CN112241973B - Image analysis boundary tracking representation method and device for intelligent assembly of power transformation equipment - Google Patents

Abstract

The invention discloses a method and a device for tracking and representing an image analysis boundary of an intelligent assembly of transformer equipment, wherein the method comprises the following steps: 1) Carrying out space color conversion on the acquired image data of the intelligent assembly of the power transformation equipment to obtain an image space; 2) Carrying out multi-scale layered processing on an image space comparing image data of the intelligent assembly of the transformer equipment by using multiple scales, and carrying out step-by-step recursive filtering model processing from a high-contrast image to a low-contrast image in each window; 3) And (3) performing binary image conversion on the filtered image data of the intelligent assembly of the power transformation equipment, and outputting a space safety monitoring image or a fire image outline image with a detection frame through a boundary tracking algorithm with scale refinement. The invention overcomes the defects of the edge detection algorithm in processing complex images with high boundary density, selects the boundary tracking algorithm with optimized scale refinement, and improves the performance of the boundary tracking algorithm in the aspects of complex retrieval and complete boundary.

Description

Boundary tracking representation method and device for image analysis of intelligent components of substation equipment

technical field

The invention relates to the technical field of intelligent detection and image processing of power transformation equipment, in particular to a method and device for image analysis and boundary tracking of intelligent components of power transformation equipment.

Background technique

The normal operation of the intelligent components of the substation equipment is very important to ensure the safe operation of the substation. Because the intelligent components of the substation equipment are in a high-voltage complex and changeable environment for a long time, they are easily affected by many factors such as heating, damage, and lightning strikes of electronic devices, causing fires. Major safety accidents bring great losses to personnel safety and the national economy. Image detection method is an important method for fire detection of intelligent components of substation equipment. When a fire occurs in the environment of the intelligent component of the substation equipment, the fire picture of the intelligent assembly of the substation equipment is collected by the camera, and then the boundary tracking method of the image filter is used to remove a large number of non-fire redundant noise interference signals in the picture, and then the fire contour features The boundary sign is extracted to effectively realize fire detection. In practical engineering applications, when a fire occurs in the intelligent components of the substation equipment, the fire feature extraction can be performed through the image filtering boundary tracking representation method to realize the safety alarm of the substation equipment intelligent components, so the boundary tracking representation method of the image filtering is in It is widely used in engineering applications.

The main problem faced by image filter boundary tracking detection in practical application is the detection of object far and near intersections, resulting in incomplete filter denoising and unclear edge contour extraction. When a fire occurs in the substation equipment, the captured pictures contain multi-element interference, such as many buildings, complex substation equipment, too much space information and other factors, which increase the difficulty of image fire removal interference information and contour extraction. Therefore, it is necessary to use Optimized Image Filtering for Boundary Tracing for Fire Detection.

Filtering fire images is the basic problem of fire feature extraction. At present, traditional methods such as average filtering, bilateral filtering and Gaussian filtering are used. However, these methods have many disadvantages. For example, Gaussian filtering can effectively remove noise close to the normal distribution of the image. However, the image contains a large number of interference signals and non-normal information, so the interference information in the image cannot be removed. It is not conducive to the extraction of fire contours.

After filtering and denoising the image, boundary detection algorithms such as Sobel algorithm, Laplacian algorithm, and Canny operator are often used. However, these algorithms have shortcomings. For example, the Bell algorithm is better for image processing with grayscale gradients and more noise, but when the edge of the image is more than one pixel, the Bell algorithm is not very accurate for edge contour positioning.

Contents of the invention

Aiming at the deficiencies of the prior art, the present invention proposes a boundary tracking representation method and device for image analysis of intelligent components of substation equipment, uses the boundary tracking algorithm of image filtering to solve the problem of multi-dimensional denoising and contour extraction of fire image space, and converts the image The filtered boundary tracking algorithm processes the image, which can reveal the essential characteristics of the multi-dimensional space of complex information in the image, facilitate the removal of multiple interfering signals in the space and improve the performance of fire contour feature extraction, thereby improving the image processing performance and realizing the safety monitoring of intelligent components of substation equipment .

