CN116109658B - Harvester control data processing method based on 5G technology - Google Patents
Harvester control data processing method based on 5G technology Download PDFInfo
- Publication number
- CN116109658B CN116109658B CN202310363307.7A CN202310363307A CN116109658B CN 116109658 B CN116109658 B CN 116109658B CN 202310363307 A CN202310363307 A CN 202310363307A CN 116109658 B CN116109658 B CN 116109658B
- Authority
- CN
- China
- Prior art keywords
- point
- minimum
- value
- pixel
- points
- 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
- 238000005516 engineering process Methods 0.000 title claims abstract description 13
- 238000003672 processing method Methods 0.000 title claims abstract description 9
- 238000012545 processing Methods 0.000 claims abstract description 15
- 239000013598 vector Substances 0.000 claims description 42
- 238000000034 method Methods 0.000 claims description 39
- 238000003780 insertion Methods 0.000 claims description 12
- 230000037431 insertion Effects 0.000 claims description 12
- 238000004458 analytical method Methods 0.000 claims description 8
- 238000012935 Averaging Methods 0.000 claims description 2
- 230000006870 function Effects 0.000 description 9
- 238000003306 harvesting Methods 0.000 description 8
- 238000003708 edge detection Methods 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 6
- 230000011218 segmentation Effects 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000005070 sampling Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000005484 gravity Effects 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 210000005069 ears Anatomy 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000010902 straw Substances 0.000 description 1
- 230000016776 visual perception Effects 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
- G06T7/10—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
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4007—Scaling of whole images or parts thereof, e.g. expanding or contracting based on interpolation, e.g. bilinear interpolation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
-
- 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
- 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
-
- 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
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
Abstract
The invention relates to the technical field of image data processing, and provides a harvester control data processing method based on a 5G technology, which comprises the following steps: collecting crop images, and dividing to obtain lodging areas; acquiring a plurality of initial minimum points according to gray values in the crop image, acquiring a plurality of first minimum points in a lodging area, and acquiring the guiding necessary degree of each first minimum point according to the gray values of the neighborhood of the first minimum points so as to acquire a plurality of second minimum points; constructing an artificial potential field model for each second minimum point according to the gray value of the adjacent pixel point, acquiring a reference point of each second minimum point according to the gravitational field and the repulsive field in the artificial potential field model, and interpolating the neighborhood of each second minimum point according to the reference point and the neighborhood pixel point to obtain a processed crop image; judging the lodging condition of crops according to the processed crop images, and controlling a reel of the harvester. The invention aims to accurately identify lodging crops and improve the working efficiency of a harvester.
Description
Technical Field
The invention relates to the technical field of image data processing, in particular to a harvester control data processing method based on a 5G technology.
Background
The intelligent agricultural machinery under the 5G network can get rid of excessive dependence on people, and control and management of remote one person to multiple machines can be conveniently realized; in the harvester or combine of the intelligent agricultural machine, a seedling pulling and supporting device is arranged on the harvester or combine, which is called a seedling pulling device; the harvester can harvest crops with the lodging dip angle of more than 10 degrees by lengthening the reel, and adopts a reverse harvesting mode according to the lodging direction of the crops during operation, thereby being beneficial to feeding the crop straws and reducing the loss of ears.
In identifying lodging crops, it is often necessary to extract and analyze the edges of the lodging crop at a higher image resolution; however, due to the high planting density of crops, the influence of weather light and the like, edges among crops in images acquired by a front-mounted camera of the harvester may not be clear, so that the lodging condition of the crops is difficult to accurately identify, and further, the adjustment of related parameters of a reel in the harvester is difficult to control; in the prior art, the definition of an image is usually improved by utilizing upsampling interpolation, however, an edge blurring region is a weak expression region with an unobvious boundary, the traditional interpolation method is influenced by pixels with larger neighborhood pixel values, and partial regions with smaller pixel values in the edge blurring region are not enhanced, so that the purpose of enhancing the weak expression region cannot be achieved, therefore, the weak expression region needs to be analyzed, and the difference value is carried out by combining the pixel values of the similar pixels with a larger range, so that the weak expression region is enhanced, the definition of the image is further improved, and the lodging condition of crops is more accurately identified.
