CN108960100A - A kind of recognition methods of the sugarcane sugarcane section based on image procossing - Google Patents

Abstract

The invention discloses a method for identifying sugarcane knots based on image processing, which includes: first, converting the original image into a grayscale image and performing median filtering and binarization; second, black and white the image obtained in step one Two-color replacement to remove the part of the nonlinear structural element; third, the background of the grayscale image in step 1 is replaced with black, and the sugarcane is replaced with white; fourth, the image obtained in step 3 is subjected to Sobel boundary detection to extract obvious boundary features; 5. Use the Hough transform to detect the straight line on the image obtained in step 4, select the longest straight line, and calculate the angle α with the horizontal direction; 6. Rotate the image obtained in step 2 according to the angle α, and then perform column pixel summation , the abscissa of the extreme point and the maximum point of the summation function is the position d of the stem node; Seventh, the position d ₁ of the stem node in the original image is obtained through conversion. The invention can accurately identify sugarcane nodes, provide accurate position signals for subsequent cutting, and prevent damage to buds and waste.

Description

A recognition method of sugarcane knots based on image processing

technical field

The invention relates to the field of sugarcane industry, in particular to a method for identifying sugarcane knots based on image processing.

Background technique

Sugarcane is an important raw material for sugar production in my country, and it plays an important role in the field of sugar production in China. In the process of sugarcane industry development, automated production methods will greatly save labor costs and improve work efficiency. Cane nodes containing cane buds can be used as sugarcane seeds. During the preparation of sugarcane seeds, the buds should not be damaged. The cane stems at both ends of the cane nodes are the main source of nutrition for the initial survival of the sugarcane seeds, and sufficient length of the cane stems must be left , so in the preparation process of sugarcane seeds, the accuracy of cutting is very important. Through the recognition technology based on image processing, the exact position of sugarcane stem nodes can be extracted, thereby providing signals for cutting work, so that the process of sugarcane seed preparation can realize automatic production.

At present, domestic research on sugarcane knot recognition technology based on image processing is not sufficient, and there are still many aspects that have not been considered, and there is still a big gap with industrial production. Moreover, the existing methods for stem node recognition through image processing often only involve the recognition of a single stem node of sugarcane, and few target features are extracted. In the process of industrial production, it is necessary to consider the influence of complex features on the surface of sugarcane, extract multiple target features, and form a set of signal sources. Moreover, the sugarcane raw materials used in industrial production are not straight, and there are often a large number of distorted raw materials, which easily lead to inaccurate identification of the position of cane nodes, and are likely to cause damage to buds in subsequent cutting and separation of cane stems and cane seeds. , wasting sugarcane seeds.

The information disclosed in this Background section is only for enhancing the understanding of the general background of the present invention and should not be taken as an acknowledgment or any form of suggestion that the information constitutes the prior art that is already known to those skilled in the art.

Contents of the invention

The object of the present invention is to provide a kind of identification method of sugarcane node based on image processing, thereby overcome the existing method for identifying node by image processing only involves the identification of a single node and the position of the identified node Easily inaccurate shortcomings.

To achieve the above object, the present invention provides a method for identifying sugarcane knots based on image processing, which includes the following steps:

Step 1, first collect the original image, then use matlab software to convert the original image into a grayscale image, carry out median filtering on the grayscale image, and then perform binarization processing;

Step 2, replace the black and white two colors in the image processed by step 1, and then use the open operation in the morphological processing to remove the part of the nonlinear structural element;

Step 3, replace the sugarcane background of the grayscale image obtained in step 1 with black, replace the position of sugarcane with white, and perform binarization;

Step 4, performing Sobel boundary detection on the image processed in step 3 to extract obvious boundary features;

Step 5, using Hough transform to detect straight lines on the image processed in step 4, selecting the longest straight line, and calculating the angle α between the longest straight line and the horizontal direction of the image;

Step 6: Use the included angle α obtained in Step 5 to rotate the processed image in Step 2, so that the nodes of the image are on the same horizontal line, and then perform column pixel summation on the image to find the extreme point of the summation function and the maximum point, the abscissa of the extreme point and the maximum point is the position d of the sugarcane stem node;

In step seven, according to the position d of the sugarcane stem node obtained after the processing in step six, the position d ₁ of the sugarcane stem node in the original image can be obtained through the conversion formula d ₁ =d xcosα.

