CN113284076A - FloodFill-based high-speed rail contact net current-carrying ring fracture abnormity detection method - Google Patents
FloodFill-based high-speed rail contact net current-carrying ring fracture abnormity detection method Download PDFInfo
- Publication number
- CN113284076A CN113284076A CN202110827688.0A CN202110827688A CN113284076A CN 113284076 A CN113284076 A CN 113284076A CN 202110827688 A CN202110827688 A CN 202110827688A CN 113284076 A CN113284076 A CN 113284076A
- Authority
- CN
- China
- Prior art keywords
- current
- carrying ring
- floodfill
- picture
- pixels
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 29
- 238000000034 method Methods 0.000 claims abstract description 31
- 238000001914 filtration Methods 0.000 claims abstract description 16
- 238000005286 illumination Methods 0.000 claims abstract description 8
- 230000011218 segmentation Effects 0.000 claims abstract description 4
- 238000004422 calculation algorithm Methods 0.000 claims description 27
- 230000002159 abnormal effect Effects 0.000 claims description 9
- 230000005856 abnormality Effects 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 4
- 230000001629 suppression Effects 0.000 claims description 4
- 230000002146 bilateral effect Effects 0.000 claims description 3
- 239000003086 colorant Substances 0.000 claims 1
- 238000009827 uniform distribution Methods 0.000 claims 1
- 238000012545 processing Methods 0.000 abstract description 12
- 238000010586 diagram Methods 0.000 abstract description 6
- 238000003745 diagnosis Methods 0.000 abstract description 2
- 230000009466 transformation Effects 0.000 abstract description 2
- 230000000694 effects Effects 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 4
- 238000004043 dyeing Methods 0.000 description 4
- 238000003708 edge detection Methods 0.000 description 4
- 241000872198 Serjania polyphylla Species 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 150000001875 compounds Chemical class 0.000 description 2
- 238000009795 derivation Methods 0.000 description 2
- 230000002708 enhancing effect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000003137 locomotive effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration using histogram techniques
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Signal Processing (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a FloodFill-based high-speed rail contact net current-carrying ring fracture abnormity detection method, and belongs to the technical field of high-speed rail current-carrying ring abnormity diagnosis. The filling detection method based on FloodFill provided by the invention firstly preprocesses the image to reduce the influence of noise and low illumination, adopts image filtering to the current-carrying ring, and improves the quality of the image through histogram equalization; then, carrying out threshold segmentation processing on the picture, and processing the picture to only contain two kinds of pixels, namely black and white; obtaining a skeleton diagram of the current-carrying ring by using distance transformation, horizontal projection and vertical projection; filling the pictures by using FloodFill; and finally, calculating the ratio of the pixel values of the white pixel and the whole picture to determine whether the current-carrying ring has fracture abnormity. The method can realize intelligent detection of the 4C detection of the vehicle-mounted flow ring fracture abnormity, and has important application value for solving the problems of unbalanced positive and negative samples and non-uniform current-carrying ring states.
Description
Technical Field
The invention relates to a FloodFill-based high-speed rail contact net current-carrying ring fracture abnormity detection method, and belongs to the technical field of high-speed rail current-carrying ring abnormity diagnosis.
Background
Once a high-speed rail contact net breaks down, the normal operation of the high-speed rail is influenced. The current carrying ring is used as an important part of a high-speed rail contact net and has the function of suspending a contact line and a catenary, so that the elasticity and sag of the contact net are guaranteed under the condition that a strut is not added to the contact net, and the contact net and a pantograph of an electric locomotive are guaranteed to take current well. If the current-carrying ring breaks abnormally, the power supply of the train is caused to be in a problem, and the safe operation of the railway is greatly influenced.
The invention designs and realizes a visual-based current-carrying ring fracture abnormity detection method aiming at a current-carrying ring picture shot by a 4C detection vehicle. And the current-carrying ring fracture abnormity detection is to judge whether the current-carrying ring is closed, if the current-carrying ring is closed, the current-carrying ring is not abnormal, and if the current-carrying ring has an opening, the current-carrying ring is judged to be abnormal in fracture.