Disclosure of Invention
The invention provides a harvester control data processing method based on a 5G technology, which aims to solve the problem that the efficiency of a harvester is affected due to inaccurate identification of existing lodging crops, and adopts the following technical scheme:
one embodiment of the invention provides a harvester control data processing method based on 5G technology, which comprises the following steps:
acquiring a crop image and acquiring a lodging area in the crop image;
acquiring a plurality of initial minimum points according to gray values in the crop image, marking the initial minimum points in the lodging area as first minimum points, acquiring the guiding degree of each first minimum point according to the gray values of the neighborhood of the first minimum points, judging and acquiring a plurality of second minimum points according to the guiding degree through a first preset threshold value, and performing linear interpolation on pixel points except the second minimum points in the crop image;
constructing an artificial potential field model for each second minimum point according to the neighboring pixel points and the gray values thereof, acquiring other second minimum points in the artificial potential field model of each second minimum point, marking the other second minimum points as similar points of each second minimum point, and acquiring gravitational field vectors of each second minimum point and each similar point according to the space coordinates and the gray values of each second minimum point and each similar point; according to the space coordinates, gray values and artificial potential field models of each second minimum point and each similar point, repulsive force field vectors of each second minimum point and each similar point are obtained;
taking the sum of vectors of the gravitational field vector and the repulsive force field vector as a combined force field of each second minimum point and each similar point, acquiring the path flatness of each second minimum point and each similar point according to the combined force field vector, taking the similar point corresponding to the minimum value in the path flatness of each second minimum point as a reference point of each second minimum point, and acquiring the final interpolation value of the position to be interpolated in the neighborhood range of each second minimum point according to the gray value of the neighborhood pixel point of each second minimum point and the gray value of the reference point, so as to finish the interpolation of all pixel points in the crop image and obtain the processed crop image;
judging the lodging condition of crops according to the processed crop images, and controlling a reel of the harvester.
Optionally, the method for obtaining a plurality of initial minimum points according to the gray value in the crop image includes the following specific steps:
traversing crop images by a first preset window at a preset step length from the upper left corner of the images, traversing one row, and traversing the next row longitudinally by the preset step length; and (3) obtaining the pixel point with the minimum gray value in the first preset window every time through traversal, and marking all the pixel points with the minimum gray value obtained through traversal as initial minimum points.
Optionally, the method for obtaining the guiding degree of each first minimum point according to the gray value of the neighborhood of the first minimum point includes the following specific steps:
wherein ,
indicating the degree of guidance of the i-th first minimum point,
the gray value representing the ith first minimum point, n represents the number of neighbor pixels of the ith first minimum point, and n=8 here because of the analysis of eight neighbor pixels;
the gray value of the j-th neighborhood pixel point representing the i-th first minimum point,
representing absolute values, th represents a hyperbolic tangent function.
Optionally, the constructing the artificial potential field model for each second minimum point according to the neighboring pixel point and the gray value thereof includes the following specific methods:
setting a second preset window, taking each second electrode small point as the center of the second preset window, and taking the pixel point in the range of the second preset window corresponding to each second electrode small point as an artificial potential field model of each second electrode small point.
Optionally, the acquiring the gravitational field vector of each second minimum point and each similar point includes the following specific methods:
wherein ,
represents the gravitational field value of the kth second pole point and its qth similarity point,
representing the euclidean distance of the kth second pole point from its qth similarity point,
represents the gray value of the kth second pole dot,
gray value of the (q) th similar point representing the (k) th second pole small point, exp [ []Represents an exponential function with a base of a natural constant,
a gain constant representing the gravitational field;
the direction of the kth second pole point to the qth similar point is recorded as the gravitational field value
The direction of the (b) is that the gravity field vector of the kth second pole small point and the q-th similar point is obtained and expressed as
。
Optionally, the specific method for obtaining the repulsive force field vector of each second minimum point and each similar point includes:
wherein ,
representing the kth second pole dot and the qth pole dotThe repulsive force vector of the s-th pixel point in the repulsive force field path of the similar point,
representing the repulsive force value of the s-th pixel point in the repulsive force field path of the kth second pole small point and the q-th similar point,
representing the number of neighbor pixels to be analyzed of the s-th pixel in the repulsive force field path of the kth second pole small point and the q-th similar point,
representing the gray value of the s-th pixel point in the repulsive force field path of the kth second pole small point and the q-th similar point,
representing the kth pixel point in the repulsive force field path of the kth second pole point and the qth similar point
Gray values of the neighbor pixels to be analyzed,
representing the gray value of the next pixel point in the direction of the v neighbor pixel point to be analyzed in the direction of the s pixel point in the repulsive force field path of the kth second dot and the q similar dot,
represents the maximum gray level in the crop image,
to find the absolute value;
representing all neighbor pixel points to be analyzed of the s-th pixel point in the repulsive force field path of the kth second pole small point and the q-th similar point
Will be the mostThe neighborhood pixel point to be analyzed corresponding to the small value points to the kth second pole small point and the kth second pole small point
The direction of the s-th pixel point in the repulsive force field path of each similar point is taken as a repulsive force vector
Is a direction of (2);
Optionally, the method for obtaining the path flatness of each second minimum point and each similar point according to the resultant field vector includes the following specific steps:
taking any second pole small point and any similar point as an example, acquiring the slope of the space coordinate between two pixel points, and recording the slope as the resultant slope; acquiring the slope of the spatial coordinates of every two adjacent pixel points on a connection path of the two pixel points, namely a repulsive force field path, averaging the slope on the path, and recording the average value as a path slope average value; and taking the absolute value of the difference between the path slope average value and the resultant slope as the path flatness of the two pixel points.