Preferably, in the above technical solution, in step 1, during the image binarization process, the average value of the maximum pixel value and the minimum pixel value in the image is set as the threshold value, and the pixel points whose grayscale is greater than or equal to the threshold value are set is 255, and the pixels smaller than the threshold are set to 0. In order to save computer storage, 0 and 255 are further mapped to 0 and 1 respectively.

Preferably, in the above technical solution, in step 2, since the stem nodes of sugarcane are in the shape of a straight line, in order to recognize and retain this feature, a flat straight line structure with a length of 15° and 90° counterclockwise from the horizontal axis is adopted Yuan, the formula is: Among them: B is the structural element, A is the processed image, and the result of the opening operation is the union of B completely matched and translated in A.

Preferably, in the above technical solution, in step 4, the Sobel boundary detection is to use two (3x3) matrices to perform convolution operation on the original image to calculate the estimated value G _x of the gray partial derivative in the horizontal direction and the vertical The estimated value G _y of the gray partial derivative in the vertical direction. The mathematical expressions of G _x and G _y are as follows:

and

Among them: A is the processed image;

The calculation process is as follows:

G _x ＝[f(x+1,y-1)+2*f(x+1,y)+f(x+1,y+1)]-[f(x-1,y-1)+ 2*f(x-1,y)+f(x-1,y+1)];

G _y =[f(x-1,y-1)+2f(x,y-1)+f(x+1,y-1)]-[f(x-1,y+1)+2* f(x,y+1)+f(x+1,y+1);

After calculating the G _x and G _y of each point, the gradient G of each point can be obtained by the following formula:

Finally, a threshold G _max is set. When the gradient G is greater than G _max , the point is a boundary point.

Preferably, in the above technical solution, in step five, the Hough transform is to transform the straight line in the image plane space to the parameter space, and its formula is:

Any point (x, y) on the straight line can form a curve in the ρ-θ space. If the curves obtained by performing the above operations on two different points intersect in the plane ρ-θ, it means that they are on the same straight line. By setting the threshold, the number of curves intersecting at one point exceeds the threshold, then the parameter (θ, ρ) representing the intersection point corresponds to a straight line in the original image; find the first 5 points that are greater than 0.3 times the maximum value in the Hough matrix, Convert to a line segment, merge the line segments whose distance is less than 20, remove the straight line segment whose length is less than 5, and obtain the longest line segment through the for loop; obtain the slope k of the line through the coordinate value of the end point of the line segment, and then use the formula α=arctank to find The angle α between the longest straight line and the horizontal direction of the image is obtained.

Preferably, in the above-mentioned technical solution, in step six, utilize the imrotate function in the matlab software to rotate the image, utilize the sum summation function, sum each row of pixels in the image, and draw the function image, utilize the findpeaks function and The max function finds the extreme point and the maximum point, and the abscissa of the extreme point and the maximum point is the position d of the sugarcane stem node.

Compared with the prior art, the present invention has the following beneficial effects:

The present invention provides a technical reference for the industrial preparation of domestic sugarcane seeds. Its core is to use the image processing technology based on matlab software to extract multiple objects on the surface of sugarcane to accurately identify the nodes of sugarcane, effectively avoiding missed detection and Error detection, thereby providing accurate position signals for subsequent cutting and separation of cane stems and cane seeds, preventing damage to buds and waste of cane seeds during the cutting process, and the present invention can also provide correction references for feeding inclinations.

Description of drawings

Fig. 1 is a flowchart of a method for identifying sugarcane knots based on image processing according to the present invention.

Fig. 2 is a schematic diagram of converting the original image into a grayscale image in step 1 according to the present invention.

