CN108665464A - A kind of foreign matter detecting method based on morphologic high tension electric tower and high-tension bus-bar - Google Patents

Abstract

The present invention provides a foreign object detection method based on morphology of high-voltage electric towers and high-voltage wires, which relates to the field of foreign object detection in electric power facilities, including: step 1: collecting and reading images _I1 of high-voltage electric towers and high-voltage electric wires; 2: Segment the image I ₁ to obtain the binary image I ₂ ; Step 3: Select the appropriate morphological structural elements for the binary image I ₂ , find the rough position of the foreign object, and obtain the image I ₃ ; Step 4: The foreign object Perform edge detection at a rough position, find the edge contour of the foreign object, and obtain image I ₄ ; step 5: perform morphological filling on image I ₄ to obtain morphologically filled image I ₅ ; step 6: morphologically filled image Step ₁₅ checks layer by layer morphology successively to obtain the number and position of foreign objects in electrical facilities; Step 7: mark all high-voltage towers and high-voltage wire foreign objects in the corresponding positions in the original picture. The invention solves the problem that the identification types of foreign matter detection in high-voltage electric towers and high-voltage wires are single.

Description

A foreign object detection method for high-voltage towers and high-voltage wires based on morphology

technical field

The invention relates to the field of foreign object identification of electric power facilities, in particular to a method for detecting foreign objects of high-voltage electric towers and high-voltage lines based on morphology.

Background technique

Electricity is the main artery of energy in our country, and the transmission line network is the main carrier of power transmission, so it is particularly important to maintain the normal operation of transmission lines. It is relatively quick to build these transmission lines, but long-term maintenance requires huge manpower, financial and material resources. In recent years, it is not uncommon for foreign objects such as kites and balloons to endanger the safety of the power grid in various places. Such foreign objects suspended on transmission lines will shorten the limit discharge distance of high-voltage electricity, and even cause serious consequences of large-scale power outages. Therefore, it is of great significance to identify foreign objects on the transmission line in time.

The existing inspection of foreign objects on transmission lines is mainly manual line inspection, but manual line inspection has the disadvantages of large safety hazards, low work efficiency, and difficult operation for some complex terrain transmission lines. In order to reduce work intensity and improve work efficiency, in recent years, there has been a method of inspecting 110-1000kV high-voltage transmission line corridors by using aircraft as a vehicle, loading visible light imaging detection equipment, and applying computer intelligence to process a large number of inspections brought back. Image data is used to judge whether there are foreign objects on the line. This technology can greatly improve the level and efficiency of inspection technology, reduce the maintenance cost of transmission lines, and have great significance for creating better economic and social benefits.

Contents of the invention

The purpose of the present invention is to solve the problem that the foreign object detection method of the current high-voltage electric tower and high-voltage wire detects a single type of foreign object, and the present invention proposes a foreign object detection method for high-voltage electric tower and high-voltage electric wire based on morphology.

Technical scheme of the present invention is as follows:

A kind of foreign matter detection method of high-voltage electricity tower and high-voltage electric wire based on morphology, comprises the following steps:

Step 1: Collect and read in the image I ₁ of high-voltage towers and high-voltage wires;

Step 2: Segment image I ₁ to obtain binary image I ₂ ;

Step 3: Select the appropriate morphological structural elements for the binary image _I2 , find the rough position of the foreign matter, and obtain the image _I3 ;

Step 4: Perform edge detection on the rough position of the foreign matter, find the edge contour of the foreign matter, and obtain the image I ₄ ;

Step 5: performing morphological filling on the image I ₄ to obtain a morphologically filled image I ₅ ;

Step 6: Check the morphologically filled image _I5 sequentially layer by layer to obtain the number and position of foreign objects in the power facility;

Step 7: Mark all high-voltage towers and foreign objects on high-voltage wires at the corresponding positions in the original picture.