Optionally, the obtaining the final insertion value of the to-be-interpolated position of each second pole small point neighborhood range includes the following specific methods:
obtaining an interpolation value for the position to be interpolated in any second pole small point neighborhood range through a linear interpolation method, and recording the interpolation value as an initial interpolation value of the position to be interpolated; taking the gray value of the reference point of the second pole small point as a reference insertion value of the position to be interpolated; and taking the average value of the initial interpolation value and the reference interpolation value as the final interpolation value of the position to be interpolated, and filling the final interpolation value into the position to be interpolated.
The beneficial effects of the invention are as follows: according to the method, an initial minimum point is calculated on a gap part in a lodging area through a local minimum value, a first minimum point is screened according to the lodging area, a second minimum point is obtained through a guiding degree, a connecting path of the second minimum point and other adjacent second minimum points is obtained through a mode of constructing an artificial potential field model, the same gap part of the two second minimum points is estimated through evaluating the flatness degree of the path, adjacent second minimum points meeting the requirements are used as reference points, and when interpolation is carried out on the adjacent positions of the second minimum points, interpolation results are recalculated to the reference points through values; the new filling value is not influenced by the pixel points with larger gray scale to be smooth, stronger correlation can be established for the pixel points in the same gap, the continuity of the region in the gap region with weaker performance after interpolation is enhanced, the region characteristics are more prominent, the edge information is easier to detect, the lodging direction of crops is accurately identified, and a plurality of parameters in the harvesting process of the harvester are adjusted, so that the grain harvesting rate is improved, and the optimal harvesting result is ensured.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the invention, and that other drawings can be obtained according to these drawings without inventive faculty for a person skilled in the art.
Fig. 1 is a schematic flow chart of a harvester control data processing method based on a 5G technology according to an embodiment of the invention;
FIG. 2 is a schematic diagram of crop lodging area identification;
FIG. 3 is a diagram of the edge detection result of a lodging region before image interpolation processing;
fig. 4 is a schematic diagram of a lodging region edge detection result after image interpolation processing.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. 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.
Referring to fig. 1, a flowchart of a method for processing harvester control data based on 5G technology according to an embodiment of the invention is shown, the method includes the following steps:
and S001, acquiring a crop image, and segmenting to acquire a lodging area.
The purpose of the embodiment is to identify and process lodging crops correspondingly in the working process of the harvester, so that crop images to be harvested are firstly required to be collected, and the crop images are segmented to obtain lodging areas in the crop images; specifically, in the embodiment, a front-mounted visual perception camera carried by a harvester is utilized to shoot crops in front of a travelling path of the harvester, and the obtained image is subjected to gray processing, and the result is recorded as a crop image; the method comprises the steps of identifying and segmenting a lodging area in a crop image through a semantic segmentation network in the prior art, training the semantic segmentation network through a large number of lodging crop image data sets, inputting the acquired crop image into the semantic segmentation network after training, and outputting the segmented lodging area in the crop image, wherein fig. 2 is a schematic diagram for identifying and segmenting the lodging area in the crop image, and the area surrounded by a curve in fig. 2 is the segmented lodging area.
Thus, the crop image is acquired, and the lodging area in the crop image is segmented and acquired.
Step S002, acquiring a plurality of initial minimum points according to gray values in the crop images, acquiring a plurality of first minimum points in the lodging area, and acquiring the guiding degree of each first minimum point according to the gray values of the neighborhood of the first minimum points so as to acquire a plurality of second minimum points.
It should be noted that, the up-sampling algorithm can enhance resolution of any image so as to achieve the purpose of highlighting detailed texture information, the nearest neighbor interpolation method in the traditional up-sampling algorithm brings obvious blockiness, the linear interpolation effectively solves the blockiness problem, the saw teeth are not obvious after interpolation, but the method only considers the influence of gray values of four neighborhood points around the interpolation position, but does not consider the influence of gray value change rate among all the neighborhood points, so that the linear interpolation smoothes the image like a low-pass filter, the linear interpolation of the detail part still causes more blurring, and the edge segmentation of lodging crops under the influence of planting density and illumination environment in the lodging area is still unfavorable;
when up-sampling is carried out by using a bilinear interpolation method, blank grids are inserted into all adjacent rows and columns in an image, namely 0 is filled in positions to be interpolated, and convolution values are filled in each position to be interpolated by convoluting an original image; calculating to-be-interpolated according to 2 x 2 = 4 known values nearest to the to-be-interpolated position, wherein the weight of each known value position is determined by the distance from the to-be-interpolated position, and the closer the distance is, the larger the weight is; bilinear interpolation is to insert a new value in the middle position of each 2 x 2 window in the original image, the middle insertion value is inevitably influenced by the pixel points with larger gray values, and the analysis of crop edges is not improved greatly for crop density;
for the gap part of crops in a lodging area, due to the influence of planting density and illumination environment, the gray value is usually smaller and a communicating domain cannot be integrally formed, but the gap part is important detail information for analyzing the edges of the crops, and the gap part needs to be expanded, namely the gap part is expanded through interpolation so as to accurately analyze the edges of the crops; however, in the traditional linear interpolation method, because the gray value of the pixel points of the crop part around the gap is larger, the interpolation value is influenced by the larger gray value, even if interpolation is completed, the gap part is not expanded, and the boundary between the gap part and the crop part is further blurred;
for the lodging of crops, the local lodging direction consistency is presented, which is caused by environmental weather, and the pixel points of the gap part in a certain direction can be continuously distributed or intermittently distributed but have the same distribution direction for the gap part; therefore, the pixel point with the minimum local gray value can be extracted according to the gray value to obtain a plurality of initial minimum points, the first minimum point in the lodging area is obtained according to the position of the initial minimum point in the image, and the second minimum point belonging to the gap part of the lodging area is judged according to the difference of the gray value of the first minimum point and the neighborhood pixel point and is used for obtaining the reference point of the second minimum point, so that the purpose of expanding the gap part is achieved.