Fig. 3 is a schematic diagram of an image after median filtering in step 1 according to the present invention.

FIG. 4 is a schematic diagram of an image after binarization processing in Step 1 according to the present invention.

Fig. 5 is a schematic diagram of an image after black-and-white two-color replacement in step 2 according to the present invention.

Fig. 6 is a schematic diagram of an image after performing an opening operation in step 2 according to the present invention.

Fig. 7 is a schematic diagram of an image after setting the sugarcane position to white and the sugarcane background to black in step 3 according to the present invention.

FIG. 8 is a schematic diagram of an image with obvious boundary features after Sobel boundary detection in Step 4 of the present invention.

FIG. 9 is a schematic diagram of an image of the longest straight line detected by Hough transform in Step 5 according to the present invention.

Fig. 10 is a schematic diagram of an image after rotating Fig. 6 in step 6 according to the present invention.

Fig. 11 is a statistical diagram of the column pixel summation of the rotated image in step 6 according to the present invention.

Fig. 12 is a schematic diagram of an image in which the positions of identified sugarcane nodes are marked in the original image in step 7 according to the present invention.

Detailed ways

The specific embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings, but it should be understood that the protection scope of the present invention is not limited by the specific embodiments.

Unless expressly stated otherwise, throughout the specification and claims, the term "comprise" or variations thereof such as "includes" or "includes" and the like will be understood to include the stated elements or constituents, and not Other elements or other components are not excluded.

Fig. 1 to Fig. 12 have shown the structure diagram of a kind of identification method of sugarcane section based on image processing according to the preferred embodiment of the present invention, with reference to Fig. 1, this identification method of sugarcane section based on image processing comprises following seven altogether step:

In step 1, an original image is first collected, which may be obtained by shooting with a camera, and the original image is a color image. Then use matlab software to convert the original image into a grayscale image, and carry out median filtering on the grayscale image, and then perform binarization processing.

Each pixel of a color image (i.e., an RGB image) can be regarded as a point in a three-dimensional space in a space Cartesian coordinate system with R, G, and B as axes, while each pixel of a grayscale image (i.e., a Gray image) can be It is represented by a point on the straight line R=G=B. Therefore, the essence of converting an RGB image to a Gray image is to find a mapping from a three-dimensional space to a one-dimensional space. In the matlab software, it can be realized by making a vertical line from a point in the RGB space to the R=G=B line. The formula is:

Gray＝0.299*R+0.587*G+0.114*B; (1)

The image after converting the original image to a grayscale image is shown in Figure 2.

When sampling or transmitting digital images, they are often disturbed by noise. For subsequent image operations, the noisy image can be filtered. Median filtering is to use the value of the image pixel point by the value of each point in a neighborhood of the point. The median is replaced, and the neighborhood in the processing of the median filter is the template M of mxn, and the formula is:

g(x,y)=med{f(x-m,y-n)},(m,n∈M); (2)

Among them: (x, y) is the coordinate of the point to be processed, m is the length of the processing range from the point (x, y), n is the width of the processing range from the point (x, y), and (m, n) is the distance Point (x, y) is a field of length m and width n. The image after median filtering is shown in Figure 3.

Image binarization processing is to set the pixel points of the image to 0 or 255, so that the entire image presents an obvious black and white effect, highlighting the features in the image. Preferably in the present invention, in step 1, during the image binarization process, the average value of the maximum pixel value and the minimum pixel value in the image is set as the threshold value, and the pixel points whose gray scale is greater than or equal to the threshold value are set to 255, Pixels smaller than the threshold are set to 0. In order to save computer storage, 0 and 255 are further mapped to 0 and 1, respectively. The image after binary processing is shown in Figure 4.

Step two, replace the black and white colors in the image processed by step one, and then use the open operation in the morphological processing to remove the part of the nonlinear structural element. The specific operation is to first find the pixel point with a pixel value of 1, change its pixel value to 0, find the pixel point with a pixel value of 0, and change its pixel value to 1, so as to realize the black and white two colors in the binarized image Replacement, the image after replacement is shown in Figure 5.