Specifically, the method of segmentation in the step 2 is the OTSU segmentation method, and the specific steps are as follows:

Step 21: Set the number of pixels of the original image I ₁ as M, the number of pixels with gray level i as n _i , normalize the gray level histogram, and the probability of the pixel with gray level i as p _i :

p _i =n _i /M

Step 22: Suppose L is the gray level of the original image I ₁ , k is its gray level threshold, select the threshold T(k)=k, and use it to threshold the total pixels of the original image I ₁ into c ₁ and c ₂ two types of pixels, P is the probability of occurrence of this type of pixel; for two types of pixels c ₁ and c ₂ , the probability of occurrence of each type is P ₁ (k), P ₂ (k):

The average gray value of the pixels assigned to c ₁ and c ₂ are m ₁ , m ₂ :

In the _formula , _C1 is the number of pixels _of c1 type pixels, _C2 is the number of pixels of c2 type pixels,

The global gray mean of the entire image I ₁ is:

is the between-class variance, defined as:

The best threshold is k*, then maximize the inter-class variance

At this point, the threshold k* is determined, the original image I ₁ is divided into two parts c ₁ and c ₂ , all c ₁ type pixels are converted into pixels with a gray level of 0, and all c ₂ type pixels are converted into gray The pixels whose degree series is 1 convert the original image I ₁ into a binary image I ₂ and complete the OTSU threshold segmentation.

Specifically, the specific steps of the step 3 are as follows:

Step 31: Assuming that the disc-shaped structural element is S, the initial size of its radius R is r, and its radius R increases as the structural element S is used again, the image after corrosion is E, and the binary image is I ₂ ;

Use the structural element S to scan the binary image, and use the structural element and the binary image I ₂ covered by the "AND" operation. If both are 1, the pixel of the corroded image E is 1, otherwise it is 0;

Use corrosion to corrode and reduce the binary image I ₂ according to the radius R of the disc-shaped structural element S, so that the gaps can be eliminated, and the image corrosion of the binary image I ₂ can be reduced, and part of the structure of the corroded power facility can be removed, and at the same time Corrosion and reduction of foreign objects in power facilities; if the disc-shaped structural element S is used again, as the size of its radius R increases, the ability to eliminate the space between the binary image I ₂ and the image corrosion reduction of the binary image I ₂ will be increased ;

Step 32: Assume that the image after etching is E, and use the disk-shaped structural element S in step 31 to expand the image E to obtain an image I ₃ ;

Use the disk-shaped structural element S to scan the binary image, and use the structural element and the binary image I ₂ covered by the "OR" operation. If both are 0, the pixel of the expanded image I ₃ will be 0, otherwise it will be 1;

Utilize expansion to expand the corroded image E with the same disc-shaped structural element S to restore the size and shape of the corroded power facility foreign matter according to the radius R of the disc-shaped structural element S, and obtain the expanded image I ₃ ; if the circle When the disk-shaped structural element S is used again, as its radius R increases, the restoration ability of the image after corrosion is increased. The size of the image is restored. The location of foreign objects in the power facilities.

Specifically, the specific steps of the step 4 are as follows:

Step 41: Perform Gaussian filtering on image I ₃ , specifically:

Assuming that the two-dimensional kernel vector is K, the variance is σ, the horizontal and vertical coordinates of the pixel of image _I3 are x, y, and the parameters are determined to obtain the two-dimensional kernel vector K;

The above formula is a discretized two-dimensional Gaussian function. After determining the parameters and obtaining the two-dimensional kernel vector, the image Gaussian filter is performed, so that the gray value of the pixel to be filtered and its neighborhood points is weighted and averaged;

Step 42: Use the two matrices s _x and s _y of the partial derivatives of the image in the x and y directions to obtain the finite difference of the first-order partial derivative, and use the first-order finite difference to approximate the gradient of the gray value of the image:

And use the finite difference of the first-order partial derivative to calculate the magnitude M and direction θ of the image I ₃ gradient;

Step 43: Use the magnitude M and direction θ of the gradient to suppress the non-maximum value of the gradient magnitude, find the local maximum value of the pixel, set the gray value corresponding to the non-maximum value point to 0, and the current position The gradient value is compared with the gradient value on both sides of the gradient direction and the gradient direction is perpendicular to the edge direction; a binary image is obtained after non-maximum suppression is completed, and the gray value of non-edge points is 0, which may be a local edge The gray value of the maximum gray point is 128;

Step 44: Use the double-threshold algorithm to detect and connect edges, select two thresholds, get an edge image according to the high threshold, link the edges into contours in the high-threshold image, when reaching the endpoint of the contour, in the 8 neighborhood of the breakpoint Find the point that satisfies the low threshold among the points, and then collect new edges based on this point until the edge of the entire image is closed, and the image I ₄ after edge detection by the canny operator is obtained.

