LU102950B1 - Tobacco plant counting method based on uav remote sensing technology and image processing technology - Google Patents
Tobacco plant counting method based on uav remote sensing technology and image processing technology Download PDFInfo
- Publication number
- LU102950B1 LU102950B1 LU102950A LU102950A LU102950B1 LU 102950 B1 LU102950 B1 LU 102950B1 LU 102950 A LU102950 A LU 102950A LU 102950 A LU102950 A LU 102950A LU 102950 B1 LU102950 B1 LU 102950B1
- Authority
- LU
- Luxembourg
- Prior art keywords
- tobacco
- image
- tobacco plant
- remote
- outline
- Prior art date
Links
- 235000002637 Nicotiana tabacum Nutrition 0.000 title claims abstract description 159
- 244000061176 Nicotiana tabacum Species 0.000 title claims abstract description 99
- 238000005516 engineering process Methods 0.000 title claims abstract description 46
- 238000000034 method Methods 0.000 title claims abstract description 35
- 241000208125 Nicotiana Species 0.000 claims abstract description 60
- 239000002689 soil Substances 0.000 claims abstract description 29
- 238000001914 filtration Methods 0.000 claims abstract description 24
- 230000000877 morphologic effect Effects 0.000 claims abstract description 9
- 230000000694 effects Effects 0.000 claims description 6
- 101100496858 Mus musculus Colec12 gene Proteins 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 4
- 101100534231 Xenopus laevis src-b gene Proteins 0.000 claims description 3
- 238000005260 corrosion Methods 0.000 claims description 3
- 230000007797 corrosion Effects 0.000 claims description 3
- 238000009826 distribution Methods 0.000 claims description 3
- 238000000926 separation method Methods 0.000 claims description 3
- 241001270131 Agaricus moelleri Species 0.000 claims description 2
- 241000196324 Embryophyta Species 0.000 abstract description 3
- 238000007689 inspection Methods 0.000 abstract description 3
- 238000012271 agricultural production Methods 0.000 abstract description 2
- 238000003706 image smoothing Methods 0.000 abstract 1
- 238000012790 confirmation Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 238000009499 grossing Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
Classifications
-
- 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
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/04—Context-preserving transformations, e.g. by using an importance map
-
- 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
- G06T5/20—Image enhancement or restoration using local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background 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
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2200/00—Indexing scheme for image data processing or generation, in general
- G06T2200/32—Indexing scheme for image data processing or generation, in general involving image mosaicing
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
-
- 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
- G06T2207/20036—Morphological 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
- G06T2207/20172—Image enhancement details
- G06T2207/20182—Noise reduction or smoothing in the temporal domain; Spatio-temporal filtering
-
- 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
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Geometry (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to a tobacco plant counting method based on a UAV remote sensing technology and an image processing technology. The method includes the following steps: acquiring visible orthoimages of a tobacco field by using the UAV remote sensing technology, and stitching the orthoimages to acquire a complete remote-sensing image map of the tobacco field; filtering out information about a soil background from the remote-sensing image map of the tobacco field by using an HSV model mask method, and extracting an outline of a tobacco plant by sequentially performing gray-scale processing, Gaussian filtering-based image smoothing, threshold binarization, a morphological operation, and open operation-based denoising; and filtering out an outline having an over-large or over-small pixel area, detecting a minimum bounding rectangle of the outline of the tobacco plant, and counting minimum bounding rectangles to acquire the quantity of tobacco plants. In the present invention, plant counting is performed by using the UAV remote sensing technology and the image processing technology. Therefore, the method has the advantages of a high inspection speed, high identification accuracy, and low costs. This brings a new mode for digitization and informationization of agricultural production of tobacco.
Description
C72P3LU i 05.05.2022 TOBACCO PLANT COUNTING METHOD BASED ON UAV 00050
[0001] The present invention belongs to the field of agricultural technologies of tobacco, and relates to a tobacco plant counting method based on an unmanned aerial vehicle (UAV) remote sensing technology and an image processing technology.
[0002] Plant counting is a key link in a tobacco production process, and refers to accurate counting after transplanting of tobacco seedlings to ensure unification of quantity, area, contract, and persons. At present, most counting methods are manual counting methods. However, the manual counting method has low efficiency. high costs, and a low capability in accuracy Statistics. In addition, information such as a planting area, plot boundaries, and persons is hard to verify simultaneously in a counting process. This brings great difficulty to statistical work. Although planting of tobacco plants has strict requirements for a row distance, if an area of a plot is computed based on a UAV remote-sensing image and the quantity of tobacco plants is estimated based on the area, it is hard to guarantee counting accuracy because plots are complex and diverse and it is hard to implement accurately the row distance in a planting process. As a result, a use effect is unsatisfactory.