Preferably, in the present invention, in step 2, since the stem nodes of sugarcane are in the shape of a straight line, in order to identify and retain this feature, a flat linear structural element with a length of 15 and 90° counterclockwise from the horizontal axis is used, the formula for:

Among them: B is the structural element, A is the processed image, and the result of the opening operation is the union of B completely matched and translated in A. The opening operation removes all the parts that cannot contain structural elements in the image, smoothes the contour of the target, breaks the small connection parts between the images, and removes the sharp protruding parts at the same time. The image after the opening operation is shown in Figure 6.

Step 3: Replace the sugarcane background of the grayscale image obtained in step 1 without median filtering and binarization processing with black, and replace the position of sugarcane with white, and perform binarization processing. The specific operation is: since the sugarcane in the original image uses a white background, in the digital image whose data type is uint8, the pixel value of the background part is 255 or close to 255, use the find function in the matlab software to find out the pixel value of the background part Pixel, and set the pixel value to 0, and the background part will appear black at this time. Similarly, if the pixel value of the sugarcane part (that is, the non-background part) is set to 255, the sugarcane part will appear white. The image after replacement is shown in Figure 7.

Step 4, carry out Sobel boundary detection to the image after step 3 processing, extract obvious boundary feature; The present invention preferably, in step 4, Sobel operator detection boundary is to use two (3x3) matrices to carry out original picture Convolution operation to calculate the estimated value G _x of the gray partial derivative in the horizontal direction and the estimated value G _y of the gray partial derivative in the vertical direction. The mathematical expressions of G _x and G _y are as follows:

Among them: A is the processed image;

The calculation process is as follows:

G _x ＝[f(x+1,y-1)+2*f(x+1,y)+f(x+1,y+1)]-[f(x-1,y-1)+ 2*f(x-1,y)+f(x-1,y+1)]; (6)

G _y =[f(x-1,y-1)+2*f(x,y-1)+f(x+1,y-1)]-[f(x-1,y+1)+ 2*f(x,y+1)+f(x+1,y+1);(7)

After calculating the G _x and G _y of each point, the gradient G of each point can be obtained by the following formula:

Finally, a threshold G _max is set. When the gradient G is greater than G _max , the point is a boundary point. The image obtained after extracting obvious boundary features is shown in Figure 8.

Step 5: Use the Hough transform to detect straight lines on the image processed in step 4, select the longest straight line, and calculate the angle α between the longest straight line and the horizontal direction of the image. Preferably in the present invention, the Hough transform is to transform the straight line in the image plane space to the parameter space, and its formula is:

Any point (x, y) on the straight line can form a curve in the ρ-θ space. If the curves obtained by performing the above operations on two different points intersect in the plane ρ-θ, it means that they are on the same straight line. By setting the threshold, the number of curves intersecting at one point exceeds the threshold, then the parameter (θ, ρ) representing the intersection point corresponds to a straight line in the original image; find the first 5 points that are greater than 0.3 times the maximum value in the Hough matrix, Convert to a line segment, merge the line segments with a distance of less than 20, remove the line segment with a length of less than 5, and obtain the longest line segment through a for loop. The position of the longest line in the grayscale image is shown in Figure 9; The slope k of the straight line is obtained from the coordinate value, and then the angle α between the longest straight line and the horizontal direction of the image can be obtained through the formula α=arctank.