In order to achieve the above purpose, the method and device for image analysis and boundary tracking of intelligent components of power transformation equipment designed by the present invention are special in that the method includes the steps of:

1) Perform space color conversion on the collected image data of the intelligent components of the substation equipment to obtain the image space;

2) Use multi-scale comparison to perform multi-scale layered processing on the image space of the image data of the substation equipment intelligent components to obtain accurate high-low contrast range window images, and then step by step from high-contrast images to low-contrast images in each window image Recursive filtering model processing;

3) Convert the image data of intelligent components of substation equipment after filtering to binary image conversion, and then output the image of spatial safety monitoring or the outline of the fire image with detection frame through the boundary tracking algorithm of scale refinement.

Preferably, the spatial color conversion method is to input the collected image data of the intelligent components of the substation equipment into the polar coordinate system of the depth in three directions of space, and the polar coordinate system corresponds to the red, green, blue R, G, B mixed cross colors , when one of the axes rotates along the circumferential direction, every time the specified angle is rotated, the represented color begins to change correspondingly, so that the space color conversion can represent the original information of the monitored multi-dimensional space object; and then through grayscale processing, the Each pixel uses a byte to store the gray value.

Preferably, the filtering model in the step 2) is a trained spatial recursive filtering model, and the operational formula of the model is:

为递推处理图像正交基。Among them, g(x,y) and f(x,y) are the processed image and the original image respectively; h and H are the recursive low contrast and high contrast values of the image space respectively,

Orthonormal basis for recursively processing images.

Preferably, the boundary tracking algorithm of step 3) medium-scale refinement includes:

(1) Carry out thinning and segmentation processing to the outer contour and inner edge of the image data;

(2) The window image is set to a binary image, and the binary image is set to have a width of p pixels and an outer frame of q, where p and q are natural numbers;

(3) Mark the point with pixel p in the window image as b0, mark the point with value p-1 in the window image as c0, and search for m points adjacent to c0 to b0 successively, m is a natural number, and the first A point whose pixel value is p is set to b1, similarly, a point pixel with a value of p-1 is set to c1, and the initialization is b=b1, c=c1.

(4) Mark the m neighborhoods detected clockwise from c to b as k1, ..., km, and mark the subsequent values as p pixels as kn, which is the same as the previous step b=kn, c=kn-1.

(5) Loop step (4), iterate until b stops at b0, and retrieve the next boundary point b1.

The present invention also proposes an image analysis boundary tracking display device for intelligent components of power transformation equipment, which is special in that the device includes:

The conversion module is used to perform space color conversion on the collected image data of the intelligent components of the substation equipment to obtain the image space;

The processing module is used to use multi-scale contrast to perform multi-scale layered processing on the image space of the image data of the substation equipment smart components, to obtain accurate high-low contrast range window images, and then pass from high-contrast images to low-contrast images in each window image Perform step-by-step recursive filtering model processing;

The output module is used to convert the filtered image data of the substation equipment intelligent components into a binary image, and then output a space safety monitoring image or a fire image outline image with a detection frame through a scale-refined boundary tracking algorithm.

Further, the conversion module includes:

The input module is used to input the collected image data of the intelligent components of the substation equipment into the polar coordinate system of depth in three directions of space;

The polar coordinate system conversion module is used to perform spatial color conversion in the polar coordinate system on the original information of the multi-dimensional space object in the image data, and the corresponding red, green and blue R, G, and B mixed cross colors in the polar coordinate system, when the When an axis rotates along the circumferential direction, every time the specified angle is rotated, the represented color starts to change correspondingly;

The grayscale processing module is used to perform grayscale processing on the image data after space color conversion to obtain an image space.