[0003] In view of this, an objective of the present invention is to provide a tobacco plant counting method based on a UAV remote sensing technology and an image processing technology, which has a high inspection speed, high identification accuracy, and low costs.
[0004] To achieve the foregoing objective, the present invention provides the following technical solutions.
[0005] A tobacco plant counting method based on a UAV remote sensing technology and an image processing technology is provided. The method includes the following steps:
[0006] S1: acquiring visible orthoimages of a tobacco field by using the UAV remote sensing technology, and stitching the orthoimages to acquire a complete remote-sensing image map of the tobacco field; 1
C72P3LU 05.05.2022
[0007] S2: filtering out information about a soil background of the remote-sensing image map of the tobacco field to separate tobacco plants from the soil background; LU102950
[0008] S3: performing gray-scale processing, filtering, and binarization on an image acquired after separation to acquire a binarized image;
[0009] S4: performing a morphological operation and open operation—based denoising on the binarized image to remove an irrelevant external white pixel point interference region around a tobacco plant;
[0010] S5: extracting an outline of the tobacco plant;
[0011] S6: computing information about the outline to detect coordinate information of an upper left corner and a lower right corner of a minimum bounding rectangle of the outline of the tobacco plant, and circling the minimum bounding rectangle on the tobacco plant on the remote- sensing image map of the tobacco field; and
[0012] S7: counting circled minimum bounding rectangles to acquire the quantity of tobacco plants in the remote-sensing image map of the tobacco field.
[0013] Further, S2 includes the following sub-steps:
[0014] S20: changing a mode of the remote-sensing image map of the tobacco field from RGB to HSV; acquiring values of hue components H. saturation components S, and brightness components V corresponding to a soil background region and a tobacco plant region in the remote-sensing image map of the tobacco field; and setting value ranges for a hue component, a saturation component, and a brightness component of a target region, such that the tobacco plant region is extracted as the target region and becomes white, and the soil background region becomes black, thereby generating a mask image; and
[0015] S22: superimposing the mask image with a corresponding original remote-sensing image of the tobacco field, such that the soil background region of the remote-sensing image of the tobacco field is shielded by using a mask shielding effect, the information about the soil background of the remote-sensing image map of the tobacco field is filtered out effectively, and information about the tobacco plants is preserved to the most extent, thereby separating the tobacco plants from the soil background.
[0016] Further, in step S20, a value range of the hue component H is 46<H<69, a value range of the saturation component S is 0<S<255, and a value range of the brightness component V is O<V<255.
[0017] Further, in step S22, the mask image is defined as a first input array src1, the remote- sensing image map of the tobacco field is defined as a second input array src2, and dst is defined as an output array having the same size as the input array, wherein all elements are 0 by default, 2
C72P3LU 05.05.2022 and an "and" operation is performed on the mask image and the remote-sensing image map of the tobacco field according to the following formula: LU102950 dst(I) = srcl(I) Asre(I) , mask(T) #0
[0018] | dst(I)=0 , mask(I)=0
[0019] wherein mask denotes an optional operation code.
[0020] Further, step S3 includes the following sub-steps:
[0021] S30: performing gray-scale processing on a superimposed image, such that a color mode of the superimposed image is changed to a gray-scale mode;
[0022] S32: performing Gaussian filtering on an image whose color mode is changed to the gray-scale mode; and
[0023] S34: performing binarization on an image acquired after the Gaussian filtering to improve extraction precision of the outline of the tobacco plant.
[0024] Further, in step S30, a gray-scale value of a single component is computed according to the following formula:
[0025] Y = 0.299R + 0.587G + 0.114B,
[0026] wherein Y denotes a gray-scale value of the single component; and R, G, and B respectively denote three channel components corresponding to an RGB color model.
[0027] Further, in step S32, a method in which Gaussian filtering is used to smooth an image is as follows:
[0028] defining an input image as S(u,v), and processing the input image by using a two- dimensional Gaussian function to acquire G(u,v) as an output image, wherein the following computing formula is used: Glu,v) = — om (W420?)
[0029] 210 ,
[0030] wherein c denotes a standard deviation of normal distribution.