Specifically, a first preset window is adopted to traverse a crop image to obtain a plurality of initial minimum points, wherein the first preset window is set to 3*3, 1 pixel is used as a preset step length, the traversal is started from the upper left corner of the image, an overlapping part exists in each traversal, one row is traversed at the same time, and the next row is traversed longitudinally and also in the preset step length; obtaining the pixel point with the minimum gray value in each first preset window through traversal, and marking all the pixel points with the minimum gray value obtained through traversal as initial minimum points; it should be noted that, the first preset window needs to be as small as possible, meanwhile, the window size of 2×2 is too large in calculation amount, and the window of 3*3 simultaneously meets the eight-neighborhood constitution of the pixel point, so that the embodiment adopts the window with the size of 3*3 for traversing; and (3) marking initial minimum points in the lodging area as first minimum points according to the lodging area acquired in the step S001, and obtaining a plurality of first minimum points.
It should be further noted that, since the difference between the gray average values of the pixels in the eight adjacent areas and the first minimum point are also different, for the first minimum point with a larger difference, it is indicated that the pixel with a smaller gray is surrounded by the pixel with a larger gray, and for the pixel with a smaller gray, the conventional linear interpolation is affected by the surrounding pixel with a larger gray, so that the expansion of the pixel with a smaller gray, that is, the expansion of the gap portion, is not realized, and therefore, a larger guiding degree is needed to continue the subsequent analysis; for the first minimum point with smaller difference, the first minimum point and the eight neighborhood pixel points are indicated to be the same part, but are not different from the crop part and the gap part, and the pixel points do not need to independently process interpolation, namely the guiding degree is smaller, and subsequent analysis processing is not needed.
Specifically, by the first
Taking the first minimum point as an example, the guiding degree is obtained
The calculation method of (1) is as follows:
wherein ,
the gray value representing the ith first minimum point, n represents the number of neighbor pixels of the ith first minimum point, and n=8 here because of the analysis of eight neighbor pixels;
the gray value of the j-th neighborhood pixel point representing the i-th first minimum point,
representing absolute values, th representing a hyperbolic tangent function; it should be noted that, the hyperbolic tangent function can normalize the numerical value, and meanwhile, the function curve of the hyperbolic tangent function on the positive x axis shows that the early period increases rapidly along with the increase of the abscissa, then increases rapidly to a position close to 1 and increases slowly, the difference of the gray average value of the first minimum point and the eight neighborhood pixel points is slightly larger, namely, a larger guiding degree is obtained, and the guiding degree is smaller only when the difference is smaller; the degree of guidance of each first minimum point is obtained according to the method described above.
Further, a first preset threshold is provided for judging the second pole point, and the first preset threshold in this embodiment adopts
Calculating, namely marking a first minimum point with the guiding degree larger than a first preset threshold value as a second minimum point, wherein the second minimum point is a gap part with more discrete distribution in a lodging area, and the gap part needs to be subjected to subsequent analysis and cannot be processed by adopting a conventional linear interpolation method; for all the pixel points except the second small points in the crop image, the conventional bilinear interpolation method is adopted to conduct linear interpolation on the pixel points.
So far, the second small point in the obtained crop image is obtained, and the pixel points outside the second small point are subjected to conventional linear interpolation.
And S003, constructing an artificial potential field model for each second minimum point according to the gray value of the adjacent pixel point, acquiring a reference point of each second minimum point according to the gravitational field and the repulsive field in the artificial potential field model, and interpolating the neighborhood of each second minimum point according to the reference point and the neighborhood pixel point to obtain the processed crop image.
It should be noted that, the second pole points are the gap part pixel points distributed in the lodging area with discrete distribution, and because of the local consistency of the lodging direction, the path judgment can be performed on other second pole points of the second pole points in a certain neighboring range; in the traditional edge detection, strong edge points with larger gradients are connected through non-maximum value inhibition, so that weak edge points similar to second pole small points are ignored, and the actual edges between the weak edge points can be expanded through interpolation in path judgment; therefore, an artificial potential field model needs to be built for each second minimum point, parameters in the artificial potential field model are obtained through the space distance and the gray value, then a connection path is judged through a gravitational field and a repulsive force field between the second minimum point and other second minimum points, other second minimum points of the optimal path are obtained to serve as reference points, and interpolation is completed.