Further, the processing module includes:

The multi-scale layered processing module is used to perform multi-scale layered processing on the image space of the image data of the intelligent component of the substation equipment by multi-scale comparison, and obtain an accurate high-low contrast range window image;

The step-by-step recursive filtering model processing module is used to perform step-by-step recursive filtering model processing from high-contrast images to low-contrast images in each window image.

Further, the output module includes:

The binary image conversion module is used to perform binary image conversion on the image data of the intelligent component of the substation equipment after filtering;

The boundary tracking operation module is used to output a space security monitoring image or a fire image outline image with a detection frame through a scale-refined boundary tracking algorithm.

The present invention further proposes a device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein,

The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor, so that the at least one processor can perform the above-mentioned image analysis applicable to the intelligent components of the substation equipment Boundary tracing representation.

The present invention further proposes a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, implements the above-mentioned image analysis boundary tracking representation method for an intelligent component of a substation equipment.

The beneficial effects of the present invention are:

1. The present invention provides spatial hierarchical filtering for smoke and fire images when a fire occurs in the intelligent components of the substation equipment, and then uses an optimized boundary tracking method to extract the fire contour, thereby improving the fire detection performance and realizing the detection of the intelligent components of the substation equipment. security monitoring.

2. The spatial hierarchical recursive variation filtering model proposed by the present invention can improve the filtering performance of spatial cross information; and optimize the boundary tracking algorithm, thereby improving the feature detection performance of fire contours.

3. In order to improve the monitoring performance of fire contour features, the present invention overcomes the shortcomings of the edge detection algorithm in dealing with complex images with high boundary density, and selects the boundary tracking algorithm with optimized scale refinement, thereby improving the performance of the boundary tracking algorithm in terms of complex retrieval and complete boundaries. performance.

4. The present invention uses the boundary tracking algorithm of image filtering to solve the problem of multi-dimensional denoising and contour extraction of fire image space, and removes complex interference redundant signals in fire images and improves the detection effect of edge contour features; the boundary tracking algorithm of image filtering Image processing can reveal the essential characteristics of multi-dimensional space of complex information in the image, facilitate the removal of multiple interfering signals in space, improve the performance of fire contour feature extraction, improve image processing performance, and realize the safety monitoring of intelligent components of substation equipment.

Description of drawings

Fig. 1 is the overall flowchart of the method of the present invention.

FIG. 2 is a flow chart of a spatial color conversion method.

Fig. 3 is a flow chart of spatial hierarchical recursive variation filtering processing.

FIG. 4 is a flow chart of Embodiment 1 of the present invention.

Fig. 5 is a schematic diagram of the implementation effect of the present invention.

Detailed ways

The present invention will be described in further detail below in conjunction with the accompanying drawings and specific examples, but the embodiments of the present invention are not limited thereto.

As shown in Figure 1, the image analysis boundary tracking representation method suitable for intelligent components of substation equipment proposed by the present invention first undergoes spatial color conversion according to the collected images, and then denoises disturbances through spatial layered recursive variation filtering Finally, the feature extraction of the fire contour is carried out by optimizing the boundary tracking algorithm, so as to realize the safety monitoring of the substation equipment, which specifically includes the following steps:

1) Transform the image data of the intelligent components of the substation equipment into space and color to obtain the image space.

The flow chart of the space color conversion method is shown in 2. The collected image data of the intelligent components of the substation equipment is input into the polar coordinate system of the depth in three directions of space, and the polar coordinate system corresponds to the mixture of red, green and blue R, G, and B. Cross color, when one of the axes rotates along the circumferential direction, every time the specified angle is rotated, the represented color starts to change correspondingly, so that the space color conversion can represent the original information of the monitored multi-dimensional space object; and then through grayscale processing , each pixel uses one byte to store the gray value, the gray range is 0-255, which can not only avoid image band distortion, but also prevent the loss of important object information.

2) Use multi-scale comparison to perform multi-scale layered processing on the image space of the image data of the substation equipment intelligent components to obtain the window image of the precise high and low contrast range, and realize the qualitative analysis of the edge component information of the object in the window, and then through each window In the image, the step-by-step recursive filtering model is processed from the high-contrast image to the low-contrast image to effectively remove the spatial cross interference filtering.