Specifically, the specific steps of the step 6 are as follows:

Step 61: Perform area statistics of all connected regions on the image I5 obtained in step ₅ , obtain the maximum and minimum areas of the connected regions, and calculate the circularity of the connected regions;

Step 62: Judging whether the maximum area of the connected region is equal to 0, if yes, proceed to step 7, otherwise proceed to step 63;

Step 63: judging whether the ratio of the maximum area of the connected region to the minimum area of the connected region belongs to the normal range, if so, proceed to step 64, otherwise, return to step 3;

Step 64: Judging whether the circularity of all foreign objects belongs to the normal range, if so, proceed to step 65, otherwise, return to step 3;

Step 65: Judging whether the minimum area of the connected region belongs to the normal range, if so, proceed to step 66, otherwise, return to step 3;

Step 66: Judging whether the maximum area of the connected region belongs to the normal range, if so, proceed to step 67, otherwise, return to step 3;

Step 67: Determine whether the number of foreign objects belongs to the normal range, if so, go to step 68, otherwise, go back to step 3.

Step 68: Counting the number and position of foreign objects in the final determined high-voltage towers and high-voltage wires.

Specifically, the specific steps of the step 7 are as follows:

Step 71: Obtain the coordinates of the center of mass of the foreign object area and store them in an array for final positioning;

Step 72: After obtaining the contour position of the foreign object area, use this position information in the original image, and change the gray value of the corresponding position into a painted marked area.

Further, in the step ₂ , before the image _I1 is segmented to obtain the binary image I2, the image _I1 is subjected to B-channel extraction processing of the RGB image.

In summary, compared with the prior art, due to the adoption of the above-mentioned technical solution, the beneficial effects of the present invention are:

(1) The present invention detects many types of foreign objects on high-voltage electric towers and high-voltage cables. For a specific comparison with the detection types of the prior art, please refer to the table and its description in the specific embodiment.

(2) The method for identifying foreign objects in high-voltage electric towers and high-voltage cables based on morphology of the present invention, because the image structures of high-voltage electric towers and high-voltage cables are special, except for foreign objects, they are all composed of staggered linear elements, and other scenes The foreign object detection method is not applicable to the scene of the present invention. If other detection methods are used to apply to the scene of the present invention, the recognition effect is very bad. The present invention can be attached to the foreign object mark, and the method of other scenarios is applied to the foreign object recognition of power facilities Without this feature, it is even impossible to locate foreign objects.

(3) When foreign matter is attached to the high-voltage tower and high-voltage cable, it will cause damage to the original structure. Using morphological structural elements and closed operations to remove the original high-voltage tower and high-voltage cable background, the image can be inspected very quickly The number of foreign objects in high-voltage towers and high-voltage wires and the position of foreign objects in high-voltage electricity towers and high-voltage wires will not damage the shape of the original foreign objects, and the subsequent marks can perfectly restore the display of foreign objects, making the results more intuitive It is clearly displayed on the image, and the follow-up work of removing foreign objects from high-voltage power towers and high-voltage wires also enables faster identification of foreign objects from high-voltage power towers and high-voltage wires.

(4) Based on morphology of the present invention, a large number of samples have been collected to study the further judgment and screening range of foreign objects in high-voltage electric towers and high-voltage cables. For high-voltage electric towers and high-voltage cables A series of conditions such as the maximum and minimum area of the foreign matter and circularity are screened, so that the foreign matter recognition rate of the present invention is greatly increased, combined with modern features, to achieve good experimental results.

