CN104992144A - Method for distinguishing transmission line from road in remote sensing image - Google Patents

Abstract

The invention discloses a method for distinguishing a transmission line from a road in a remote sensing image. A vertical line is added for each identified parallel line target in a remote sensing image, wherein the spacing of the vertical lines is d picture elements, the width of the vertical lines is f picture elements, the center of the vertical lines is in the center of a parallel line area at the vertical point, the length of the vertical lines is K times of the maximum parallel line spacing of the parallel line targets, the number of the vertical lines is recorded as M, and the number of pixels on each vertical line is recorded as N. The two pixels, intersecting with the edge parallel lines of the parallel line targets, of each vertical line divides the vertical line into three parts. The pixels in the corresponding part of all the vertical lines are combined into a set, there are totally three pixel sets, the similarity of pixel gray values of the three pixel sets is calculated, and thus a transmission line and a road are distinguished. According to the invention, a transmission line and a road can be distinguished accurately in a remote sensing image based on the fact that the pixel gray value difference between a transmission line and surface features at the two sides thereof and the pixel gray value difference between a road and surface features at the two sides thereof are different.

Description

The differentiating method of power transmission line and highway in remote sensing images

Technical field

The invention belongs to digital image understanding technical field, more specifically say, relate to the differentiating method of power transmission line and highway in a kind of remote sensing images.

Background technology

Extra high voltage network is responsible for the important task of China's electric power transfer always, is the element task effectively ensureing electric line and device security thereof for polling transmission line management.And along with the fast development of spationautics, spatial surface information obtaining ability and level promote greatly.The spatial surface resolution of many commercial satellite remotely-sensed datas reaches sub-meter grade, and revisiting period shortens to one day.High resolving power commercial satellite conventional at present comprises QuickBird, GeoEye and Worldview, and its highest resolution has reached 0.41 meter, and believe that more high resolving power also can be come out soon, this makes satellite remote sensing technology be applied to electric inspection process becomes possibility.

With regard to acquisition cycle, cost and efficiency, satellite high-resolution optical remote sensing image has irreplaceable huge advantage.Describe one in Chinese patent " a kind of linear goal identification and extraction method based on image; CN103761524A; 2014.04.30 " and extract linear weak signal target based on linear feature, as the method for line of electric force, the method is by Linear feature extraction, then rejects non-flat line to reach the object identifying line of electric force.But in real process, the performance of highway in Linear feature extraction is often similar to transmission line of electricity, be all a parallel lines race, the method rejected by non-flat line well can not distinguish both.

Summary of the invention

The object of the invention is to overcome the deficiencies in the prior art, the differentiating method of power transmission line and highway in a kind of remote sensing images is provided, distinguish power transmission line and highway by the grey scale pixel value of parallel lines region atural object, improve the accuracy to power transmission line identification in remote sensing images.

For achieving the above object, the differentiating method of power transmission line and highway in remote sensing images of the present invention, comprises the following steps:

S1: vertical line interpolation is carried out to the parallel lines target recognized in remote sensing images, the method that vertical line adds is: the spacing between vertical line is d pixel, vertical line width is f pixel, vertical line center is in the center vertically locating parallel lines region, length of perpendicular is K times of parallel lines target maximum parallel lines spacing, and note vertical line quantity is M, and the pixel quantity on every bar vertical line is N, every bar vertical line arranges pixel by equidirectional, obtains grey scale pixel value sequence X _i=(x _{i, 1}, x _{i, 2}..., x _i,j..., x _i,N), wherein x _i,jrepresent the gray-scale value of a jth pixel in i-th vertical line, the span 1≤j≤N of the span 1≤i of i≤M, j;