Disclosure of Invention
The invention provides a FloodFill-based high-speed rail contact net current-carrying ring fracture abnormity detection method, which comprises the steps of firstly preprocessing an image to reduce the influence of noise and illumination, filtering the current-carrying ring by adopting an image, and balancing a histogram to improve the quality of the image; then, carrying out threshold segmentation processing on the picture, and processing the picture to only contain two kinds of pixels, namely black and white; obtaining a skeleton diagram of the current-carrying ring by using distance transformation, horizontal projection and vertical projection; filling the pictures by using FloodFill; and finally, calculating the ratio of the pixel values of the white pixel and the whole picture to determine whether the current-carrying ring is abnormally broken or not, and realizing intelligent detection of the abnormal breakage of the 4C detection vehicle-mounted flow ring. The method has important application value for solving the problems of unbalanced positive and negative samples and non-uniform current-carrying ring states.
In order to achieve the purpose, the invention adopts the following technical scheme:
a FloodFill-based high-speed rail contact net current-carrying ring fracture abnormality detection method comprises the following steps:
preprocessing an input current-carrying ring picture, zooming the picture to 256 × 256 resolution, keeping the original proportion of a current-carrying ring while ensuring the calculation efficiency of an algorithm, and removing the influence of illumination and noise in the picture so as to improve the detection effect;
performing threshold segmentation on the preprocessed picture, wherein the processed picture only has black and white pixels, the black represents a background, and the white represents a foreground, namely a current-carrying ring;
extracting the contour of the current-carrying ring by using a Canny algorithm;
filling white (RGB values are R:255, G:255 and B: 255) from the upper left corner, the upper right corner, the lower left corner and the lower right corner of the outline picture of the current-carrying ring in sequence by using a FloodFill filling method;
calculating the number of the filled white pixels and the ratio of the white pixels to all pixels of the picture, and judging whether the current-carrying ring is broken abnormally according to the ratio;
further, the influence of illumination is removed by adopting histogram equalization, wherein the histogram equalization is to convert the histogram of the original image into a uniformly distributed form and increase the dynamic range of pixel gray values, so that the effect of enhancing the overall contrast of the image is achieved, and the detection precision is improved. Histogram equalization can be expressed by the following equation,
wherein MN is the total number of image pixels,is a gray scale ofThe number of the pixels of (a) is,l is the number of image gray levels by which the gray level value of a pixel in the output image can be determined from the gray level of a pixel in the input imageIs mapped asAnd then obtaining the compound.
Removing picture noise using gaussian filtering, gaussian filtering formulaU, v denote the coordinates of the pixel,is the standard deviation of a normal distribution. Gaussian filtering:
1. the central element of the relevant kernel is moved so that it is located directly above the pixels to be processed of the input image.
2. The pixel values of the input image are multiplied by the correlation kernel as weights.
3. And adding the results obtained in the above steps as output.
Furthermore, bilateral filtering can be adopted to remove picture noise, so that the fuzzy edge information of the image after normal Gaussian filtering can be kept clear, and the image edge is smoother. The concrete formula is as follows
As a result of the current pixel weight value,as the information of the current pixel is,is the current pixel domain mean value;as information on the position of the current pixel,as the average position information of the current pixel,andrespectively, the standard deviation of the current pixel information and the current pixel position information.
Further, the picture is divided by a fixed threshold value, the fixed threshold value is used for separating the current-carrying ring from the background in the picture of the current-carrying ring, and the target and the background have strong contrast,Is a threshold value, the following formula is satisfied when the image is divided:
wherein the threshold valueSetting the gray value of the pixel to be 5, setting all the pixels with the gray value of more than or equal to 5 to be 255, and representing the current-carrying ring pixel; all the others are set to 0, indicating background.
Further, the Canny algorithm is used to extract the contour of the current-carrying ring, and the steps are preferably as follows:
firstly, the gradient amplitude direction of the image is calculated by using first-order partial derivative finite difference, Roberts operators of edges are searched by using local difference operators, and the sharpness of the edges is determined by the gradient of the image gray.
And carrying out non-maximum suppression on the gradient amplitude, and reserving the point with the maximum local gradient value.
The edges are detected and connected using hysteresis thresholds, and the contours of the current-carrying rings are extracted.
Further, the picture is filled using the FloodFill method. A four-way FloodFill algorithm, which is specifically as follows, or an eight-way FloodFill algorithm may be used: filling white from upper left, upper right, lower left and lower right corners of the picture by using a FloodFill filling method, and searching pixel points by using a four-way FloodFill algorithmIf not, filling the four adjacent pixel points, and continuously searching the four connected pixels until the closed area is completely filled with new color.