(5) The morphology-based method for identifying foreign objects on high-voltage towers and high-voltage lines of the present invention uses the sky as the most important background picture, and when the sky is in dark blue, it will have the largest deviation impact on other processing methods , but the image of the present invention extracts the B channel of the RGB channel, which can greatly improve the recognition rate of the image obtained after threshold processing, and reduces the dependence on the surrounding environment of high-voltage electric towers and high-voltage cables, reducing the cost of obtaining image data The cost and technology greatly meet the needs of inspection images for foreign object detection and recognition.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required in the embodiments. Obviously, the accompanying drawings in the following description are only some of the present invention. Embodiments, for those of ordinary skill in the art, other drawings can also be obtained based on these drawings without any creative effort. The above and other objects, features and advantages of the present invention will be more clearly illustrated by the accompanying drawings. Like reference numerals designate like parts throughout the drawings. The drawings are not intentionally scaled according to the actual size, and the emphasis is on illustrating the gist of the present invention.

Fig. 1 is a schematic flow sheet of the present invention;

Fig. 2 is initial image of the present invention;

Fig. 3 is OTSU threshold segmentation figure among the present invention;

Fig. 4 is the result figure of closing operation and morphological restriction operation of marking circularity;

Fig. 5 is the figure of recognition result;

Figure 6 is a diagram of the program running results;

Fig. 7 is the image that directly carries out threshold segmentation without carrying out B channel extraction among the present invention;

Fig. 8 is the image after performing B channel extraction in the present invention and then performing threshold segmentation;

Fig. 9 is an original picture of a bird's nest identified in the present invention;

Fig. 10 is the result figure of the identification mark of Fig. 9 by the method that Anhui Jiyuan Software Company adopts in the present invention;

Fig. 11 adopts the method of the present invention in the present invention to the identification mark result figure of Fig. 9;

Figure 12 is the original picture of a kite identified in the invention;

Fig. 13 adopts the method of the present invention in the present invention to the identification mark result figure of Fig. 12;

Figure 14 The original picture of a balloon identified in the invention;

Fig. 15 is a diagram of the recognition mark result of Fig. 14 by using the method of the present invention in the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments It is a part of embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

As shown in Figure 1, a kind of morphology-based high-voltage electric tower and high-voltage wire foreign object detection method of the present invention, as shown in Figure 1, comprises the following steps:

Step 1: Collect and read the image I ₁ of the power facility, as shown in Fig. 2 .

Step 2: Extract the B channel of the RGB image on the image I ₁ , and then use the OTSU segmentation method to segment the image to obtain a binary image I ₂ , as shown in Figure 3, specifically:

Step 21: Set the number of pixels of the original image I ₁ as M, the number of pixels with gray level i as n _i , normalize the gray level histogram, and the probability of the pixel with gray level i as p _i :

p _i =n _i /M

Step 22: Suppose L is the gray level of the original image I ₁ , k is its gray level threshold, select the threshold T(k)=k, and use it to threshold the total pixels of the original image I ₁ into c ₁ and c ₂ two types of pixels, P is the probability of occurrence of this type of pixel; for two types of pixels c ₁ and c ₂ , the probability of occurrence of each type is P ₁ (k), P ₂ (k):

The average gray value of the pixels assigned to c ₁ and c ₂ are m ₁ , m ₂ :

In the _formula , _C1 is the number of pixels _of c1 type pixels, and _C2 is the number of pixels of c2 type pixels;

The global mean of the entire image _I1 is:

is the between-class variance, defined as:

Thus the optimal threshold is k*, maximizing the inter-class variance

At this point, the threshold k* is determined, the original image I ₁ is divided into two parts c ₁ and c ₂ , all c ₁ type pixels are converted into pixels with a gray level of 0, and all c ₂ type pixels are converted into gray The pixels whose degree series is 1 convert the original image I ₁ into a binary image I ₂ and complete the OTSU threshold segmentation.

Step 3: Select the appropriate morphological structural elements for the binary image I ₂ , find the rough position of the foreign object, and obtain the image I ₃ , the specific steps are as follows:

Step 31: Assuming that the disk-shaped structural element is S, the initial size of its radius R is r, and its radius R increases as the structural element S is used again, the image after erosion is E, and the binary image is I ₂ .