S2: remember that every bar vertical line two pixel sequence numbers crossing with the sides aligned parallel line of parallel lines target are n _i1and n _i2, n _i1< n _i2, every bar vertical line is divided into three parts: $X_{i,2}=(x_{i,n_{i1}+1},x_{i,n_{i1}+2},...,x_{i,n_{i2}}),X_{i,3}=(x_{i,n_{i2}+1},x_{i,n_{i2}+2},...,x_{i,N}),$ By the Part I X of all vertical lines _{i, 1}pixel set be designated as part II X _{i, 2}pixel set be designated as $C_2=(c_{2,1},c_{2,2},...,c_{2,D_2}),$ Part III X _{i, 3}pixel set be designated as $C_3=(c_{3,1},c_{3,2},...,c_{3,D_3}),$ Wherein $D_1=\underset{i=1}{\overset{M}{\&Sigma;}}{n}_{i1},D_2=\underset{i=1}{\overset{M}{\&Sigma;}}(n_{i2}-n_{i1}),D_3=\underset{i=1}{\overset{M}{\&Sigma;}}(N-n_{i2});$

S3: calculate pixel set C ₁, C ₂and C ₃similarity S, if similarity is greater than predetermined threshold value T, then parallel lines target is power transmission line, otherwise is highway.

The differentiating method of power transmission line and highway in remote sensing images of the present invention, vertical line interpolation is carried out to the parallel lines target recognized in remote sensing images, spacing between vertical line is d pixel, vertical line width is f pixel, vertical line center is in the center vertically locating parallel lines region, length of perpendicular is K times of parallel lines target maximum parallel lines spacing, note vertical line quantity is M, pixel quantity on every bar vertical line is N, every bar vertical line is divided into three parts by every bar vertical line two pixels crossing with the sides aligned parallel line of parallel lines target, the pixel groups of all vertical line corresponding parts is synthesized a set, amount to three pixel set, calculate the similarity of the grey scale pixel value of three pixel set, distinguish power transmission line and highway.The present invention utilizes different respectively with both sides atural object grey scale pixel value difference of power transmission line and highway, in remote sensing images, distinguish power transmission line and highway exactly.

Accompanying drawing explanation

Fig. 1 is the process flow diagram of the differentiating method of power transmission line and highway in remote sensing images of the present invention;

Fig. 2 is remote sensing images exemplary plot to be identified;

Fig. 3 is the recognition result figure of parallel linear target in remote sensing images shown in Fig. 2;

Fig. 4 Shi Duitu3Zhong highway carries out the exemplary plot after vertical line interpolation;

Fig. 5 carries out the exemplary plot after vertical line interpolation to power transmission line in Fig. 3;

Fig. 6 is that vertical line divides schematic diagram;

Fig. 7 is the grey scale pixel value statistical graph of every root vertical line in Fig. 4;

Fig. 8 is the grey scale pixel value statistical graph of every root vertical line in Fig. 5;

The power transmission line target that Fig. 9 obtains after being through differentiation.

Embodiment

Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described, so that those skilled in the art understands the present invention better.Requiring particular attention is that, in the following description, when perhaps the detailed description of known function and design can desalinate main contents of the present invention, these are described in and will be left in the basket here.

Embodiment

Fig. 1 is the process flow diagram of the differentiating method of power transmission line and highway in remote sensing images of the present invention.As shown in Figure 1, in remote sensing images of the present invention, the differentiating method of power transmission line and highway comprises the following steps:

S101: vertical line adds:

Vertical line interpolation is carried out to the parallel lines target recognized in remote sensing images, the method that vertical line adds is: the spacing between vertical line is d pixel, vertical line width is f pixel, vertical line center is in the center vertically locating parallel lines region, length of perpendicular is K times of parallel lines target maximum parallel lines spacing, the span of K is K > 1, note vertical line quantity is M, pixel quantity on every bar vertical line is N, every bar vertical line arranges pixel by equidirectional, obtains grey scale pixel value sequence X _i=(x _{i, 1}, x _{i, 2}..., x _i,j..., x _i,N), wherein x _i,jrepresent the gray-scale value of a jth pixel in i-th vertical line, the span 1≤j≤N of the span 1≤i of i≤M, j.