Step six, the included angle α obtained in step five is the angle between the longest straight line of the sugarcane in the image and the horizontal direction of the image, which is the inclination of the sugarcane relative to the workbench. Use the included angle α obtained in step 5 to rotate the image shown in Figure 6 after the opening operation in step 2, so that the sugarcane nodes of the image are located on the same horizontal straight line, and the rotated image is shown in Figure 10. Then, sum the column pixels of the sugarcane image to find the extreme point and a maximum point of the summation function. The abscissa of the extreme point and the maximum point is the position d of the sugarcane stem node. Preferably in the present invention, in step 6, utilize the imrotate function in the matlab software to rotate the image after step 2 is opened and processed, utilize the sum summation function to sum each row of pixels in the image, and draw the function image, Figure 11 shows the statistical diagram for the column pixel summation. Use the findpeaks function and the max function to find the extreme point and the maximum point, and the abscissa of the extreme point and the maximum point is the position d of the sugarcane stem node. As shown in Figure 10, the part of the sugarcane node is white, and the value of all pixels in this part is 1, and the rest is black, and the value of the pixel is 0. Therefore, when the column pixels are added, the pixel values of the white part are accumulated to obtain The maximum value or the maximum value, the cumulative value of the rest of the pixel values is still 0, so the abscissa of the extreme point and the maximum point is the position d of the sugarcane stem node, and there are multiple extreme points and maximum points in one image. It means that there are multiple sugarcane stem nodes.

Step 7, according to the position d of the sugarcane stem node obtained after the processing of step 6, calculate by the following conversion formula:

d ₁ = dx cos α; (10)

Then the position d ₁ of the sugarcane stem node in the original image can be obtained. According to the obtained cane node position d ₁ , it is marked in the original image as shown in Fig. 12 .

Below in conjunction with the concrete operation of matlab software, the present invention will be further described:

Firstly, grayscale transformation is performed on the collected image, and then median filtering is performed on the grayscale image to eliminate the noise in the image, and then the image is binarized to convert it into a black and white two-color image.

A＝imread('ganzhe.jpg');% read in the original image

figure(1),imshow(A);

B=rgb2gray(A);% converted to grayscale image

figure(2),imshow(B);% image as shown in Figure 2

C=medfilt2(B,[3,3]);% uses 3X3 matrix to carry out median filtering to the grayscale image, figure(3),imshow(C);% image as shown in Figure 3

Imax=max(max(C));

Imin=min(min(C));

T=round(Imax-(Imax-Imin)/2); % select the median value of the pixel as the threshold

D=(C)>=T; in %C, the pixel greater than T takes 1, and the pixel smaller than T takes 0

figure(4), imshow(D);% image as shown in Figure 4

Second, the black and white colors in the binary image are replaced, and then the linear structural element with a length of 15 and 90 degrees to the horizontal direction is used for opening operation to remove the part of the non-linear structural element in the image.

index1=find(D>0);% Find the white pixels

index2=find(D<1);% find black pixels

counter1=sum(index1);% the number of white pixels

counter2=sum(index2);% the number of black pixels

figure(5), imshow(E);% image as shown in Figure 5

se=strel('line',15,90); % choose a straight-line structural element, the length is 15, and the horizontal angle is 90 degrees

O＝imopen(E,se); % perform morphological open operation

figure(6), imshow(O);% image as shown in Figure 6

Third, replace the background of sugarcane in the grayscale image with black, replace the position of sugarcane with white, and then perform binarization.

index = find(B>250);

F(index)=0;

G=F;

index1 = find(F>0);

G(index1)=255;

Imax1=max(max(G));

Imin1=min(min(G));

T1=round((Imax1-Imin1)/2);

H1=(G)>=T1;

figure(7),imshow(H1);% image as shown in Figure 7

Fourth, the above images are subjected to Sobel boundary detection to extract images with obvious boundary features.

R1=edge(H1,'Sobel');

figure(8), imshow(R1);% image as shown in Figure 8

Fifth, use the Hough transform to detect and select the longest straight line, and calculate the angle between the longest straight line and the horizontal direction of the image.

[H, Theta, Rho] = hough(R1);

P=houghpeaks(H,5,'threshold',ceil(0.3*max(H(:))));% Find the first 5 peaks in the Hough matrix that are greater than 0.3 times the maximum value in the Hough matrix

lines=houghlines(R1,Theta,Rho,P,'FillGap',20,'MinLength',5);% Merge line segments whose distance is less than 20, and discard all straight line segments whose length is less than 5

end% determines the longest line segment

figure(9), imshow(A);% image as shown in Figure 9

plot(xy_long(:,1),xy_long(:,2),'LineWidth',2,'Color','r');

gradient=(xy_long(2,2)-xy_long(1,2))/(xy_long(2,1)-xy_long(1,1));

x = atan(gradient);

jiaodu=x*180/pi;

Sixth, use the included angle to rotate the image, perform column pixel summation on the processed sugarcane, and find the extreme point of the summation function and the abscissa of the extreme point.