[0031] Further, in step S34, a method for binarizing a threshold includes:
[0032] setting a gray-scale threshold T1 to 26; setting a gray-scale value that is less than or equal to the threshold 26 to 0; and setting a gray-scale value that is greater than the threshold 26 to a maximum value maxVal, such that in the image, a gray-scale value of the background region is changed to 0, and a gray-scale value of the tobacco plant region is changed to maxVal.
[0033] Further, step S4 includes the following sub-steps: 3
C72P3LU 05.05.2022 | 10 1 11 LU102950
[0034] S40: defining a 3*3 cross-shaped structure element 01 ; ; performing the morphological operation based on the structure element and the image; and feeding back a structure element having a specified shape and size; and
[0035] S42: performing operations of corrosion first and then expansion by using an open operation; separating the tobacco plants from some background regions; and removing the irrelevant external white pixel point interference region around the tobacco plant, with the other portions unchanged.
[0036] Further, step S5 includes the following sub-steps:
[0037] S50: reserving only coordinates of end points of a horizontal direction, a vertical direction, and diagonal directions by compressing elements in these directions: outputting external outline information of the tobacco plant; and drawing the outline of the tobacco plant based on the outline information; and
[0038] S52: computing, based on the outline information, a pixel area of a shape enclosed by the outline; and filtering out an over-large or over-small outline based on the pixel area, thereby removing interference information.
[0039] In the present invention, plant counting is performed by using the UAV remote sensing technology and the image processing technology. Therefore, the method has the advantages of a high inspection speed, high identification accuracy, and low costs; and work such as area verification, boundary confirmation, and matching between persons and plot information can be performed during counting. This brings a new mode for digitization and informationization of agricultural production of tobacco.
[0040] To make the objectives, technical solutions, and advantages of the present invention clearer, the following preferably describes the present invention in detail with reference to the accompanying drawings.
[0041] FIG. 1 is a flowchart of a preferred embodiment of a tobacco plant counting method based on a UAV remote sensing technology and an image processing technology according to the present invention;
[0042] FIG. 2 shows a remote-sensing image map of a tobacco field acquired through stitching after UAV remote sensing of a tobacco field;
[0043] FIG. 3 is a schematic diagram of a remote-sensing image map of a tobacco field whose color model is changed to an HSV color model; 4
C72P3LU 05.05.2022
[0044] FIG. 4 shows a color mask image generated based on the HSV color model;
[0045] FIG. 5 shows an HSV mask image acquired based on the color mask image; 7192950
[0046] FIG. 6 shows a gray-scale image acquired after gray-scale processing;
[0047] FIG. 7 shows a Gaussian filtering image acquired after an image is smoothed through Gaussian filtering;
[0048] FIG. 8 shows a binarized image acquired after threshold binarization;
[0049] FIG. 9 shows a denoised image of a tobacco plant acquired after a morphological operation and open operation-based denoising;
[0050] FIG. 10 shows an outline image of a tobacco plant acquired after outline detection and extraction; and
[0051] FIG. 11 is a diagram of a detection result of circling minimum bounding rectangles on tobacco plants on a remote-sensing image map of a tobacco field.
[0052] The following describes implementation of the present invention with specific embodiments. Those skilled in the art can easily learn other advantages and effects of the present invention based on the content disclosed in the Description. The present invention may also be implemented or applied in other different specific implementation manners. Details in this specification may also be modified or changed based on different viewpoints and application, without departing from the spirit of the present invention. It should be noted that, figures provided in the following embodiments are merely intended to illustratively describe the basic concept of the present invention. In the case of no conflict, the following embodiments and features in the embodiments may be combined with each other.
[0053] As shown in FIG. 1, a preferred embodiment of a tobacco plant counting method based on a UAV remote sensing technology and an image processing technology according to the present invention includes: first, changing an RGB color model of a remote-sensing image of a tobacco plant to an HSV color model by using an open source computer vision library (OpenCV); processing the changed remote-sensing image by using an HSV model mask method to filter out unnecessary background information; acquiring a binarized image of an outline of the tobacco plant through binaryzation; extracting the outline of the tobacco plant through morphological denoising and open operation—based denoising; and finally. acquiring a complete tobacco plant image with a clear outline by computing an area of a bounding shape of the outline of the tobacco plant and filtering out unnecessary interference information, thereby numbering the outline of each tobacco plant, and acquiring the final quantity of tobacco plants by counting numbers of the tobacco plants. The specific steps are as follows.
