CN100545857C - Standing tree based on single image blocks the branch recognition methods - Google Patents

Abstract

The invention provides a kind of standing tree and block the branch recognition methods based on single image, described method relates to the forest survey field, described method comprises: after the standing tree image of taking is carried out the image pre-service, carry out the background information that image segmentation is removed described standing tree image; Trunk part is removed, obtained a plurality of branch chart pictures; Image thinning is obtained a plurality of branch skeletons; Described each branch skeleton is carried out fitting a straight line, draw the straight-line equation of described each branch skeleton; Calculate the similarity of corresponding two branches of any two branch skeletons according to the intercept of any two branch skeleton straight-line equations and slope meter, during greater than setting value, judge described two branches for blocking branch in described similarity.Whether the method for the invention has the branch of judgement for blocking branch, reduces the advantage of the maloperation of training robot.

Description

Standing tree based on single image blocks the branch recognition methods

Technical field

The present invention relates to forest inventory investigation, relate in particular to a kind of standing tree and block the branch recognition methods based on single image.

Background technology

Forest is a renewable resources important on the earth, be human survival in necessary condition, forest is a most important life support system on the earth for multiple biology provides the environment of perching.And trees are elements of forest, in order to allow trees better grow, trees need be pruned in the reality.

Training operation at present is generally by manually finishing, but its operating efficiency is low, and labour intensity is big, and human cost is very high, and certain potential safety hazard is arranged.In order to address the above problem, prior art provides a kind of Remote Control Automatic standing tree training machine, and this training machine can be watched trees attentively but need manually face upward at the scene head by the manually-operated remote control operation of pruning, when needs are pruned, by the operation of pruning of remote control training machine.In order to address the above problem, Beijing Forestry University provides a kind of intelligent robot that prunes, and this training robot discerns the standing tree branch automatically by the robotic vision system, finishes the training operation.

In realizing process of the present invention, the inventor finds prior art, and there are the following problems:

Training of the prior art robot discerns the standing tree branch automatically by vision system, and training robotic vision system is when photographic images, also the branch of other trees around the trees to be pruned also can be taken together in image, generally the branches of other trees is called and block branch; And existing training robot can't to tell those by photographic images be to block branch, especially can't tell those by single image is to block branch, has caused the maloperation of training robot more.

Summary of the invention

In view of above-mentioned existing in prior technology problem, the specific embodiment of the present invention provides a kind of standing tree based on single image to block the branch recognition methods, and said method can be told and block branch, thereby reduces the maloperation of training robot.

The present invention is achieved by the following technical solutions:

The specific embodiment of the invention provides a kind of standing tree based on single image to block the branch recognition methods, and described method comprises:

After the standing tree image of taking carried out the image pre-service, carry out the background information that image segmentation is removed described standing tree image;

To remove the trunk part of the described standing tree image of background information and remove, obtain a plurality of branch chart pictures;

A plurality of described branch image thinnings are obtained a plurality of branch skeletons;

During glyph zygonema sexual intercourse on judging each branch skeleton, described each branch skeleton is carried out fitting a straight line, draw the straight-line equation of described each branch skeleton;

Calculate the similarity of corresponding two branches of any two branch skeletons according to the intercept of any two branch skeleton straight-line equations and slope meter, during greater than setting value, judge described two branches for blocking branch in described similarity;

The similarity that the intercept of any two the branch skeleton straight-line equations of described basis and slope meter are calculated corresponding two branches of any two branch skeletons comprises:

According to

$K=\left|\frac{A1+A2}{2(A1-A2)}\right|\×\left|\frac{B1+B2}{2(B1-B2)}\right|$

Calculate the similarity of branch, wherein K is a similarity, and A1, B1 are the intercept and the slope of a branch skeleton straight-line equation, and A2, B2 are the intercept and the slope of another branch skeleton straight-line equation.

The technical scheme that is provided by the specific embodiment of the invention described above as can be seen, the described technical scheme of specific embodiments of the invention is by calculating the similarity degree of branch, and judge branch whether for blocking branch, thereby reduce the maloperation of training robot according to the value of the branch similarity degree that calculates.

Description of drawings

Fig. 1 is the process flow diagram of the described method of the specific embodiment of the invention.

Fig. 2 is the process flow diagram of the specific embodiment of the invention 1 described method.

Fig. 3 is the single image of shooting in the specific embodiment of the invention 1.

Fig. 4 is the standing tree image that removes background information in the specific embodiment of the invention 1.

Fig. 5 is a plurality of branch chart pictures in the specific embodiment of the invention 1.

Fig. 6 is the branch skeleton image in the specific embodiment of the invention 1.