Length between perpendiculars and vertical line width determine the complexity of subsequent calculations, can arrange according to actual needs.The multiple K of length of perpendicular and parallel lines target maximum parallel lines spacing can be arranged according to actual needs, but in order to the unitarity of parallel lines target proximity atural object, K value also should not arrange excessive, and the span usually arranging K is 1 < K≤4.Arrange K=2 in the present embodiment, what length between perpendiculars parameter d was got is 1/4th of length of perpendicular, i.e. N/4 pixel, vertical line width parameter f=1.

Fig. 2 is remote sensing images exemplary plot to be identified.Fig. 3 is the recognition result figure of parallel linear target in remote sensing images shown in Fig. 2.As shown in Figure 3, in the remote sensing images shown in Fig. 2, obtaining two parallel linear targets, is power transmission line and highway respectively.Fig. 4 Shi Duitu3Zhong highway carries out the exemplary plot after vertical line interpolation.Fig. 5 carries out the exemplary plot after vertical line interpolation to power transmission line in Fig. 3.

S102: three part divisions are carried out to vertical line:

First need to divide vertical line.Fig. 6 is that vertical line divides schematic diagram.As shown in Figure 6, remember that every bar vertical line two pixel sequence numbers crossing with the sides aligned parallel line of parallel lines target are n _i1and n _i2, n _i1< n _i2, every bar vertical line is divided into three parts: $X_{i,1}=(x_{i,1},x_{i,2},...,x_{i,n_{i1}}),X_{i,2}=(x_{i,n_{i1}+1},x_{i,n_{i1}+2},...,x_{i,n_{i2}}),$ by the Part I X of all vertical lines _{i, 1}pixel set be designated as $C_1=(c_{1,1},c_{1,2},...,c_{1,D_1}),$ Part II X _{i, 2}pixel set be designated as $C_2=(c_{2,1},c_{2,2},...,c_{2,D_2}),$ Part III X _{i, 3}pixel set be designated as $C_3=(c_{3,1},c_{3,2},...,c_{3,D_3}),$ Wherein $D_1=\underset{i=1}{\overset{M}{\&Sigma;}}{n}_{i1},D_2=\underset{i=1}{\overset{M}{\&Sigma;}}(n_{i2}-n_{i1}),$ $D_3=\underset{i=1}{\overset{M}{\&Sigma;}}(N-n_{i2}).$

S103: judge that the similarity of three partial pixel set obtains recognition result:

Calculate pixel set C ₁, C ₂and C ₃similarity S, if similarity is greater than predetermined threshold value T, then parallel lines target is power transmission line, otherwise is highway.

In practice, highway and its both sides atural object often difference are very large, and remote sensing images show as the widely different of pixel gray-scale value; Meanwhile, power transmission line is maked somebody a mere figurehead often, and the impact of image unit gray-scale value is little over the ground, so the pixel gray-scale value of power transmission line region and its both sides atural object relatively.Therefore, by the average pixel gray scale of comparison object and its both sides atural object, highway and power transmission line can be distinguished.That is, the difference degree of the average images light intensity value of three parts obtained by comparison step S102, can distinguish highway and power transmission line, middle and both sides atural object average images light intensity value difference great Wei highway, what the middle and average images light intensity value in both sides was basically identical is power transmission line.

Fig. 7 is the grey scale pixel value statistical graph of every root vertical line in Fig. 4.Fig. 8 is the grey scale pixel value statistical graph of every root vertical line in Fig. 5.As shown in Figure 7 and Figure 8, the vertical line grey scale pixel value of highway target obviously presents graded, and the change of the vertical line grey scale pixel value of power transmission line target is little, therefore by three part average gray values relatively just can judge obtain power transmission line target.Its concrete grammar is:

Calculate the average gray value of all vertical lines three parts respectively as follows:

$P_1=\frac{1}{D_1}\underset{g=1}{\overset{D_1}{\&Sigma;}}{c}_{1,g},P_2=\frac{1}{D_2}\underset{g=1}{\overset{D_2}{\&Sigma;}}{c}_{2,g},P_3=\frac{1}{D_3}\underset{g=1}{\overset{D_3}{\&Sigma;}}{c}_{3,g}$

Calculate average gray value P ₁, P ₂and P ₃adjacent part difference judge, that is, calculate Δ ₁=| P ₁-P ₂|, Δ ₂=| P ₂-P ₃|, order similarity S=1/ Δ _max.If S is greater than predetermined threshold value T, then parallel lines target is power transmission line, otherwise is highway.