C72P3LU 05.05.2022
[0054] In step S1, UAV remote sensing is performed on a tobacco field. First, UAV remote sensing is performed on a tobacco field to which tobacco seedlings are transplanted about a 02950 days ago to acquire visible orthoimages of the tobacco field: and then, the orthoimages are stitched to acquire a complete remote-sensing image map of the tobacco field in an RGB format by using image stitching software. It is required that a spatial resolution of the remote-sensing image map is not less than 1.25 cm/pixel, as shown in FIG. 2.
[0055] In step S2, information about a soil background of the remote-sensing image map of the tobacco field is filtered out to separate tobacco plants from the soil background. Step S2 may include the following sub-step.
[0056] In step S20, a mode of the remote-sensing image map of the tobacco field is changed from RGB to HSV; values of hue components H, saturation components S, and brightness components V corresponding to a soil background region and a tobacco plant region in the remote-sensing image map of the tobacco field are acquired; and value ranges for a hue component, a saturation component, and a brightness component of a target region are set, such that the tobacco plant region is extracted as the target region and becomes white, and the soil background region becomes black, thereby generating a mask image.
[0057] As shown in FIG. 3, which is a schematic diagram of a remote-sensing image map of a tobacco field whose color model is changed to an HSV color model, first, values of hue components, saturation components, and brightness components corresponding to a soil background region and a tobacco plant region in the remote-sensing image map of the tobacco field are acquired from the HSV color model. It is found, through observation and statistics on the values of the hue components, the saturation components, and the brightness components, that when a region whose hue component, saturation component, and brightness component meet formula (1) is used as the target region, the soil background can be separated from the tobacco plants to the most extent, thereby accurately counting the tobacco plants. According to formula (1), the tobacco plant region may be extracted as the target region and become white; and a soil portion that is not extracted is used as the background region and becomes black, thereby generating a mask image, as shown in FIG. 4. It can be learned from the mask image that, only two colors (black and white) can be displayed, that is, the target region is white, and the background region is black.
46 < H<69 | 0<S<255
[0058] O<V<255 1),
[0059] in the formula, H denotes a hue component (Hue), S denotes a saturation component (Saturation), and V denotes a brightness component (Value).
6
C72P3LU 05.05.2022
[0060] In step S22, the mask image is superimposed with a corresponding original remote- sensing image of the tobacco field, such that the soil background region of the remote-sensing 02950 image of the tobacco field is shielded by using a mask shielding effect, the information about the soil background of the remote-sensing image map of the tobacco field is filtered out effectively, and information about the tobacco plants is preserved to the most extent. thereby separating the tobacco plants from the soil background. To accurately remove the soil background, when the mask image is superimposed with the remote-sensing image map of the tobacco field, the soil background region of the remote-sensing image map of the tobacco field is shielded by using a mask shielding effect. Because the two images have the same size, the mask image is defined as a first input array srcl, the remote-sensing image map of the tobacco field is defined as a second input array src2, and dst is defined as an output array having the same size as the input array, wherein all elements are 0 by default, and an "and" operation is performed on the mask image and the remote-sensing image map of the tobacco field according to formula (2): dst(I) = srcl(I) A sre(T) , mask(l) #0
[0061] | dst(I)=0 , mask(I)=0 ©.
[0062] wherein mask denotes an optional operation code. À region where mask(I)#0, namely, a white region of the mask image, displays content of a corresponding region of the remote- sensing image map of the tobacco field. A region where mask(I)= 0, namely, a black region of the mask image, blocks a corresponding region of the remote-sensing image map of the tobacco field. Through superimposition of the two images, the information about the soil background of the remote-sensing image map of the tobacco field can be effectively filtered out; and the information about the tobacco plants can be reserved to the most extent. Therefore, the tobacco plants are separated from the soil background, as shown in FIG. 5.
[0063] In step S3. gray-scale processing, filtering, and binarization are performed on an image acquired after separation to acquire a binarized image. Step S3 may specifically include the following sub-steps.
[0064] In step S30, gray-scale processing is performed on a superimposed image, such that a color mode of the superimposed image is changed to a gray-scale mode. During gray-scale processing of an image. a gray-scale value of each component is computed first; then, color space is changed from an RGB mode to the gray-scale mode through gray-scale processing, thereby compressing original data of the image. speeding up subsequent processing of the image. reducing interference, and avoiding band distortion. A gray-scale value Y of each component is acquired according to formula (3). An image acquired after gray-scale processing is shown in FIG. 6. 7
C72P3LU 05.05.2022
[0065] Y = 0.299R + 0.587G + 0.114B (3).