The images collected by real-time monitoring of intelligent components of substation equipment have the difficulties of multi-dimensional space interlacing and the complexity of indoor and outdoor obstacles. Therefore, removing spatial obstacles and interfering filtering is an important key to feature extraction of image contours. Because the working environment of the smart components of the substation equipment is duplicated, and there are too many obstacles, it is very important to remove the interference of obstacles. The changes in the monitoring environment are complex and diverse. For example, the early detection of fire is characterized by smoke, and then the characteristics of flames change. For example, objects in the monitoring screen cross from far to near, and there are too many obstacles between them. Conditions make filtering difficult.

Therefore, the flow chart of the training method for image space recursive variation filtering is shown in Figure 3: the images collected by the monitoring equipment through space color conversion are processed by multi-scale and layered processing of the image space by multi-scale contrast, and accurate high-low contrast range images are obtained. window, to achieve qualitative analysis of the quality of the object edge component information in the window, and then perform step-by-step recursive filtering model processing through the contrast images from high H to low h in each window to effectively remove spatial cross-interference filtering. The spatial recursive variation filtering model is shown in Table 1.

Table 1 Filtering model representation

为小波基；空间递推滤波模型中h、H分别为图像空间递推低对比和高对比值，

为递推处理图像正交基。In the formula shown in Table 1, g(x, y) and f(x, y) are the processed image and the original image respectively; , (x, y) is the point in the image, med() is the median filter reduction function; in the mean filter model, M is the total number of pixels; in the threshold filter model, w _{i, j} are image gradient values, and λ is Optimum threshold; in the wavelet filter model, A _j-1 is the low-frequency sparseness of the J-1 layer, D _n,j is the high-frequency coefficient of the J layer,

is the wavelet base; in the spatial recursive filtering model, h and H are the image spatial recursive low-contrast and high-contrast values respectively,

Orthonormal basis for recursively processing images.

3) Convert the image data of intelligent components of substation equipment after filtering to binary image conversion, and then output the spatial safety monitoring image or the fire image contour image with detection frame through the boundary tracking algorithm with scale refinement. In order to improve the monitoring performance of fire contour features, overcome the disadvantages of edge detection algorithm in processing complex and high-density images, and optimize the scale-refined boundary tracking algorithm, so as to improve the performance of boundary tracking algorithm in terms of complex retrieval and complete boundaries. The process is as follows:

(1) Carry out scale refinement segmentation processing on the outer contour and inner edge of the image data;

(2) The window image is set as a binary image, and the binary image is set as an outer frame with a width of 1 pixel and a value of 0;

(3) Set the point of 1 pixel in the image as b0, mark the point with a value of 0 on its adjacent west side as c0, search for 8 points adjacent to c0 to b0 in turn, and set the point with the first pixel value of 1 Mark it as b1, similarly set the 0-value pixel point as c1, and initialize it as b=b1, c=c1.

(4) Mark the 8 neighborhoods detected clockwise from c to b as k1, ..., k8, and mark the subsequent value as 1 pixel as kn, which is the same as the previous step b=kn, c=kn-1.

(5) Loop step (4), iterate until b stops at b0, and retrieve the next boundary point b1.

Embodiment 1: Fire image detection method for substation equipment

The invention belongs to an image processing method. The input is the collected image monitored by the intelligent component of the substation equipment, and the output is the extracted fire contour feature. The flowchart of the algorithm is shown in Figure 4. The biggest difference from the existing technology is that the existing image filters such as median filter, threshold filter and wavelet filter cannot effectively denoise the spatial interference information of complex images, but the paper proposes a spatial layered recursive variation filter model that can improve the spatial Cross-information filtering performance. And the boundary tracking algorithm is optimized to improve the performance of detecting fire contour features. The application effect is shown in Figure 5, which includes four maps, namely, the fire map of the intelligent component of the substation equipment, the spatial color conversion model map, the spatial layered recursive variation filter map, and the fire contour map extracted by the optimal boundary algorithm.