Use the structural element S to scan the binary image, and use the structural element and the binary image I ₂ covered by the "AND" operation. If both are 1, the pixel of the corroded image E will be 1. Otherwise 0.

Then use corrosion to corrode and reduce the binary image I ₂ according to the radius R of the disc-shaped structural element S, so that the interval gaps can be removed, and the image corrosion of the binary image I ₂ can be reduced, that is, part of the structure of the corroded power facility can be removed. At the same time, it also reduces the corrosion of foreign objects in power facilities. If the disk-shaped structural element S is used again, as its radius R increases, the ability to eliminate gaps in the binary image I ₂ and image erosion and reduction in the binary image I ₂ will increase.

Structural element, the so-called structural element refers to the basic structure with a certain shape, and its selection generally requires that it has rotation invariance or mirror invariance, so that the origin of the structural element is at its geometric center, and the surrounding pixels are symmetrical about the origin.

Step 32: Assume that the image after etching is E, and use the disk-shaped structural element S in step 31 to expand the image E to obtain an image I ₃ .

Use the disk-shaped structural element S to scan the binary image, and use the structural element and the binary image I ₂ covered by the "OR" operation. If both are 0, the pixel of the expanded image I ₃ will be 0. 1 otherwise.

Then use expansion to expand and restore the corroded image E with the same disc-shaped structural element S according to the radius R of the disc-shaped structural element S to obtain the expanded image I ₃ . If the disk-shaped structural element S is used again, as its radius R increases, the restoration ability of the image after corrosion is increased, and the image size is restored. After the image is corroded and dilated, part of the structure and background of the power facility is removed, and a rough location of foreign objects in the power facility is obtained.

Step 4: Perform canny operator edge detection on the image obtained in step 4 to obtain the edge profile of foreign objects or other interference objects. Step 4 is specifically:

Assuming that the two-dimensional kernel vector is K, the variance is σ, the horizontal and vertical coordinates of the pixel of image _I3 are x, y, and the parameters are determined to obtain the two-dimensional kernel vector K;

The above formula is a discretized two-dimensional Gaussian function. After determining the parameters and obtaining the two-dimensional kernel vector, the image Gaussian filter is performed, so that the gray value of the pixel to be filtered and its neighborhood points is weighted and averaged.

Step 42: Use the two matrices s _x and s _y of the partial derivatives of the image in the x and y directions to obtain the finite difference of the first-order partial derivative, and use the first-order finite difference to approximate the gradient of the gray value of the image:

And use the finite difference of the first order partial derivative to calculate the magnitude M and direction θ of the image I ₃ gradient.

Step 43: Use the magnitude M and direction θ of the gradient to suppress the non-maximum value of the gradient magnitude, find the local maximum value of the pixel, set the gray value corresponding to the non-maximum value point to 0, and the current position The gradient value is compared with the gradient value on both sides of the gradient direction and the gradient direction is perpendicular to the edge direction; a binary image is obtained after non-maximum suppression is completed, and the gray value of non-edge points is 0, which may be a local edge The gray value of the maximum gray point is 128;

Step 44: Use the double-threshold algorithm to detect and connect edges, select two thresholds, get an edge image according to the high threshold, link the edges into contours in the high-threshold image, when reaching the endpoint of the contour, in the 8 neighborhood of the breakpoint Find the point that satisfies the low threshold among the points, and then collect new edges based on this point until the edge of the entire image is closed, and the image I ₄ after edge detection by the canny operator is obtained.

Step 5: Perform morphological filling on the image I ₄ to obtain a morphologically filled image I ₅ .

Step 6: The morphologically filled image I ₅ is sequentially inspected layer by layer to obtain the number and position of foreign objects in the electrical facility.

Step 6: Mark all the foreign objects of the power facilities at the corresponding positions in the original picture. After this step is completed, the resulting graph is shown in Figure 4.