[0066] in the formula, R, G, and B respectively denote three channel components corresponding. 02950 to the RGB color model.
[0067] In step S32, Gaussian filtering is performed on an image whose color mode is changed to the gray-scale mode, thereby smoothing the image through Gaussian filtering. A linear smoothing filter is selected as a Gaussian filter based on a shape of a Gaussian function to eliminate an error caused by noise in subsequent application of a digital image. A specific method is as follows:
[0068] establishing a two-dimensional UV coordinate system; defining an input image as S(u,v); replacing a value of a center pixel of a template with a weighted average gray-scale value of pixels in a neighborhood; and processing the image by using a two-dimensional Gaussian function to acquire G(u,v) as an output image, as shown in formula (4). An image shown in FIG. 7 acquired after Gaussian filtering is acquired according to formula (4).
Glu,v) = — e- (+2 )/20?)
[0069] 210 (#),
[0070] wherein o denotes a standard deviation of normal distribution.
[0071] In step S34, binarization is performed on an image acquired after the Gaussian filtering to acquire a binarized image. thereby improving extraction precision of the outline of the tobacco plant. During binarization, a gray-scale threshold Ty is set to 26; a gray-scale value that is less than or equal to the threshold 26 is set to 0; and a gray-scale value that is greater than the threshold 26 is set to a maximum value maxVal (generally, the maximum value is equal to 255), as shown in formula (5): hi Val sre(x,y)>T; dst(x,y) =
[0072] 0 sre(x,y)<T 6).
[0073] wherein dst(x,y) denotes a gray-scale value of a pixel output after threshold binarization; and src(x,y) denotes a gray-scale value of the pixel before threshold binarization. After binarization, the gray-scale value of the background region in the image becomes 0, but the gray- scale value of the tobacco plant region in the image becomes 255, which greatly reduces data amount. Therefore, the outline of the tobacco plant can be highlighted, as shown in FIG. 8.
[0074] In step S4, a morphological operation and open operation—based denoising are performed on the binarized image to remove an irrelevant external white pixel point interference region around a tobacco plant. Step S4 specifically includes the following sub-steps. 8
C72P3LU 05.05.2022 | 1 | 1 1 1 LU102950
[0075] In step S40, a 3*3 cross-shaped structure element 010 is defined; the morphological operation is performed based on the structure element and the image; and a structure element having a specified shape and size is fed back.
[0076] In step S42, operations of corrosion first and then expansion are performed by using an open operation; the tobacco plants are separated from some background regions; and the irrelevant external white pixel point interference region around the tobacco plant are removed, with the other portions unchanged, as shown in FIG. 9.
[0077] In step S5, an outline of the tobacco plant is extracted. Step S5 specifically includes the following sub-steps.
[0078] In step S50, only coordinates of end points of a horizontal direction, a vertical direction, and diagonal directions are reserved by compressing elements in these directions: external outline information of the tobacco plant is output; and the outline of the tobacco plant is drawn based on the outline information. as shown in FIG. 10.
[0079] In step S52, a pixel area of a shape enclosed by the outline is computed based on the outline information; and an outline having an over-large or over-small pixel area is filtered out by setting pixel area thresholds. thereby removing interference information, and acquiring a complete tobacco plant image with a clear outline. The following method may be used:
[0080] computing a pixel area ContoursArea of a shape enclosed by the outline based on the outline information; setting a minimum pixel area threshold T2 and a maximum pixel area threshold T3; and filtering out an outline having an over-large or over-small pixel area (ContoursArea<<T or ContoursAÂrea T3) based on the pixel area thresholds, thereby removing interference information. Because there is no large-area interference information in this embodiment, only T>=10 is set, and the maximum pixel area threshold T3 is not set, as shown in formula (6).
Contours +1 ContoursArea>T, Contours = oma +0 ContoursArea<T,
[0081] I (6),
[0082] wherein Contours denotes the quantity of outlines. Certainly, values of Tz and T3 can be adjusted according to actual conditions, provided that the interference information can be filtered out.