The invention also proposes an image analysis boundary tracking and display device for intelligent components of electric substation equipment, which includes a conversion module, a processing module and an output module. in,

The conversion module is used to perform space color conversion on the collected image data of the intelligent components of the substation equipment to obtain the image space;

The processing module is used to use multi-scale contrast to perform multi-scale layered processing on the image space of the image data of the substation equipment smart components, to obtain accurate high-low contrast range window images, and then pass from high-contrast images to low-contrast images in each window image Perform step-by-step recursive filtering model processing;

The output module is used to convert the filtered image data of the substation equipment intelligent components into a binary image, and then output a space safety monitoring image or a fire image outline image with a detection frame through a scale-refined boundary tracking algorithm.

The conversion module includes:

The input module is used to input the collected image data of the intelligent components of the substation equipment into the polar coordinate system of depth in three directions of space;

The polar coordinate system conversion module is used to perform spatial color conversion in the polar coordinate system on the original information of the multi-dimensional space object in the image data, and the corresponding red, green and blue R, G, and B mixed cross colors in the polar coordinate system, when the When an axis rotates along the circumferential direction, every time the specified angle is rotated, the represented color starts to change correspondingly;

The grayscale processing module is used to perform grayscale processing on the image data after space color conversion to obtain an image space.

The processing modules include:

The multi-scale layered processing module is used to perform multi-scale layered processing on the image space of the image data of the intelligent component of the substation equipment by multi-scale comparison, and obtain an accurate high-low contrast range window image;

The step-by-step recursive filtering model processing module is used to perform step-by-step recursive filtering model processing from high-contrast images to low-contrast images in each window image.

The output modules include:

The binary image conversion module is used to perform binary image conversion on the image data of the intelligent component of the substation equipment after filtering;

The boundary tracking operation module is used to output a space security monitoring image or a fire image outline image with a detection frame through a scale-refined boundary tracking algorithm.

The present invention further proposes a device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein,

The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor, so that the at least one processor can perform the above-mentioned image analysis applicable to the intelligent components of the substation equipment Boundary tracing representation.

The present invention further proposes a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, implements the above-mentioned image analysis boundary tracking representation method for an intelligent component of a substation equipment.

The content not described in detail in this specification belongs to the prior art known to those skilled in the art.

Finally, it should be noted that the above specific embodiments are only used to illustrate the technical solution of the patent and not to limit it. Although the patent has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the patent can be Modifications or equivalent replacements without departing from the spirit and scope of the technical solution of this patent shall be covered by the claims of this patent.

Claims (9)

1. A method for representing intelligent assembly image analysis boundary tracking of transformer equipment is characterized by comprising the following steps: the method comprises the following steps:

1) Carrying out space color conversion on the acquired image data of the intelligent assembly of the power transformation equipment to obtain an image space;

2) Carrying out multi-scale layered processing on an image space for comparing image data of the intelligent assembly of the transformer equipment by using multiple scales to obtain window images in a precise high-low contrast range, and then carrying out step-by-step recursive filtering model processing from a high-contrast image to a low-contrast image in each window image; the filtering model is a trained space recursive filtering model, and the operational formula of the model is as follows:

wherein g (x, y) and f (x, y) are respectively a processed image and an original image; h. h are the image space recursive low and high contrast values respectively,

processing the image orthogonal basis for recursion;

3) And (3) performing binary image conversion on the filtered intelligent assembly image data of the power transformation equipment, and outputting a space safety monitoring image or a fire image outline image with a detection frame through a boundary tracking algorithm with scale refinement.

2. The method for image analysis boundary tracking representation of a power transformation device intelligent component according to claim 1, wherein: the space color conversion method comprises the steps that collected image data of the intelligent assembly of the power transformation equipment are input into a polar coordinate system with three spatial directions of depth, the colors of red, green and blue R, G and B in the polar coordinate system are mixed and crossed correspondingly, when one shaft rotates along the circumferential direction, the color represented by the rotation starts to be correspondingly converted every time the shaft rotates by a specified angle, so that the original information representation of a monitored multi-dimensional space object is carried out by space color conversion; and then, storing the gray value of each pixel point by using one byte through gray processing.