Step 61: Perform area statistics of all connected regions on the image I5 obtained in step ₅ , obtain the maximum and minimum areas of the connected regions, and calculate the circularity of the connected regions;

Step 62: Judging whether the maximum area of the connected region is equal to 0, if yes, proceed to step 7, otherwise proceed to step 63;

Step 63: Determine whether the ratio of the maximum area value of the connected region to the minimum area value of the connected region belongs to the normal range, that is, whether the ratio of the maximum area value to the minimum area value of the connected region is less than 3 times, if so, then enter step 64, otherwise , return to step 3;

Step 64: Judging whether the circularity of all foreign objects belongs to the normal range, that is, whether the circularity of all foreign objects is 0.5 to 0.8, if yes, proceed to step 65, otherwise, return to step 3;

Step 65: Determine whether the minimum area of the connected region belongs to the normal range, that is, whether the minimum area of the connected region is less than one-thousandth of the area of the original image _I1 , if so, then enter step 66, otherwise, return to step 3;

Step 66: Determine whether the maximum area of the connected region belongs to the normal range, that is, whether the maximum area of the connected region is greater than 1/750 of the area of the original image, if so, enter step 67, otherwise, return to step 3;

Step 67: Determine whether the number of foreign objects belongs to the normal range, that is, whether the number of foreign objects is less than six, if so, go to step 68, otherwise, go back to step 3.

Step 68: Counting the number and position of foreign objects in the final determined high-voltage towers and high-voltage wires.

Step 7: Mark all high-voltage towers and foreign objects on high-voltage wires at the corresponding positions in the original picture. Specifically include:

Step 71: Locate the centroid of the position corresponding to the foreign matter of the high-voltage electric tower and high-voltage wire in the original image, obtain the coordinates of the center of the foreign matter, and store it in an array as the final position.

Step 72: Obtain the contour position of the foreign matter, use this contour position in the original image, and change the gray value of the corresponding position of the foreign matter in the original image into a colored mark. After this step is completed, the result is shown in Figure 5.

After step 7 is completed, in order to make statistics more convenient in the specific implementation, the data should also be packaged and interface designed, and finally displayed on the independent application interface, in terms of image screening, as shown in Figure 6. This step is well known to those skilled in the art, and will not be described in detail here.

The present invention detects many types of foreign matter on high-voltage electric towers and high-voltage cables, and the specific comparison table is shown in Table 1. Figure 10 is the identification effect diagram of the bird's nest in Figure 9 by the method adopted by Anhui Jiyuan Software Company. It can be seen in Figure 10 that only simple line marking is carried out in this figure, and the marking is rough, while Figure 11 is the identification of Figure 9 In the effect picture, it can be seen that the mark of the bird's nest in Figure 11 is accurately covered on the bird's nest, and the detection effect is very good. Fig. 2 and Fig. 5 are the detection original picture and result picture of another bird's nest according to the present invention. Fig. 12 and Fig. 13 are the original and effect diagrams of kite recognition according to the present invention, and Fig. 14 and Fig. 15 are the effect diagram and original diagram of balloon recognition according to the present invention.

Table 1

The method for identifying foreign objects in high-voltage electric towers and high-voltage cables based on morphology of the present invention collects a large number of samples to study the further judgment and screening range of foreign objects in high-voltage electric towers and high-voltage electric wires. A series of conditions such as area, minimum area and circularity are screened, so that the foreign object recognition rate of the present invention is greatly increased, combined with modern features, a good experimental effect is achieved, and the specific results are shown in icon 2.

Table 2

Foreign body type Number of samples Accuracy the bird's nest 76 97.4% Kite twenty one 95.2% balloon twenty four 100% No foreign matter 114 96.9% foreign body 121 97.5%

It is worth noting that the morphology-based recognition method for foreign objects on high-voltage electric towers and high-voltage lines of the present invention uses the sky as the most important background image. The color of the sky is sometimes gray, sometimes light blue, and sometimes dark blue. More preferably implement the present invention, when the sky is dark blue, if the B channel is not extracted from the original image, the final result will be as shown in Figure 8, and the method of the present invention is used to extract the image after the B channel of the RGB channel , the final result is shown in Figure 9, which can greatly improve the recognition rate of the image obtained after threshold processing, and reduce the dependence on the surrounding environment of high-voltage power towers and high-voltage cables, reducing the cost and technology of image data acquisition, It greatly meets the needs of inspection images for foreign object detection and recognition.