Carry out knowledge method for distinguishing for average gray value, known by calculating, in the present embodiment, the average gray value of Tu4Zhong highway target vertical line three parts is difference P ₁=117, P ₂=85, P ₃in=49, Fig. 5, the average gray value of power transmission line target vertical line three parts is respectively P ₁=72, P ₂=73, P ₃=73.The average gray value of known Fig. 4 perpendicular bisector three parts differs greatly, and the average gray value difference of Fig. 5 perpendicular bisector three parts is very little, therefore can judge to obtain the parallel lines destination object that Fig. 4 perpendicular bisector adds is highway, is power transmission line, is consistent with actual conditions in Fig. 5.The power transmission line target that Fig. 9 obtains after being through differentiation.

Present invention also offers the detection of a kind of employing otherness and obtain pixel set C ₁, C ₂and C ₃the method of similarity S, concrete steps comprise:

Calculate pixel set C ₁, C ₂and C ₃in the total sum of squares SS of all grey scale pixel values _t, computing formula is:

${\mathrm{SS}}_T=\underset{k=1}{\overset{3}{\&Sigma;}}\underset{g=1}{\overset{D_k}{\&Sigma;}}{c_{k,g}}^2-(G^2/Q)$

Wherein, c _k,grepresent g grey scale pixel value in a kth pixel set, Q=M × N;

Calculate sum of squares between groups SS _b, computing formula is:

${\mathrm{SS}}_B=\underset{k=1}{\overset{3}{\&Sigma;}}\&lsqb;D_k{(P_k-\stackrel{\&OverBar;}{G})}^2\&rsqb;$

Wherein, $\stackrel{\&OverBar;}{G}=G/Q;$

Quadratic sum SS in calculating group _w, computing formula is:

${\mathrm{SS}}_W=\underset{k=1}{\overset{3}{\&Sigma;}}\underset{g=1}{\overset{D_k}{\&Sigma;}}{(c_{k,g}-P_k)}^2$

Calculate three quadratic sum SS _t, SS _band SS _wcorresponding degree of freedom df _t, df _band df _w:

df _T＝Q-1

df _B＝r-1＝3-1＝2

df _W＝Q-r＝Q-3

Calculate mean square MS _band MS _w:

${\mathrm{MS}}_B=\frac{{\mathrm{SS}}_B}{{\mathrm{df}}_B},{\mathrm{MS}}_W=\frac{{\mathrm{SS}}_W}{{\mathrm{df}}_W}$

Calculate F value:

$F=\frac{{\mathrm{MS}}_B}{{\mathrm{MS}}_W}$

Make similarity S=F, search the threshold value F that F value table obtains F value _aas threshold value T.

Although be described the illustrative embodiment of the present invention above; so that those skilled in the art understand the present invention; but should be clear; the invention is not restricted to the scope of embodiment; to those skilled in the art; as long as various change to limit and in the spirit and scope of the present invention determined, these changes are apparent, and all innovation and creation utilizing the present invention to conceive are all at the row of protection in appended claim.

Claims (4)

1., based on the power transmission line of remote sensing images and a differentiating method for highway, it is characterized in that, comprise the following steps:

S1: vertical line interpolation is carried out to the parallel lines target recognized in remote sensing images, the method that vertical line adds is: the spacing between vertical line is d pixel, vertical line width is f pixel, vertical line center is in the center vertically locating parallel lines region, length of perpendicular is K times of parallel lines target maximum parallel lines spacing, and note vertical line quantity is M, and the pixel quantity on every bar vertical line is N, every bar vertical line arranges pixel by equidirectional, obtains grey scale pixel value sequence X _i=(x _{i, 1}, x _{i, 2}..., x _i,j..., x _i,N), wherein x _i,jrepresent the gray-scale value of a jth pixel in i-th vertical line, the span 1≤j≤N of the span 1≤i of i≤M, j;