Embodiment

The specific embodiment of the invention provides a kind of standing tree based on single image to block the branch recognition methods, and described method may further comprise the steps as shown in Figure 1:

Step 11, the standing tree image of taking carried out the image pre-service after, carry out the background information that image segmentation is removed described standing tree image;

The pretreated method of image in this step can adopt the method for filtering to reduce noise, and with the influence of reducing noise to the later image Treatment Analysis, the method for this filtering can adopt medium filtering; The image segmentation of carrying out in this step is removed the background information of described standing tree image and can be realized by the following method, adopts the iteration threshold dividing method to extract standing tree and branch, removes background information.

Step 12, the trunk part that will remove the described standing tree image of background information are removed, and obtain a plurality of branch chart pictures;

Realize this step method can for, the standing tree image that removes background information is adopted the fundamental operation-corrosion of mathematical morphology and the extraction that expansion realizes branch.The concrete operations of its realization can for, at first, to the standing tree image that removes background information carry out N time the corrosion, branch is eliminated.And then this image is carried out N+1 time expand, thereby obtain the trunk image; Standing tree figure image subtraction trunk image with removing background information has just obtained only having a plurality of branch chart pictures.

Step 13, a plurality of described branch image thinnings are obtained a plurality of branch skeletons;

Realize this step method can for, by formula (1) branch chart is looked like to carry out refinement, obtain the skeleton of branch.

$A\⊗\left\{B\right\}=(\·\·\·((A\⊗B_1)\⊗B_2)\·\·\·)\⊗B_n---\left(1\right)$

When step 14, the glyph zygonema sexual intercourse on judging each branch skeleton, described each branch skeleton is carried out fitting a straight line, draw the straight-line equation of described each branch skeleton;

The glyph zygonema sexual intercourse of judging on each branch skeleton in this step can realize by following method, calculate the linear system of each section branch skeleton two-dimensional coordinate according to formula (2), the linear coefficient that is calculating | r|＞0.9 o'clock, judge the glyph zygonema sexual intercourse on this section branch skeleton.

$\mathrm{\γ}=\frac{\mathrm{n\Σxy}-\mathrm{\Σx\Σy}}{\sqrt{(\mathrm{n\Σ}{x}^2-{\left(\mathrm{\Σx}\right)}^2)\·(\mathrm{n\Σ}{y}^2-{\left(\mathrm{\Σy}\right)}^2)}}---\left(2\right)$

In this step described each branch skeleton is carried out fitting a straight line, draw described each branch skeleton straight-line equation implementation method can for, according to formula (3) to described | the branch skeleton of r|＞0.9 carries out fitting a straight line, draws the straight-line equation Y=A of each branch skeleton _i+ B _iX.

$\left\{\begin{array}{c}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{y}_i=\mathrm{nA}+B\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{x}_i\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{x}_i{y}_i=A\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{x}_i+B\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{x}_i^2\end{array}\right.---\left(3\right)$

Step 15, calculate the similarity of corresponding two branches of any two branch skeletons according to the intercept of any two branch skeleton straight-line equations and slope meter;

Realize this step method can for, calculate similarity according to formula (4).

$k_{\mathrm{ij}}=\left|\frac{A_i+A_j}{2(A_i-A_j)}\right|\×\left|\frac{B_i+B_j}{2(B_i-B_j)}\right|---\left(4\right)$

K wherein _IjThe similarity of representing i branch and j branch, A _i, B _iBe the intercept and the slope of i branch skeleton straight-line equation, Aj, Bj are the intercept and the slope of j branch skeleton straight-line equation.

Step 16, whether judge described similarity, greater than setting value the time, judge described two branches for blocking branch greater than setting value; When being less than or equal to setting value, judging described two branches is not to block branch.

Setting value user in this step can rule of thumb set different seeds, and the preferred settings value is 100 here, and this setting value is by the known branch that blocks being calculated, draw when being when blocking branch, and the value of similarity is generally all greater than 100; The specific embodiment of the invention is not limited to the concrete numerical value of setting.

The device of realizing above-mentioned steps can be training robot or automatic standing tree training machine, and the specific embodiment of the invention is not limited to the concrete device of realizing this step.