[0083] In step S6, coordinate information of an upper left corner and a lower right corner of a minimum bounding rectangle of the outline of the tobacco plant is detected based on selected outline information, and the minimum bounding rectangle is circled on the tobacco plant on the 9
C72P3LU 05.05.2022 remote-sensing image map of the tobacco field to acquire a detection result, as shown in FIG. 11. To facilitate statistics, the rectangles may be numbered while being circled. LU102950
[0084] In step S7, the quantity of the minimum bounding rectangles, namely, the quantity of the tobacco plants in the remote-sensing image map of the tobacco field is recorded based on numbers of the minimum bounding rectangles. In addition, an analysis result may be visually displayed and output. Therefore, work such as area verification. boundary confirmation, and matching between persons and plot information can be performed during counting.
[0085] Compared with manual tobacco plant counting, tobacco plant counting performed based on the UAV remote sensing technology and the image processing technology in this embodiment has the following advantages. :
[0086] Work efficiency can be greatly improved: work efficiency of conventional manual tobacco plant counting is 20 mu/person/day by average (1 mu=666.67 m”). A capability of a UAV tobacco plant counting system is determined based on a data acquisition capability, a data processing capability, and a data analyzing capability. The data acquisition capability is 5000 mu/UAV/day. The data processing capability is 2500 mu/hour. An identification capability is 1500 mu/minute (this value differs with an operational capability of a server). Based on the above factors, a capability of a single-UAV system is 5000 mu/day; and statistical accuracy of the system is higher than 93 %.
[0087] Use costs are reduced: costs of conventional manual tobacco plant counting are RMB 100 /person/day, and costs of UAV tobacco plant counting is about RMB 2 /mu.
[0088] During conventional manual tobacco plant counting, statistic on an area of a tobacco field and planting density cannot be collected; and accuracy of a counting result is hard to verify. During UAV tobacco plant counting, the quantity of tobacco plants, an area of a tobacco field, and planting density can be computed accurately; and accuracy of a counting result is higher than 93%.
[0089] Moreover, work such as area verification, boundary confirmation, and matching between persons and plot information can be performed during counting.
[0090] Finally, it should be noted that the above embodiments are merely used for describing, rather than restricting, the technical solutions of the present invention. Although the present invention is described in detail with reference to the preferred embodiments, a person of ordinary skill in the art should understand that the technical solutions of the present invention may be modified or equivalently replaced without departing from the objective and scope of the present invention, all of which shall fall within the scope of claims of the present invention.
Claims (10)
1. A tobacco plant counting method based on an unmanned aerial vehicle (UAV) remote sensing technology and an image processing technology, comprising the following steps: S1: acquiring visible orthoimages of a tobacco field by using the UAV remote sensing technology, and stitching the orthoimages to acquire a complete remote-sensing image map of the tobacco field; S2: filtering out information about a soil background of the remote-sensing image map of the tobacco field to separate tobacco plants from the soil background: S3: performing gray-scale processing, filtering, and binarization on an image acquired after separation to acquire a binarized image; S4: performing a morphological operation and open operation-based denoising on the binarized image to remove an irrelevant external white pixel point interference region around a tobacco plant; S35: extracting an outline of the tobacco plant; S6: computing information about the outline to detect coordinate information of an upper left corner and a lower right corner of a minimum bounding rectangle of the outline of the tobacco plant, and circling the minimum bounding rectangle on the tobacco plant on the remote- sensing image map of the tobacco field; and S7: counting circled minimum bounding rectangles to acquire a quantity of tobacco plants in the remote-sensing image map of the tobacco field.
2. The tobacco plant counting method based on a UAV remote sensing technology and an image processing technology according to claim 1, wherein S2 comprises the following sub- steps: S20: changing a mode of the remote-sensing image map of the tobacco field from RGB to HSV; acquiring values of hue components H, saturation components S, and brightness components V corresponding to a soil background region and a tobacco plant region in the remote-sensing image map of the tobacco field; and setting value ranges for a hue component, a saturation component, and a brightness component of a target region, such that the tobacco plant region is extracted as the target region and becomes white, and the soil background region becomes black, thereby generating a mask image; and S22: superimposing the mask image with a corresponding original remote-sensing image of the tobacco field, such that a soil background region of the remote-sensing image of the tobacco 11
C72P3LU 05.05.2022 field is shielded by using a mask shielding effect, the information about the soil background of the remote-sensing image map of the tobacco field is filtered out effectively, and information 02050 about the tobacco plants is preserved to the most extent, thereby separating the tobacco plants from the soil background.