Specifically, a second preset window is set, the second preset window in this embodiment adopts a size of 9*9, each second electrode small point is taken as the center of the second preset window, and a pixel point in the range of the second preset window corresponding to each second electrode small point is taken as an artificial potential field model of each second electrode small point; the artificial potential field model is a traditional artificial potential field method, a search path is controlled through a gravitational field and a repulsive force occasion force direction, and the artificial potential field model is commonly used in path obstacle avoidance planning, and is constructed through gray values and space distances; taking any second minimum point as an example, acquiring a plurality of other second minimum points in the artificial potential field model of the second minimum point, and marking the other second minimum points as similar points of the second minimum points; and obtaining a plurality of similar points of each second pole small point according to the method.
Further, in the first step
Second pole point and first pole point
For example, a similar point is the gravitational field vector between two pixel points
The calculation method of (1) is as follows:
wherein ,
represents the gravitational field value of the kth second pole point and its qth similarity point,
representing the euclidean distance of the kth second pole point from its qth similarity point,
represents the gray value of the kth second pole dot,
gray value of the (q) th similar point representing the (k) th second pole small point, exp [ []Representing a finger with a base of natural constantA function of the number of the data,
represents the gain constant of the gravitational field, which is used in this embodiment
Calculating; the Euclidean distance and the gray level difference between two pixel points are used for calculating the Euclidean norm, so that the gravitational field is quantized, and the smaller the Euclidean distance is, the smaller the gray level difference is, the larger the gravitational force between the two pixel points is, and therefore the gravitational field is obtained by
To present the inverse proportion relation and simultaneously utilize for the subsequent unification of dimension
Normalization processing is carried out, and an implementer can select an inverse proportion function and a normalization function according to actual conditions; the direction of the kth second pole point to the qth similar point is recorded as the gravitational field value
The direction of the (b) is that the gravity field vector of the kth second pole small point and the q-th similar point is obtained and expressed as
The method comprises the steps of carrying out a first treatment on the surface of the Obtaining the first according to the method
The gravitational field vector of the second pole and each of its similarity points.
Further, in the first step
Second pole point and first pole point
For example, a repulsive force field vector between two pixel points
The calculation method of (1) is as follows:
wherein ,
representing the repulsive force vector of the s-th pixel point in the repulsive force field path of the kth second pole small point and the q-th similar point,
representing the repulsive force value of the s-th pixel point in the repulsive force field path of the kth second pole small point and the q-th similar point,
representing the number of neighbor pixels to be analyzed of the s-th pixel in the repulsive force field path of the kth second pole small point and the q-th similar point,
representing the gray value of the s-th pixel point in the repulsive force field path of the kth second pole small point and the q-th similar point,
representing the kth pixel point in the repulsive force field path of the kth second pole point and the qth similar point
Gray values of the neighbor pixels to be analyzed,
representing the gray value of the next pixel point in the direction of the v neighbor pixel point to be analyzed in the direction of the s pixel point in the repulsive force field path of the kth second dot and the q similar dot,
representing gray maxima in crop images，
To find the absolute value;
representing all neighbor pixel points to be analyzed of the s-th pixel point in the repulsive force field path of the kth second pole small point and the q-th similar point
The minimum value of the (B) points the neighborhood pixel point to be analyzed corresponding to the minimum value to the kth second electrode small point and the kth second electrode small point
The direction of the s-th pixel point in the repulsive force field path of each similar point is taken as a repulsive force vector
Is a direction of (2);
representing the number of pixel points in the repulsive force field path of the kth second pole dot and its qth similar dot,
the gain constant representing the repulsive field is used in this embodiment
Calculating; note that, the number of pixel points in the repulsive force field path
Excluding the kth second minimum point and its qth similarity point, when
In the time-course of which the first and second contact surfaces,
representing the repulsive force vector of the first pixel point pointing to the kth second pole point in the repulsive force field path.
Each pixel point on the repulsive force field path and a certain neighborhood pixel point thereof, and the pixel point points to the next pixel point on the neighborhood pixel point direction to form a connecting line of three pixel points, the gradient change rate is quantified through the gray level difference of two adjacent pixel points on the connecting line, the larger the gradient change rate is, the larger the break-through cost in the direction is indicated, the larger the corresponding repulsive force is, the smaller the repulsive force obtained by the repulsive force field path is, the direction with the minimum break-through cost is required to be selected to form the repulsive force field path, namely, the pixel point with the minimum break-through cost in eight neighborhood eight directions is taken as the next pixel point of the pixel point on the repulsive force field path; for the next pixel point of each pixel point on the repulsive force field path, the larger the gray value of the next pixel point is, the larger the breakthrough cost is, the further limit on the gradient change rate is avoided when the repulsive force field path is constructed, the breakthrough cost is reduced when the next pixel point is avoided, and the corresponding repulsive force is smaller; and obtaining the repulsive force field vector of the kth second pole small point and each similar point thereof according to the method.