For better describing the described method of embodiment of the present invention, existing 2～7 pairs of the specific embodiment of the present invention in conjunction with the accompanying drawings describe:

Embodiment 1: the embodiment of the invention 1 provides a kind of standing tree based on single image to block the branch recognition methods, the technical background of present embodiment 1 is: suppose that Fig. 3 is the single image of shooting, method in the present embodiment 1 is all handled Fig. 3, setting value can be 100, and this method comprises the steps:

Step 21, the single image Fig. 3 that shoots is carried out adopting the iteration threshold dividing method to extract standing tree and branch after medium filtering reduces noise, remove background information; The standing tree image that removes background information as shown in Figure 4;

Step 22, adopt mathematical morphology fundamental operation-corrosion and expansion to remove trunk part, obtain a plurality of branch chart pictures the standing tree image graph 4 of removing background information; The a plurality of branch chart pictures that draw as shown in Figure 5;

Step 23, a plurality of branch chart image pattern 5 refinements are obtained a plurality of branch skeletons; The a plurality of branch skeletons that obtain as shown in Figure 6;

The linear coefficient of step 24, a plurality of branch skeletons of calculating; Carry out following step at linear coefficient;

Realize this step method can for, calculate the linear coefficient of a plurality of branch skeleton L1, L2, L3, L4 according to formula (2), the absolute value of L1, the L2 that calculates according to formula (2), the linear coefficient of L3, L4 is all greater than 0.9.

Step 25, the branch skeleton in a plurality of branch skeleton diagrams 6 is carried out the straight-line equation that fitting a straight line draws the branch skeleton;

Realize this step concrete grammar can for, the branch skeleton L1 among Fig. 6, L2, L3, L4 are carried out fitting a straight line according to formula (3), it is as follows to draw straight-line equation:

L1：Y＝-0.664X+293.664

L2：Y＝-0.252X+277.274

L3：Y＝-0.807X+136.411

L4：Y＝-0.736X+126.107

Step 26, calculate the similarity of corresponding two branches of any two branch skeletons according to the intercept of any two branch skeleton straight-line equations and slope meter;

Realize this step concrete grammar can for:

Draw according to the straight-line equation in the step 25:

A1＝293.664，B1＝-0.664

A2＝277.274，B2＝-0.252

A3＝136.411，B3＝-0.807

A4＝126.107，B4＝-0.736

Calculate similarity according to formula (4), its result calculated is as follows:

The similarity of L1 and L2 is: K12=19.36

The similarity of L1 and L3 is: K13=7.03

The similarity of L1 and L4 is: K14=12.18

The similarity of L2 and L3 is: K23=1.40

The similarity of L2 and L4 is: K24=1.36

The similarity of L3 and L4 is: K34=138.42

The similarity that the similarity that step 27, basis calculate is judged L3 and L4 belongs to and blocks branch greater than 100.

The described method of the specific embodiment of the invention by calculating the similarity degree of branch, and is judged branch whether for blocking branch according to the value of the branch similarity degree that calculates, thereby is reduced the maloperation of training robot.

The described technical scheme of the specific embodiment of the invention can be judged branch whether for blocking branch, and reduces the maloperation of training robot.

The above; only for the preferable embodiment of the present invention, but protection scope of the present invention is not limited thereto, and anyly is familiar with those skilled in the art in the technical scope that the present invention discloses; the variation that can expect easily or replacement all should be encompassed within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with the protection domain of claim.

Claims (2)

1, a kind of standing tree based on single image blocks the method for branch identification, it is characterized in that described method comprises:

After the standing tree image of taking carried out the image pre-service, carry out the background information that image segmentation is removed described standing tree image;

To remove the trunk part of the described standing tree image of background information and remove, obtain a plurality of branch chart pictures;

A plurality of described branch image thinnings are obtained a plurality of branch skeletons;

During glyph zygonema sexual intercourse on judging each branch skeleton, described each branch skeleton is carried out fitting a straight line, draw the straight-line equation of described each branch skeleton;

Calculate the similarity of corresponding two branches of any two branch skeletons according to the intercept of any two branch skeleton straight-line equations and slope meter, during greater than setting value, judge described two branches for blocking branch in described similarity;

The similarity that the intercept of any two the branch skeleton straight-line equations of described basis and slope meter are calculated corresponding two branches of any two branch skeletons comprises:

According to

$K=\left|\frac{A1+A2}{2(A1-A2)}\right|\×\left|\frac{B1+B2}{2(B1-B2)}\right|$

Calculate the similarity of branch, wherein K is a similarity, and A1, B1 are the intercept and the slope of a branch skeleton straight-line equation, and A2, B2 are the intercept and the slope of another branch skeleton straight-line equation.

2, method according to claim 1 is characterized in that, the described trunk part that will remove the described standing tree image of background information is removed, and obtains a plurality of branch charts and looks like to comprise:

The described standing tree image that will remove background information adopts mathematical morphology fundamental operation-corrosion and expansion to remove trunk part, obtains a plurality of branch chart pictures.

Images