3. The tobacco plant counting method based on a UAV remote sensing technology and an image processing technology according to claim 2, wherein in step S20, a value range of the hue component H is 46<H<69, a value range of the saturation component S is 0<S<255, and a value range of the brightness component V is 0<V<255.
4. The tobacco plant counting method based on a UAV remote sensing technology and an image processing technology according to claim 2, wherein in step S22, the mask image is defined as a first input array srcl, the remote-sensing image map of the tobacco field is defined as a second input array src2, and dst is defined as an output array having the same size as the input array, wherein all elements are 0 by default, and an "and" operation is performed on the mask image and the remote-sensing image map of the tobacco field according to the following formula: dst(T) = srcl(I) A sre(T) , mask(l) #0 | dst(I)=0 , mask(I)=0 wherein mask denotes an optional operation code.
5. The tobacco plant counting method based on a UAV remote sensing technology and an image processing technology according to claim 2, wherein S3 comprises the following sub- steps: S30: performing gray-scale processing on a superimposed image, such that a color mode of the superimposed image is changed to a gray-scale mode; S32: performing Gaussian filtering on an image whose color mode is changed to the gray- scale mode; and S34: performing binarization on an image acquired after the Gaussian filtering to improve extraction precision of the outline of the tobacco plant.
6. The tobacco plant counting method based on a UAV remote sensing technology and an image processing technology according to claim 5, wherein in step S30, a gray-scale value of a single component is computed according to the following formula: Y = 0.299R + 0.587G + 0.114B, 12
C72P3LU 05.05.2022 wherein Y denotes a gray-scale value of the single component; and R, G, and B respectively denote three channel components corresponding to an RGB color model. LU102950
7. The tobacco plant counting method based on a UAV remote sensing technology and an image processing technology according to claim 5, wherein in step S32, a method in which Gaussian filtering is used to smooth an image is as follows: defining an input image as S(u,v), and processing the input image by using a two- dimensional Gaussian function to acquire G(u,v) as an output image, wherein the following computing formula is used: Gv) = 1 e- (W472 )/(20?) 2n0* , wherein 6 denotes a standard deviation of normal distribution.
8. The tobacco plant counting method based on a UAV remote sensing technology and an image processing technology according to claim 5, wherein in step $34, a method for binarizing a threshold comprises: setting a gray-scale threshold T1 to 26; setting a gray-scale value that is less than or equal to the threshold 26 to 0; and setting a gray-scale value that is greater than the threshold 26 to a maximum value max Val, such that in the image, a gray-scale value of the background region is changed to 0, and a gray-scale value of the tobacco plant region is changed to max Val.
9. The tobacco plant counting method based on a UAV remote sensing technology and an image processing technology according to claim 1, wherein step S4 comprises the following sub- steps: ; 1 | 1 11 S40: defining a 3*3 cross-shaped structure element 010 ; performing the morphological operation based on the structure element and the image; and feeding back a structure element having a specified shape and size; and S42: performing operations of corrosion first and then expansion by using an open operation; separating the tobacco plants from some background regions; and removing the irrelevant external white pixel point interference region around the tobacco plant, with the other portions unchanged.
13 ra
C72P3LU 05.05.2022
10. The tobacco plant counting method based on a UAV remote sensing technology and an image processing technology according to claim 1, wherein step S5 comprises the following sup 102950 steps: S50: reserving only coordinates of end points of a horizontal direction, a vertical direction, and diagonal directions by compressing elements in these directions; outputting external outline information of the tobacco plant; and drawing the outline of the tobacco plant based on the outline information; and S52: computing, based on the outline information, a pixel area of a shape enclosed by the outline; and filtering out an over-large or over-small outline based on the pixel area, thereby removing interference information.