S2: remember that every bar vertical line two pixel sequence numbers crossing with the sides aligned parallel line of parallel lines target are n _i1and n _i2, n _i1< n _i2, every bar vertical line is divided into three parts: $X_{i,1}=(x_{i,1},x_{i,2},\mathrm{...},x_{i,n_{i1}}),$ $X_{i,2}=(x_{i,n_{i1}+1},x_{i,n_{i1}+2},\mathrm{...},x_{i,n_{i2}}),X_{i,3}=(x_{i,n_{i2}+1},x_{i,n_{i2}+2},...,x_{i,N}),$ By the Part I X of all vertical lines _{i, 1}pixel set be designated as part II X _{i, 2}pixel set be designated as part III X _{i, 3}pixel set be designated as wherein $D_1=\underset{i=1}{\overset{M}{\&Sigma;}}{n}_{i1},D_2=\underset{i=1}{\overset{M}{\&Sigma;}}(n_{i2}-n_{i1}),D_3=\underset{i=1}{\overset{M}{\&Sigma;}}(N-n_{i2});$

S3: calculate pixel set C ₁, C ₂and C ₃similarity S, if similarity is greater than predetermined threshold value T, then parallel lines target is power transmission line, otherwise is highway.

2. the differentiating method of power transmission line according to claim 1 and highway, is characterized in that, in described S1, the span of K is 1 < K≤4.

3. the differentiating method of highway according to claim 1, is characterized in that, in described step S3, the computing method of similarity are:

Calculate three pixel set C as follows respectively ₁, C ₂and C ₃average gray value P ₁, P ₂and P ₃:

$P_1=\frac{1}{D_1}\underset{g=1}{\overset{D_1}{\&Sigma;}}{c}_{1,g},P_2=\frac{1}{D_2}\underset{g=1}{\overset{D_2}{\&Sigma;}}{c}_{2,g},P_3=\frac{1}{D_3}\underset{g=1}{\overset{D_3}{\&Sigma;}}{c}_{3,g};$

Calculate Δ ₁=| P ₁-P ₂|, Δ ₂=| P ₂-P ₃|, make Δ _max=max (Δ ₁, Δ ₂), similarity S=1/ Δ _max.

4. the differentiating method of power transmission line according to claim 1 and highway, is characterized in that, in described step S3, the computing method of similarity are:

Calculate pixel set C ₁, C ₂and C ₃in the total sum of squares SS of all grey scale pixel values _t, computing formula is:

${\mathrm{SS}}_T=\underset{k=1}{\overset{3}{\&Sigma;}}\underset{g=1}{\overset{D_k}{\&Sigma;}}{c_{k,g}}^2-(G^2/Q)$

Wherein, q=M × N;

Calculate sum of squares between groups SS _b, computing formula is:

${\mathrm{SS}}_B=\underset{k=1}{\overset{3}{\&Sigma;}}\&lsqb;D_k{(P_k-\stackrel{\&OverBar;}{G})}^2\&rsqb;$

Wherein, $\stackrel{\&OverBar;}{G}=G/Q;$

Quadratic sum SS in calculating group _w, computing formula is:

${\mathrm{SS}}_W=\underset{k=1}{\overset{3}{\&Sigma;}}\underset{g=1}{\overset{D_k}{\&Sigma;}}{(c_{k,g}-P_k)}^2$

Calculate three quadratic sum SS _t, SS _band SS _wcorresponding degree of freedom df _t, df _band df _w:

df _T＝Q-1

df _B＝2

df _W＝Q-3

Calculate mean square MS _band MS _w:

${\mathrm{MS}}_B=\frac{{\mathrm{SS}}_B}{{\mathrm{df}}_B},{\mathrm{MS}}_W=\frac{{\mathrm{SS}}_W}{{\mathrm{df}}_W}$

Calculate F value:

$F=\frac{{\mathrm{MS}}_B}{{\mathrm{MS}}_W}$

Make similarity S=F, search the threshold value F that F value table obtains F value _aas threshold value T.