Claims (7)

1. a kind of foreign matter detecting method based on morphologic high tension electric tower and high-tension bus-bar, which is characterized in that including following step Suddenly:

Step 1：Acquire and read in the image I of high tension electric tower and high-tension bus-bar₁；

Step 2：To image I₁It is split to obtain bianry image I₂；

Step 3：To bianry image I₂It carries out choosing suitable morphological structuring elements, finds the rough position where foreign matter, obtain Image I₃；

Step 4：Edge detection is carried out to foreign matter rough position, the edge contour of foreign matter is found, obtains image I₄；

Step 5：To image I₄The image I after morphology filling is obtained after carrying out morphologic filling₅；

Step 6：Image I after being filled to morphology₅Successively morphological examination successively, obtains electric power facility foreign matter number and location；

Step 7:Corresponding position marks all high tension electric towers and high-tension bus-bar foreign matter in artwork.

2. a kind of and morphologic high tension electric tower and high-tension bus-bar foreign matter detecting method according to claim 1, feature It is, the method divided in the step 2 is OTSU split plot designs, is as follows：

Step 21：If original image I₁Pixel number be M, the pixel number that gray level is i is n_i, grey level histogram is carried out Normalization, the pixel probability that gray scale is i is p_i：

p_i=n_i/M

Step 22：Assuming that L is original image I₁Number of greyscale levels, k is its number of greyscale levels threshold value, selected threshold T (k)=k, and making With it original image I₁Total pixel thresholding is divided into c₁And c₂Two class pixels, P are the probability that such pixel occurs；For two Class pixel c₁And c₂, it is respectively P per a kind of probability occurred₁(k),P₂(k)：

It is assigned to c₁And c₂The average gray value of pixel be respectively m₁,m₂：

In formula, C₁For c₁The number of pixels of class pixel, C₂For c₂The number of pixels of class pixel,

Whole image I₁Global gray average be：

For inter-class variance, it is defined as：

Optimal threshold is k*, class variance between then maximizing

This is arrived, threshold value k*, by original image I₁It is divided into c₁And c₂Two parts, by whole c₁Class pixel is converted into gray level 0 pixel, by whole c₂Class pixel is converted into the pixel that number of greyscale levels is 1, makes original image I₁It is converted into bianry image I₂, complete OTSU Threshold segmentations.

3. it is according to claim 1 a kind of based on morphologic high tension electric tower and high-tension bus-bar foreign matter detecting method, it is special Sign is that the step 3 is as follows：

Step 31：Assuming that collar plate shape structural element is S, the initial size of radius R sizes is r, and its radius R is with structure Increase when element S reuses, image is E, bianry image I after corrosion₂；

Bianry image, the bianry image I covered with it with structural element are scanned using structural element S₂All it is if doing with operation 1, the pixel of image E is 1 after as a result corroding, and is otherwise 0；

Using corrosion by bianry image I₂Corrosion reduction is carried out according to the radius R of collar plate shape structural element S, interstitial spaces is made to obtain It rejects, and makes bianry image I₂Image erosion reduce, the part-structure of corrosion electric power facility is rejected, while also to electric power facility Foreign matter carries out corrosion reduction；If collar plate shape structural element S is reused, as its radius R sizes increase, increase bianry image I₂The rejecting of interstitial spaces and bianry image I₂Image erosion reduce ability；

Step 32：Assuming that image is E after corrosion, and with the collar plate shape structural element S of step 31, image E is expanded to obtain Image I₃；

Bianry image, the bianry image I covered with it with structural element are scanned using collar plate shape structural element S₂Do OR operation such as Fruit is all 0, image I after as a result expanding₃The pixel be 0, be otherwise 1；

Using expansion by the image E after corrosion with identical collar plate shape structural element S to the big of the electric power facility foreign matter after corrosion Small shape restored according to the radius R expansions of collar plate shape structural element S, image I after being expanded₃；If collar plate shape structural elements When plain S is reused, as its radius R sizes increase, the restorability image size of image is restored after increasing corrosion, After image rejects some electrical power facilities section structure and background after excessive erosion expansion process, rough electric power facility foreign matter is obtained Position.