3. The method for image analysis boundary tracking representation of a power transformation device intelligent component according to claim 1, wherein: the boundary tracking algorithm of the mesoscale refinement in the step 3) comprises the following steps:

(1) Carrying out scale thinning segmentation processing on the external contour and the internal edge of the image data;

(2) Setting the window image as a binary image, setting the binary image as an outer frame with the width of p pixels and the value of p-1, wherein p is a natural number;

(3) Marking a point with a pixel of p in a window image as b0, marking a point with a median of p-1 in the window image as c0, sequentially and adjacently searching m points from c0 to b0, wherein m is a natural number, setting a point with a first pixel value of p as b1, and setting a point with a value of p-1 as c1 in the same way, and initializing the point with b = b1 and c = c1;

(4) Detecting m neighborhoods clockwise from c to b as k1, \ 8230;, km, retrieving the subsequent value as p pixels and as kn, and b = kn, c = kn-1 as the previous step;

(5) And (4) looping the step until b stops at the position b0, and retrieving the next boundary point b1.

4. A boundary tracking representation device for intelligent assembly image analysis of power transformation equipment is characterized in that: the device comprises:

the conversion module is used for carrying out space color conversion on the acquired image data of the intelligent assembly of the power transformation equipment to obtain an image space;

the processing module is used for carrying out multi-scale layered processing on an image space for comparing image data of the intelligent assembly of the transformer equipment by using multiple scales to obtain window images in a precise high-low contrast range, and then carrying out step-by-step recursive filtering model processing from a high-contrast image to a low-contrast image in each window image; the filtering model is a trained space recursive filtering model, and the operational formula of the model is as follows:

wherein g (x, y) and f (x, y) are respectively a processed image and an original image; h. h are the image space recursive low and high contrast values respectively,

processing the image orthogonal basis for recursion;

and the output module is used for performing binary image conversion on the filtered intelligent assembly image data of the power transformation equipment, and outputting a space safety monitoring image or a fire image outline image with a detection frame through a boundary tracking algorithm with scale refinement.

5. The apparatus according to claim 4, wherein the apparatus comprises: the conversion module includes:

the input module is used for inputting the acquired image data of the intelligent assembly of the power transformation equipment to a polar coordinate system with three spatial depths;

the polar coordinate system conversion module is used for carrying out space color conversion on original information of a multi-dimensional space object in image data in a polar coordinate system, wherein the polar coordinate system is used for mixing cross colors corresponding to R, G and B of red, green and blue, and when one of the polar coordinate system rotates along the circumferential direction, the expressed color starts to be correspondingly converted every time an appointed angle is rotated;

and the gray processing module is used for carrying out gray processing on the image data after the space color conversion to obtain an image space.

6. The apparatus according to claim 4, wherein the apparatus comprises: the processing module comprises:

the multi-scale layering processing module is used for carrying out multi-scale layering processing on the image space of the image data of the intelligent assembly of the power transformation equipment through multi-scale comparison to obtain a window image of an accurate high-low comparison range;

and the step-by-step recursive filtering model processing module is used for performing step-by-step recursive filtering model processing from the high-contrast image to the low-contrast image in each window image.

7. The apparatus according to claim 4, wherein the apparatus comprises: the output module includes:

the binary image conversion module is used for performing binary image conversion on the filtered intelligent assembly image data of the power transformation equipment;

and the boundary tracking operation module is used for outputting a space safety monitoring image or a fire image outline image with a detection frame through a boundary tracking algorithm with scale refinement.

8. An apparatus, comprising:

at least one processor; and the number of the first and second groups,

a memory communicatively coupled to the at least one processor; wherein,

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 3.

9. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the method of any one of claims 1 to 3.

Images