The eight-pass FloodFill algorithm is specifically as follows: filling the pictures from the upper left point, the upper right point, the lower left point and the lower right point of the pictures by using a FloodFill filling method in sequence by using white, and calculating pixel points by using an eight-way FloodFill algorithmAnd recursively looks for their octal connected pixel fill until the region is completely filled with new color.
If the current-carrying ring is broken, the whole picture is changed into white after filling; if the current-carrying ring is not broken, namely the current-carrying ring is annular, after filling, the middle of the current-carrying ring cannot be filled with white, and the part is black with background color.
And calculating the number of white pixel values in the picture after each filling, and calculating the ratio of the number of white pixels to the total number of pixels of the picture, wherein the fracture abnormity occurs in the current-carrying ring as long as the ratio is more than or equal to 0.98 at any time in multiple filling.
According to the method, the intelligent detection of the 4C detection of the vehicle-mounted flowing ring fracture abnormity can be realized. In addition, the method has important application value for solving the problems of unbalance of positive and negative samples and non-uniform current-carrying ring states.
Drawings
FIG. 1 is a flow chart of a method according to a first embodiment of the present invention;
FIG. 2 is a schematic diagram of four sectors as described in embodiments one and two of the present invention;
FIG. 3 is a diagram of the first and second embodiments of the present inventionA schematic of a neighborhood;
FIG. 4 is a schematic diagram of the eight-pass process in accordance with a first embodiment of the present invention;
FIG. 5 is a flowchart of a method according to a second embodiment of the present invention;
FIG. 6 is a schematic diagram of the four-pass method according to the second embodiment of the present invention.
Detailed Description
In order to clearly illustrate the technical features of the present invention, the present invention will be explained in detail by the following embodiments and the accompanying drawings. The following disclosure provides many different embodiments, or examples, for implementing different features of the invention. To simplify the disclosure of the present invention, the components and arrangements of specific examples are described below. Furthermore, the present invention may repeat reference numerals and/or letters in the various examples. This repetition is for the purpose of simplicity and clarity and does not in itself dictate a relationship between the various embodiments and/or configurations discussed. Descriptions of well-known components and processing techniques and procedures are omitted so as to not unnecessarily limit the invention.
Example one
As shown in fig. 1, the method for detecting the abnormal breakage of the current-carrying ring of the high-speed rail contact network based on the FloodFill comprises the following steps:
s1, preprocessing the input current-carrying ring picture, scaling the picture to 256 × 256 resolution, firstly, using histogram equalization to remove the influence of illumination on the picture, and then using Gaussian filtering to remove noise in the picture so as to improve the detection effect; the histogram equalization is to transform the histogram of the original image into a uniformly distributed form and increase the dynamic range of the pixel gray value, thereby achieving the effect of enhancing the overall contrast of the image and improving the detection precision. Histogram equalization can be expressed by the following equation,
wherein MN is the total number of image pixels (MN =65535),is a gray scale ofL is the number of image grey levels (e.g. L =256 for an 8-bit image), by which the grey value of a pixel in the output image can be determined from the grey value of the pixel in the input imageIs mapped asAnd then obtaining the compound.
Removing picture noise using gaussian filtering, gaussian filtering formulaU, v denote the coordinates of the pixel,is the standard deviation of a normal distribution. Gaussian filtering: (1) the central element of the relevant kernel is moved so that it is located directly above the pixels to be processed of the input image. (2) The pixel values of the input image are multiplied by the correlation kernel as weights. (3) And adding the results obtained in the above steps as output.
Wherein the content of the first and second substances,as in the imageThe gray-scale value of the point or points,is the gaussian filtered value for that point.
S2, carrying out threshold processing on the preprocessed picture, separating the current-carrying ring from the background in the picture of the current-carrying ring, wherein the method adopted in the embodiment is to divide the current-carrying ring by a fixed threshold value and set the original image as the target and the background have stronger contrast,Is a threshold value, the following formula is satisfied when the image is divided:
wherein the threshold valueSet to 5, all of the pixels having a gradation value of 5 or more are 255, indicating thatA current-carrying ring pixel; all the others are set to 0, indicating background. The processed picture only has black and white pixels, wherein the black pixel represents the background, and the white pixel represents the foreground, namely the current-carrying ring;
s3, extracting the contour of the current carrying ring by using a Canny algorithm, and the steps are as follows: first, the gradient magnitude direction of the image is calculated by using a first-order partial derivative finite difference, and a first-order differential operator is generally completed by convolution through a spatial differential operator, but actually derivation in the digital image is performed by using a differential approximate differential.