Further, taking the kth second pole point and the qth similar point as examples, the resultant force field vector between two pixel points
The calculation method of (1) is as follows:
wherein ,
represent the first
The gravitational field vector of the second dot and the q-th similar dot,
represent the first
Second polesA repulsive force field vector of the small point and the q-th similar point; the vector sum of the gravitational field vector and the repulsive force field vector is used for obtaining a resultant force field vector between two pixel points, and meanwhile, the repulsive force field path is used as a connecting path of the two pixel points; acquiring a force field vector and a corresponding connection path of the kth second pole small point and each similar point thereof according to the method; it should be noted that, in this embodiment, the gravitational field and the repulsive field are not attraction and repulsion in the physical model, but are only indexes for feature analysis constructed according to the artificial potential field model.
Further, taking the kth second pole small point and the qth similar point as examples, acquiring the slope of the space coordinate between the two pixel points, and recording the slope as the resultant slope; acquiring the slope of the space coordinates of every two adjacent pixel points on the connecting path, calculating the average value of the slope on the path, and recording the average value as the path slope average value; taking the absolute value of the difference between the path slope mean value and the resultant slope as the path flatness of the kth second pole small point and the q-th similar point; obtaining the path flatness of the kth second electrode small point and each similar point according to the method, taking the similar point corresponding to the minimum value of the path flatness as the reference point of the kth second electrode small point, and marking the corresponding connection path as the optimal path of the kth second electrode small point; the smaller the slope change of the connecting path is, the flatter the connecting path is, the greater the possibility that two pixel points are intermittently distributed in the same direction is, the greater the possibility that the two pixel points belong to the same gap part is, and the greater the consultability of interpolation calculation is; and acquiring the reference point of each second minimum point according to the method.
Further, interpolation is carried out on the four-neighborhood range of the second pole small point, and for any position to be interpolated, an interpolation value is obtained through a conventional linear interpolation method and is recorded as an initial interpolation value of the position to be interpolated; taking the gray value of the reference point of the second pole small point as a reference insertion value of the position to be interpolated; taking the average value of the initial interpolation value and the reference interpolation value as the final interpolation value of the position to be interpolated, and filling the final interpolation value into the position to be interpolated; finishing interpolation on the four-neighborhood range of all the second pole small points according to the method, obtaining a crop image with the interpolation finished, and recording the crop image as a processed crop image; it should be noted that, the final insertion value is obtained by the average value of the initial insertion value and the reference insertion value, that is, the initial insertion value and the reference insertion value are considered to be as important as the position to be interpolated in the embodiment, that is, the weights are all 0.5, and the implementer can adjust the weights of the two insertion values according to the actual situation.
Thus, an image which is obtained by interpolating all pixel points in the crop image, namely the processed crop image, is obtained.
And S004, judging the lodging condition of the crops according to the processed crop images, and controlling a reel of the harvester.
Performing edge detection on the crop image and the processed crop image; referring to fig. 3, an edge detection result of a crop image lodging area before interpolation processing is shown; referring to fig. 4, an edge detection result of a crop image lodging area after interpolation processing is shown; more edge information can be obtained from fig. 4 than from fig. 3, and although the up-sampling interpolation can lead to unavoidable occurrence of many smaller saw teeth in the edge information, the segmentation of crops in the crop lodging area and judgment of lodging direction are not affected.
Judging the lodging direction of crops according to the direction of the edge line on the edge detection result, which is the prior art, and the embodiment is not described in detail; the harvester's reel is adjusted according to the lodging direction of the crop, for example, the reel needs to reel in the reverse direction of the lodging direction of the crop, if the lodging angle is lower, the height of the reel needs to be reduced as well, and meanwhile, the harvester needs to adjust many control parameters such as the height and angle of the header, the running speed of the harvester, the rotating wheel speed, etc. for the lodging crop harvesting, which are the prior art adjustment of lodging crop harvesting, and this embodiment will not be repeated.