E＝imrotate(O, jiaodu, 'nearest', 'crop'); % will rotate the operation image

figure(10), imshow(E); % image as shown in Figure 10

y=sum(E);% Statistical diagram of the sum of pixels in the image row after rotation

figure('NumberTitle','off','Name','Pixel statistical maximum point'),plot(y);

title('column pixel summation chart');

xlabel('x');

ylabel('y');

[a,b]=findpeaks(y,'minpeakdistance',50,'minpeakheight',9);% find the maximum value

hold on;

[c,d]=max(y); %% looking for the maximum value point, d is the coordinate of the maximum value

hold on;

plot(b,a,'ro');% draw the extreme points

plot(d,c,'ro');% draw the most value points

Seventh, the position of the sugarcane stem node is obtained and marked through formula calculation.

figure('NumberTitle','off','Name','mark of sugarcane stem node'),imshow(A);

b1=b*cosd(abs(jiaodu));

for i=1:length(b1)

rectangle('Position',[b1(i)-8,70,20,150]);

end

d=d*cosd(abs(jiaodu));

rectangle('Position', [d-8,70,20,150]);

30 groups of yellow cane samples were randomly selected for the experiment, numbered from 1 to 30, and the number of nodes on each sugarcane was different, with a total of 275 stem nodes. Recognize the sugarcane images numbered 1-30 respectively, count the number of identified sugarcane nodes, compare it with the actual number, and calculate the error rate. The experimental data are shown in the following table:

It can be seen from the table that the detection rate of 17 sugarcane nodes is 100%, and the number of missed detection of all sugarcane nodes is no more than 1. There are a total of 275 sugarcane nodes, and a total of 263 sugarcane nodes have been detected. Up to 95.6%. Through the analysis of the test phenomenon, the missed detection parts are mainly distributed in the first or last cane node of sugarcane. The color of the sugarcane nodes in these two parts is lighter than that of other parts. In the image processing, the features are not obvious, and the detection is missed. Therefore, the program can be further optimized to enhance the contrast of the image, thereby reducing the phenomenon of missed detection.

The present invention provides a technical reference for the industrial preparation of domestic sugarcane seeds. Its core is to use the image processing technology based on matlab software to extract multiple objects on the surface of sugarcane to accurately identify the nodes of sugarcane, effectively avoiding missed detection and Error detection, thereby providing accurate position signals for subsequent cutting and separation of cane stems and cane seeds, preventing damage to buds and waste of cane seeds during the cutting process, and the present invention can also be used for raw materials that are tilted during feeding through image rotation. Correction, thereby improving the accuracy of recognition.

The foregoing descriptions of specific exemplary embodiments of the present invention have been presented for purposes of illustration and description. These descriptions are not intended to limit the invention to the precise form disclosed, and obviously many modifications and variations are possible in light of the above teaching. The exemplary embodiments were chosen and described in order to explain the specific principles of the invention and its practical application, thereby enabling others skilled in the art to make and use various exemplary embodiments of the invention, as well as various Choose and change. It is intended that the scope of the invention be defined by the claims and their equivalents.