The gradient corresponds to the first derivative and the gradient operator is the first derivative operator. For a continuous functionIt is in positionThe gradient may be represented as a vector
For digital images, the derivatives can be approximated by differences, and the gradient can be expressed as:
in practice, a small region template convolution is often used to approximate the calculation. To pairAndone template each, so two templates are required to be combined to form a gradient operator. Depending on the size of the template, where the element (coefficient) values differ, a large number of templates can be proposed, constituting a large number of detection operators.
The operator proposed by Roberts is a method of exploiting local differencesThe sub-operators search operators of edges, and the sharpness of the edges is determined by the gradient of the image gray level. The gradient is a vector of the magnitude of the gradient,indicating the fastest direction and number of gray scale changes.
The simplest edge detection operator is therefore to approximate the gradient operator with the vertical and horizontal difference of the image:
the vector of the above formula is calculated for each pixel, its absolute value is calculated, and then compared with a threshold, and by using this idea, the Roberts cross operator is obtained:
the above equation can provide a better invariant edge orientation. For edges of equal length but different orientations, the resultant amplitude variation obtained by applying the Roberts maximum operator is smaller than that obtained by applying the Roberts crossover operator. It is clear that the Roberts first differential is not alongAxial direction andaxial differential, but taking rotationThe sum of the differential values in two directions of 45 degrees.
And carrying out non-maximum suppression on the gradient amplitude, and reserving the point with the maximum local gradient value. The above steps only result in a global gradient and are not sufficient to determine the edge, so to determine the edge, the point where the local gradient is maximal must be preserved, while suppressing non-maxima.
Using simple edge calculationsThe sub-vector can obtain vertical and horizontal direction edges, and the direction can be rotated by using Roberts cross operatorEdges in two directions of 45 degrees, therefore, we can combine the two operators to obtain edge information in four directions. As shown in FIGS. 2 and 3, the four directional edges may be represented by a pie chart, labeled with sector numbersToTo indicate the pixels that compute the edge utilization in four directions, which corresponds toFour possible combinations of neighborhoods. Sector 0 corresponds to pixels 8 and 4 in fig. 3, sector 1 corresponds to pixels 7 and 3 in fig. 3, sector 2 corresponds to pixels 6 and 2 in fig. 3, and sector 3 corresponds to pixels 5 and 1 in fig. 3.
At each point, the central pixel of the neighborhoodCompared to two pixels along the gradient line. If it is notThe gradient value of (a) is not larger than the gradient values of two adjacent pixels along the gradient line, then. Namely, it is。
Edges are detected and connected using hysteresis thresholds. The choice of the threshold size affects the quality of the detected edge. This step is to convert the image into a non-black, i.e. white, edge map, defining two parameters:and。
(1) if the amplitude of the point is lower than Low, the point is not an edge point, is removed and is set to be black.
(2) If the amplitude of the point is higher than High, it is an edge point, and the point is set to white.
(3) If the amplitude of the point is between Low and High, then:
a) if the point is directly adjacent to a point having a magnitude greater than High or the point may be indirectly adjacent through a point also between Low and High and a point having a magnitude greater than High. The point is considered to be on the edge, is an edge point, and is set to white.
b) Otherwise, the edge point is not considered to be the edge point, and the color is set to be black.
Through the processing, the final image is converted into an edge map which is not black or white and is suitable for computer processing, and the purpose of edge detection is achieved, namely the edge map of the current-carrying ring is obtained.
And S4, filling the picture by using a FloodFill filling method. The FloodFill method fills a communication area with a certain color, and achieves different filling effects by setting the upper limit and the lower limit of a connectable pixel and a communication mode. The FloodFill algorithm accepts three parameters: the start node, the target color and the alternate color. The algorithm traverses all nodes to find nodes connected to the starting node (connected by a path of the target color) and then changes their color to the alternate color.