4. it is according to claim 1 a kind of based on morphologic high tension electric tower and high-tension bus-bar foreign matter detecting method, it is special Sign is that the step 4 is as follows：

Step 41：To image I₃Gaussian filtering is carried out, specially：

Assuming that two-dimensional nucleus vector is K, variance σ, image I₃Pixel transverse and longitudinal coordinate be x, y, determine parameter obtain two-dimensional nucleus vector K；

Above formula is the two-dimensional Gaussian function of discretization, determines and carries out image gaussian filtering after parameter obtains two-dimensional nucleus vector, makes to wait for The gray value of pixel and its neighborhood point of filtering is weighted average；

Step 42：With two matrix s of image partial derivative in the x and y direction_xAnd s_yThe finite difference of single order local derviation is obtained, about The gradient of gray value of image is carried out approximate using first difference point：

And calculate image I using the finite difference of single order local derviation₃The amplitude M and direction θ of gradient；

Step 43：Using the amplitude M and direction θ of gradient, non-maxima suppression is carried out to gradient magnitude, finds pixel part most Gray value corresponding to non-maximum point is set to 0 by big value, the Grad of both sides on the Grad and gradient direction of current location It is compared with gradient direction perpendicular to edge direction；A bianry image is obtained after completing non-maxima suppression, non-edge Point gray value is 0, may be the local gray level maximum point gray scale at edge is 128；

Step 44：Edge is detected and connected with dual threashold value-based algorithm, selects two threshold values, an edge graph is obtained according to high threshold Picture, when reaching the endpoint of profile, is found full edge link at profile in high threshold image in 8 neighborhood points of breakpoint The point of sufficient Low threshold collects new edge further according to this point, until whole image edge closure, obtains the inspection of canny operators edge Image I after survey₄。

5. it is according to claim 1 a kind of based on morphologic high tension electric tower and high-tension bus-bar foreign matter detecting method, it is special Sign is that the step 6 is as follows：

Step 61：To step 5 gained image I₅The area statistics for carrying out all connected regions, the area for obtaining connected region are maximum Value and minimum value, and calculate the circularity of connected region；

Step 62：Judge whether connected region maximum area is equal to 0, if so, and enter step 7, otherwise then enter step 63；

Step 63：Judge whether the Maximum Area of connected region and the ratio of the area minimum value of connected region belong to normal model It encloses, if so, enter step 64, otherwise, then return to step 3；

Step 64：Judge whether the circularity of all foreign matters belongs to normal range (NR), if so, entering step 65, otherwise, then returns Step 3；

Step 65：Judge whether the area minimum value of connected region belongs to normal range (NR), if so, entering step 66, otherwise, then Return to step 3；

Step 66：Judge whether the Maximum Area of connected region belongs to normal range (NR), if so, entering step 67, otherwise, then Return to step 3；

Step 67：Judge whether foreign matter number belongs to normal range (NR), if so, enter step 68, otherwise, then return to step 3.

Step 68：The number and location of foreign matter in high tension electric tower and high-tension bus-bar that statistics finally determines.

6. it is according to claim 1 a kind of based on morphologic high tension electric tower and high-tension bus-bar foreign matter detecting method, it is special Sign is that the step 7 is as follows：

Step 71：Foreign matter region center-of-mass coordinate is obtained, deposit array is used as final positioning；

Step 72：After the outline position for obtaining foreign matter region, this location information is used for artwork, by the ash for changing corresponding position Angle value is changed into marked region of tinting.

7. one kind according to claim 1-6 is based on morphologic high tension electric tower and high-tension bus-bar foreign matter detecting method, It is characterized in that, in the step 2, to image I₁It is split to obtain bianry image I₂Before, first to image I₁Carry out RGB figures The channel B extraction process of picture.