 2020-10-02 | 桂林市思奇通信设备有限公司 | Cotton terminal bud positioning method based on vector calibration |
| CN112017231A (en) * | 2020-08-27 | 2020-12-01 | 中国平安财产保险股份有限公司 | Human body weight identification method and device based on monocular camera and storage medium |
Non-Patent Citations (2)
| Title |
|---|
| 史红栩: "基于深度学习的香蕉病害诊断***及株数统计算法的实现", 《中国优秀硕士论文全文数据库（农业科技辑）》 * |
| 王侨等: "基于机器视觉的玉米苗期多条作物行线检测算法" * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113888397A (en) * | 2021-10-08 | 2022-01-04 | 云南省烟草公司昆明市公司 | Tobacco pond cleaning and plant counting method based on unmanned aerial vehicle remote sensing and image processing technology |
| CN114742822A (en) * | 2022-05-20 | 2022-07-12 | 青岛农业大学 | Construction method and application of strawberry identification and counting model |
| CN118097121A (en) * | 2024-04-18 | 2024-05-28 | 浙江双元科技股份有限公司 | Target recognition and counting method and device based on image segmentation and deep learning |
| CN118097121B (en) * | 2024-04-18 | 2024-06-28 | 浙江双元科技股份有限公司 | Target recognition and counting method and device based on image segmentation and deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113313692B (en) | 2023-04-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jia et al. | Detection and segmentation of overlapped fruits based on optimized mask R-CNN application in apple harvesting robot | |
| Gai et al. | A detection algorithm for cherry fruits based on the improved YOLO-v4 model | |
| US8340420B2 (en) | Method for recognizing objects in images | |
| CN110569786B (en) | Fruit tree identification and quantity monitoring method and system based on unmanned aerial vehicle data acquisition | |
| Li et al. | An overlapping-free leaf segmentation method for plant point clouds | |
| US20150310305A1 (en) | Learning painting styles for painterly rendering | |
| CN110120065B (en) | Target tracking method and system based on hierarchical convolution characteristics and scale self-adaptive kernel correlation filtering | |
| CN112990103B (en) | String mining secondary positioning method based on machine vision | |
| CN107220647B (en) | Crop center point positioning method and system under blade crossing condition | |
| CN108319894A (en) | Fruit recognition methods based on deep learning and device | |
| CN109711268B (en) | Face image screening method and device | |
| CN113313692A (en) | Automatic banana young plant identification and counting method based on aerial visible light image | |
| Lv et al. | A visual identification method for the apple growth forms in the orchard | |
| CN113298023B (en) | Insect dynamic behavior identification method based on deep learning and image technology | |
| US11880981B2 (en) | Method and system for leaf age estimation based on morphological features extracted from segmented leaves | |
| CN114758132B (en) | Fruit tree disease and pest identification method and system based on convolutional neural network | |
| CN114359199A (en) | Fish counting method, device, equipment and medium based on deep learning | |
| CN113298768B (en) | Cotton detection, segmentation and counting method and system | |
| CN115330833A (en) | Fruit yield estimation method with improved multi-target tracking | |
| CN111738310B (en) | Material classification method, device, electronic equipment and storage medium | |
| CN117095153A (en) | Multi-mode fruit perception system, device and storage medium | |
| CN110853052A (en) | Tujia brocade pattern primitive segmentation method based on deep learning | |
| CN115995017A (en) | Fruit identification and positioning method, device and medium | |
| CN114937030A (en) | Phenotypic parameter calculation method for intelligent agricultural planting of lettuce | |
| CN110276260B (en) | Commodity detection method based on depth camera |
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 |