Therefore, the edge segmentation of the lodging crops is more accurate by carrying out interpolation processing on the crop images and carrying out special interpolation processing on the crop gap part of the lodging area, and the working efficiency of harvesting the crops by the harvester is further improved.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (5)
1. The harvester control data processing method based on the 5G technology is characterized by comprising the following steps of:
acquiring a crop image and acquiring a lodging area in the crop image;
acquiring a plurality of initial minimum points according to gray values in the crop image, marking the initial minimum points in the lodging area as first minimum points, acquiring the guiding degree of each first minimum point according to the gray values of the neighborhood of the first minimum points, judging and acquiring a plurality of second minimum points according to the guiding degree through a first preset threshold value, and performing linear interpolation on pixel points except the second minimum points in the crop image;
constructing an artificial potential field model for each second minimum point according to the neighboring pixel points and the gray values thereof, acquiring other second minimum points in the artificial potential field model of each second minimum point, marking the other second minimum points as similar points of each second minimum point, and acquiring gravitational field vectors of each second minimum point and each similar point according to the space coordinates and the gray values of each second minimum point and each similar point; according to the space coordinates, gray values and artificial potential field models of each second minimum point and each similar point, repulsive force field vectors of each second minimum point and each similar point are obtained;
taking the sum of vectors of the gravitational field vector and the repulsive force field vector as a combined force field of each second minimum point and each similar point, acquiring the path flatness of each second minimum point and each similar point according to the combined force field vector, taking the similar point corresponding to the minimum value in the path flatness of each second minimum point as a reference point of each second minimum point, and acquiring the final interpolation value of the position to be interpolated in the neighborhood range of each second minimum point according to the gray value of the neighborhood pixel point of each second minimum point and the gray value of the reference point, so as to finish the interpolation of all pixel points in the crop image and obtain the processed crop image;
judging the lodging condition of crops according to the processed crop images, and controlling a reel of the harvester;
the method comprises the following steps of: traversing crop images by a first preset window at a preset step length from the upper left corner of the images, traversing one row, and traversing the next row longitudinally by the preset step length; obtaining the pixel point with the minimum gray value in each first preset window through traversal, and marking all the pixel points with the minimum gray value obtained through traversal as initial minimum points;
the guiding degree of each first minimum point is obtained according to the gray value of the neighborhood of the first minimum point, and the guiding degree is specifically:
wherein ,
indicating the degree of guidance of the i first minimum point,/->
The gray value representing the ith first minimum point, n represents the number of neighbor pixels of the ith first minimum point, and n=8 here because of the analysis of eight neighbor pixels; />
Gray value of the jth neighborhood pixel representing the ith first minimum point,/>
Representing absolute values, th representing a hyperbolic tangent function;
the step of obtaining the final insertion value of the position to be interpolated of each second pole small point neighborhood range comprises the following steps: obtaining an interpolation value for the position to be interpolated in any second pole small point neighborhood range through a linear interpolation method, and recording the interpolation value as an initial interpolation value of the position to be interpolated; taking the gray value of the reference point of the second pole small point as a reference insertion value of the position to be interpolated; and taking the average value of the initial interpolation value and the reference interpolation value as the final interpolation value of the position to be interpolated, and filling the final interpolation value into the position to be interpolated.
2. The harvester control data processing method based on the 5G technology according to claim 1, wherein the constructing an artificial potential field model for each second minimum point according to the neighboring pixel point and the gray value thereof comprises the following specific steps:
setting a second preset window, taking each second electrode small point as the center of the second preset window, and taking the pixel point in the range of the second preset window corresponding to each second electrode small point as an artificial potential field model of each second electrode small point.
3. The method for processing harvester control data based on 5G technology according to claim 1, wherein the obtaining the gravitational field vector of each second minimum point and each similar point thereof comprises the following specific steps:
wherein ,
gravitational field value representing the kth second dot and its qth similarity, +.>
Representing the Euclidean distance of the kth second pole point from its qth similarity point,>
gray value representing kth second pole dot, is->
Gray value of the (q) th similar point representing the (k) th second pole small point, exp [ []Expressed as natural constantExponential function of the base>
A gain constant representing the gravitational field;
the direction of the kth second pole point to the qth similar point is recorded as the gravitational field value
The direction of the (d) is then given by the gravitational field vector of the kth second pole point and its qth similarity point, expressed as +.>
。
4. The method for processing the control data of the harvester based on the 5G technology according to claim 1, wherein the specific method for obtaining the repulsive force field vector of each second minimum point and each similar point thereof comprises the following steps:
wherein ,
a repulsive force vector representing the s-th pixel point in the repulsive force field path of the kth second pole small point and the q-th similar point thereof,/for the second pole small point>
Representing the repulsive force value of the s-th pixel point in the repulsive force field path of the kth second pole small point and the q-th similar point thereof,/for the second pole small point>
A repulsive force field representing the kth second pole point and the qth similar pointThe number of neighbor pixels to be analyzed of the s-th pixel in the path, < +.>
Gray value of the s-th pixel point in the repulsive force field path representing the kth second dot and its qth similar dot,/th>
A (th) pixel point in a repulsive force field path representing a kth second pole small point and a (th) similar point thereof>