Claims (6)

1. a kind of recognition methods of the sugarcane sugarcane section based on image procossing, which comprises the steps of:

Step 1 first acquires original image, original image is then converted to gray level image with matlab software, to gray level image Median filtering is carried out, then carries out binary conversion treatment；

Step 2 will replace through the black-and-white two color in step 1 treated image, recycle in Morphological scale-space and open operation Remove the part of nonlinear organization member；

The sugarcane background of gray level image obtained in step 1 is replaced into black by step 3, and sugarcane position is replaced into white, and Carry out binary conversion treatment；

Step 4 will carry out Sobel border detection through step 3 treated image, and extract apparent boundary characteristic；

Step 5 will detect straight line using Hough transform through step 4 treated image, and choose longest straight line, and calculate The angle α of longest straight line and image level direction；

Step 6, the angle α obtained using step 5 make the sugarcane section of image to treated that image rotates through step 2 In on same level straight line, then column pixel summation is carried out to image, finds the extreme point and maximum of points of summing function, extreme point Abscissa with maximum of points is the position d of cane stalk；

Step 7 passes through reduction formula d according to the position d of the cane stalk obtained after step 6 is handled₁=d x cos α, just Position d of the cane stalk in original image can be obtained₁。

2. the recognition methods of the sugarcane sugarcane section according to claim 1 based on image procossing, which is characterized in that in step 1 In, when image binaryzation is handled, the average value of max pixel value and minimum pixel value in image is set as threshold value, gray scale Pixel more than or equal to threshold value is set as 255, and the pixel less than threshold value is set as 0, in order to save the storage of computer, then 0 and 255 are further each mapped to 0 and 1.

3. the recognition methods of the sugarcane sugarcane section according to claim 1 based on image procossing, which is characterized in that in step 2 In, it is 15 using length, from trunnion axis to identify and retain this feature since the stipes of sugarcane is the shape of straight line Start 90 ° counterclockwise of flat linear structure member, formula are as follows:Wherein: B is structure Member, A be processed image, open operation the result is that the union that B is exactly matched and translated in A.

4. the recognition methods of the sugarcane sugarcane section according to claim 1 based on image procossing, which is characterized in that in step 4 In, Sobel border detection is to carry out convolution algorithm to original image using two (3x3) matrixes, to calculate the ash of horizontal direction Spend the estimated value G of local derviation_xWith the estimated value G of the gray scale local derviation of vertical direction_y。G_xAnd G_yMathematic(al) representation it is as follows:

With

Wherein: A is processed image；

Calculating process is as follows:

G_x=[f (x+1, y-1)+2*f (x+1, y)+f (x+1, y+1)]-[f (x-1, y-1)+2*f (x-1, y)+f (x-1, y+1)]；

G_y=[f (x-1, y-1)+2*f (x, y-1)+f (x+1, y-1)]-[f (x-1, y+1)+2*f (x, y+1)+f (x+1, y+1)；

Calculate the G of each point_xAnd G_yAfterwards, the gradient G of each point can be obtained by formula below:

Last set threshold value G_max, when gradient G is greater than G_max, which is a boundary point.

5. the recognition methods of the sugarcane sugarcane section according to claim 1 based on image procossing, which is characterized in that in step 5 In, Hough transform is that the straight line in plane of delineation space is converted into parameter space, formula are as follows:

Any point (x, y) can form a curve in the space ρ-θ on straight line, if two differences carry out above-mentioned behaviour The curve obtained after work intersects in plane ρ-θ, this means that they are located at same straight line, by the way that threshold value is arranged, meets at one The curve quantity of point has been more than threshold value, then the parameter (θ, ρ) for just representing this intersection point corresponds to the straight line in original image；It seeks First 5 points for being greater than 0.3 times of maximum value in Hough matrix are looked for, line segment, line segment of the combined distance less than 20, removal length are transformed to The straightway less than 5 is spent, and recycles to obtain longest line segment by for；The oblique of straight line is acquired by the coordinate value of line segment endpoint Rate k, then by formula α=arctank, acquire the angle α of longest straight line Yu image level direction.

6. the recognition methods of the sugarcane sugarcane section according to claim 1 based on image procossing, which is characterized in that in step 6 In, image is rotated using the imrotate function in matlab software, it, will be every in image using sum summing function The summation of column pixel, and drafting function image, find extreme point and maximum of points, extreme value using findpeaks function and max function Point and the abscissa of maximum of points are the position d of cane stalk.