14
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111171584.5A CN113888397A (en) | 2021-10-08 | 2021-10-08 | Tobacco pond cleaning and plant counting method based on unmanned aerial vehicle remote sensing and image processing technology |
Publications (1)
Publication Number | Publication Date |
---|---|
LU102950B1 true LU102950B1 (en) | 2022-12-01 |
Family
ID=79005502
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
LU102950A LU102950B1 (en) | 2021-10-08 | 2022-05-12 | Tobacco plant counting method based on uav remote sensing technology and image processing technology |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN113888397A (en) |
LU (1) | LU102950B1 (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114820578B (en) * | 2022-05-26 | 2024-05-28 | 云南省烟草公司红河州公司 | Unmanned aerial vehicle remote sensing monitoring equipment for tobacco planting area |
CN114998770B (en) * | 2022-07-06 | 2023-04-07 | 中国科学院地理科学与资源研究所 | Highway identifier extraction method and system |
CN116824401A (en) * | 2023-05-22 | 2023-09-29 | 云南省烟草公司红河州公司 | Method for rapidly detecting chlorophyll content of tobacco based on unmanned aerial vehicle and RGB image |
Family Cites Families (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106503695B (en) * | 2016-12-02 | 2019-07-09 | 汕头大学 | A kind of tobacco plant identification and method of counting based on Aerial Images |
KR20180096966A (en) * | 2017-02-22 | 2018-08-30 | 전남대학교산학협력단 | Automatic Counting Method of Rice Plant by Centroid of Closed Rice Plant Contour Image |
CN109447945B (en) * | 2018-09-21 | 2021-11-30 | 河南农业大学 | Quick counting method for basic wheat seedlings based on machine vision and graphic processing |
CN109741357A (en) * | 2018-12-19 | 2019-05-10 | 浙江大学 | A kind of brassica napus inflorescence method of counting and device based on image procossing |
JP7212936B2 (en) * | 2019-03-22 | 2023-01-26 | 国立大学法人信州大学 | Heading center recognition method, recognition device, and harvester of headed vegetables |
JP6774544B2 (en) * | 2019-10-18 | 2020-10-28 | マクセルホールディングス株式会社 | Plant information acquisition system, plant information acquisition device, plant information acquisition method, crop management system and crop management method |
CN112131952A (en) * | 2020-08-26 | 2020-12-25 | 航天信德智图(北京)科技有限公司 | Corn seedling stage plant number information extraction based on unmanned aerial vehicle remote sensing image |
CN112149543B (en) * | 2020-09-16 | 2024-02-27 | 北京工业大学 | Building dust recognition system and method based on computer vision |
CN112381819A (en) * | 2020-12-07 | 2021-02-19 | 云南省烟草公司昆明市公司 | HSV color model-based plant protection droplet detection method |
CN113313692B (en) * | 2021-06-03 | 2023-04-25 | 广西大学 | Automatic banana young plant identification and counting method based on aerial visible light image |
-
2021
- 2021-10-08 CN CN202111171584.5A patent/CN113888397A/en active Pending
-
2022
- 2022-05-12 LU LU102950A patent/LU102950B1/en active
Also Published As
Publication number | Publication date |
---|---|
CN113888397A (en) | 2022-01-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
LU102950B1 (en) | Tobacco plant counting method based on uav remote sensing technology and image processing technology | |
CN110008969B (en) | Method and device for detecting image saliency region | |
CN109447945B (en) | Quick counting method for basic wheat seedlings based on machine vision and graphic processing | |
CN107316077A (en) | A kind of fat cell automatic counting method based on image segmentation and rim detection | |
CN107066952A (en) | A kind of method for detecting lane lines | |
CN110852323B (en) | Angular point-based aerial target detection method | |
CN103745104A (en) | Examination paper marking method based on augmented reality technology | |
CN112381819A (en) | HSV color model-based plant protection droplet detection method | |
CN107358184A (en) | The extracting method and extraction element of document word | |
CN109978895B (en) | Tongue image segmentation method and device | |
CN110691226A (en) | Image processing method, device, terminal and computer readable storage medium | |
CN112861654A (en) | Famous tea picking point position information acquisition method based on machine vision | |
CN105139408A (en) | Plant green feature identification method based on Kinect sensor | |
CN108711160A (en) | A kind of Target Segmentation method based on HSI enhancement models | |
US20080075366A1 (en) | Apparatus and program for evaluating images | |
CN110796654A (en) | Guide wire detection method, device, equipment, tyre crane and medium | |
CN113487473A (en) | Method and device for adding image watermark, electronic equipment and storage medium | |
CN102855025A (en) | Optical multi-touch contact detection method based on visual attention model | |
CN113537140B (en) | Weed detection method based on deep neural network | |
Xu et al. | Computer assistance image processing spores counting | |
CN109035306A (en) | Moving-target automatic testing method and device | |
CN114463440A (en) | Single-camera target positioning method, system, equipment and storage medium | |
Krishna et al. | Digital image processing techniques in character recognition-a survey | |
CN109934215B (en) | Identification card identification method | |
CN111105394B (en) | Method and device for detecting characteristic information of luminous pellets |