Gray values of the pixels in the neighborhood to be analyzed, < >>
Representing the gray value of the next pixel point in the direction of the v neighbor pixel point to be analyzed in the direction of the s pixel point in the repulsive force field path of the kth second dot and the q similar dot,
representing the gray maximum in the crop image, +.>
To find the absolute value; />
Representing all neighbor pixel points to be analyzed of the s-th pixel point in the repulsive force field path of the kth second pole small point and the q-th similar point
The minimum value of the (2) points the neighborhood pixel point to be analyzed corresponding to the minimum value to the kth second electrode small point and the kth +.>
The direction of the s-th pixel point in the repulsive force field path of each similar point is taken as a repulsive force vector
Is a direction of (2);
5. The method for processing harvester control data based on 5G technology according to claim 4, wherein the obtaining the path flatness of each second minimum point and each similar point according to the combined force field vector comprises the following specific steps:
taking any second pole small point and any similar point as an example, acquiring the slope of the space coordinate between two pixel points, and recording the slope as the resultant slope; acquiring the slope of the spatial coordinates of every two adjacent pixel points on a connection path of the two pixel points, namely a repulsive force field path, averaging the slope on the path, and recording the average value as a path slope average value; and taking the absolute value of the difference between the path slope average value and the resultant slope as the path flatness of the two pixel points.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310363307.7A CN116109658B (en) | 2023-04-07 | 2023-04-07 | Harvester control data processing method based on 5G technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310363307.7A CN116109658B (en) | 2023-04-07 | 2023-04-07 | Harvester control data processing method based on 5G technology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116109658A CN116109658A (en) | 2023-05-12 |
| CN116109658B true CN116109658B (en) | 2023-06-20 |
Family
ID=86260050
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310363307.7A Active CN116109658B (en) | 2023-04-07 | 2023-04-07 | Harvester control data processing method based on 5G technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116109658B (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112650237A (en) * | 2020-12-21 | 2021-04-13 | 武汉理工大学 | Ship path planning method and device based on clustering processing and artificial potential field |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8233705B2 (en) * | 2008-10-03 | 2012-07-31 | Eastman Kodak Company | Potential field-based gamut mapping |
| CN107065866A (en) * | 2017-03-24 | 2017-08-18 | 北京工业大学 | A kind of Mobile Robotics Navigation method based on improvement optical flow algorithm |
| CN111123913B (en) * | 2019-11-29 | 2023-05-16 | 广东工业大学 | Double-layer path planning and collaborative obstacle avoidance method for multiple unmanned boats |
| CN113960996A (en) * | 2020-07-20 | 2022-01-21 | 华为技术有限公司 | Planning method and device for obstacle avoidance path of driving device |
| CN112677153B (en) * | 2020-12-16 | 2022-09-06 | 东北林业大学 | Improved RRT algorithm and industrial robot path obstacle avoidance planning method |
| CN113050652B (en) * | 2021-03-25 | 2022-08-23 | 上海海事大学 | Trajectory planning method for automatic berthing of intelligent ship |
| CN114815831A (en) * | 2022-04-28 | 2022-07-29 | 西安交通大学 | Robot path planning method based on improved potential field grid |
| CN115167465B (en) * | 2022-08-10 | 2023-04-28 | 哈尔滨理工大学 | Unmanned submersible vehicle three-dimensional path planning method based on artificial potential field grid method |
| CN115409833B (en) * | 2022-10-28 | 2023-01-31 | 一道新能源科技(衢州)有限公司 | Hot spot defect detection method of photovoltaic panel based on unsharp mask algorithm |
| CN115861135B (en) * | 2023-03-01 | 2023-05-23 | 铜牛能源科技(山东)有限公司 | Image enhancement and recognition method applied to panoramic detection of box body |
-
2023
- 2023-04-07 CN CN202310363307.7A patent/CN116109658B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112650237A (en) * | 2020-12-21 | 2021-04-13 | 武汉理工大学 | Ship path planning method and device based on clustering processing and artificial potential field |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116109658A (en) | 2023-05-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116310360B (en) | Reactor surface defect detection method | |
| CN111798462B (en) | Automatic delineation method of nasopharyngeal carcinoma radiotherapy target area based on CT image | |
| CN106327507B (en) | A kind of color image conspicuousness detection method based on background and foreground information | |
| CN115908371B (en) | Plant leaf disease and pest degree detection method based on optimized segmentation | |
| CN107730536B (en) | High-speed correlation filtering object tracking method based on depth features | |
| CN109035289B (en) | Purple soil image segmentation and extraction method based on Chebyshev inequality H threshold | |
| CN106157330B (en) | Visual tracking method based on target joint appearance model | |
| US20060274946A1 (en) | Method and arrangement for determining an object contour | |
| CN110443775B (en) | Discrete wavelet transform domain multi-focus image fusion method based on convolutional neural network | |
| CN114548265B (en) | Crop leaf disease image generation model training method, crop leaf disease identification method, electronic equipment and storage medium | |
| CN111199245A (en) | Rape pest identification method | |
| CN109086823B (en) | Automatic statistical method for wheat scab ear disease rate | |
| CN110211117B (en) | Processing system for identifying linear tubular objects in medical image and optimized segmentation method | |
| CN114724022A (en) | Culture fish school detection method, system and medium fusing SKNet and YOLOv5 | |
| CN114842337A (en) | Fruit picking point identification method based on deep learning and multidimensional information fusion clustering | |
| CN106529441A (en) | Fuzzy boundary fragmentation-based depth motion map human body action recognition method | |
| CN115439493A (en) | Method and device for segmenting cancerous region of breast tissue section | |
| CN114120359A (en) | Method for measuring body size of group-fed pigs based on stacked hourglass network | |
| CN117952960A (en) | Electric power iron tower part defect detection method based on artificial intelligence | |
| CN116109658B (en) | Harvester control data processing method based on 5G technology | |
| CN116486273B (en) | Method for extracting water body information of small sample remote sensing image | |
| CN114782455B (en) | Cotton row center line image extraction method for agricultural machine embedded equipment | |
| CN116740337A (en) | Safflower picking point identification positioning method and safflower picking system | |
| CN117132802A (en) | Method, device and storage medium for identifying field wheat diseases and insect pests | |
| CN115861768A (en) | Honeysuckle target detection and picking point positioning method based on improved YOLOv5 |
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 |