In this embodiment, an eight-connected filling algorithm is adopted, as shown in fig. 4, that is, a pixel point is foundUp, down, left, right, top left, bottom left, top right, bottom right eight neighboring pixels, if not filled, fill them, and recursively find their eight connected pixel fills until the region is completely filled with the new color.
The algorithm comprises the following steps:
(1) and finding a pixel point which is not dyed, dyeing the pixel point into the designated color, and finishing the algorithm if the pixel point is not dyed.
(2) Initializing an empty queue and inserting the pixel points of the first step into the queue.
(3) And continuously obtaining the value of the head element of the queue and popping up, and point-dyeing the uncolored pixel adjacent to the head element of the queue into the specified color and adding the uncolored pixel into the queue.
(4) And repeating the first step until all the pixel points are dyed, and finishing the algorithm.
Filling the pictures by white (RGB values are R:255, G:255 and B: 255) from upper left, upper right, lower left and lower right corners of the pictures in sequence by using a FloodFill filling method; if the current-carrying ring is broken, the whole picture is changed into white after filling; if the current-carrying ring is not broken, namely the current-carrying ring is annular, after filling, the middle of the current-carrying ring cannot be filled with white, and the part is black with background color.
And S5, calculating the number of white pixel values in the picture after each filling, calculating the ratio of the number of white pixels to the total number of pixels of the picture (65536), and if the ratio is more than or equal to 0.98 in any one filling of four times, determining that the carrier ring has a fracture abnormity.
Example two
As shown in fig. 5, a method for detecting abnormal breakage of a current-carrying ring of a high-speed rail contact network based on FloodFill comprises the following steps:
s1, adjusting the input picture to 256 × 256 resolution, using histogram equalization to solve the effect of illumination change in the current-carrying ring not affected by the force anomaly detection (same as in the first embodiment), and filtering the image to remove the picture noise.
And the bilateral filtering is used for removing picture noise, so that the fuzzy edge information of the image after normal Gaussian filtering is kept clear, and the image edge is smoother. The concrete formula is as follows,
as a result of the current pixel weight value,as the information of the current pixel is,is the current pixel domain mean value;as information on the position of the current pixel,as the average position information of the current pixel,andrespectively, the standard deviation of the current pixel information and the current pixel position information.
S2, carrying out threshold processing on the preprocessed picture, separating the current-carrying ring from the background in the picture of the current-carrying ring, wherein the method adopted in the embodiment is to divide the current-carrying ring by a fixed threshold value and set the original image as the target and the background have stronger contrast,Is a threshold value, the following formula is satisfied when the image is divided:
wherein the threshold valueSetting the gray value of the pixel to be 5, setting all the pixels with the gray value of more than or equal to 5 to be 255, and representing the current-carrying ring pixel; all the others are set to 0, indicating background. The processed picture only has black and white pixels, wherein the black pixel represents the background, and the white pixel represents the foreground, namely the current-carrying ring;
s3, extracting the contour of the current carrying ring by using a Canny algorithm, and the steps are as follows:
first, the gradient magnitude direction of the image is calculated by using a first-order partial derivative finite difference, and a first-order differential operator is generally completed by convolution through a spatial differential operator, but actually derivation in the digital image is performed by using a differential approximate differential.
The gradient corresponds to the first derivative and the gradient operator is the first derivative operator. For a continuous functionIt is in positionThe gradient may be represented as a vector
For digital images, the derivatives can be approximated by differences, and the gradient can be expressed as:
in practice, a small region template convolution is often used to approximate the calculation. To pairAndone of the templates is used for each,two templates are required to be combined to form a gradient operator. Depending on the size of the template, where the element (coefficient) values differ, a large number of templates can be proposed, constituting a large number of detection operators.
The operator proposed by Roberts is an operator that finds edges using local difference operators, the sharpness of the edges being determined by the gradient of the image grey scale. The gradient is a vector of the magnitude of the gradient,indicating the fastest direction and number of gray scale changes.
The simplest edge detection operator is therefore to approximate the gradient operator with the vertical and horizontal difference of the image:
the vector of the above formula is calculated for each pixel, its absolute value is calculated, and then compared with a threshold, and by using this idea, the Roberts cross operator is obtained:
the above equation can provide a better invariant edge orientation. For edges of equal length but different orientations, the resultant amplitude variation obtained by applying the Roberts maximum operator is smaller than that obtained by applying the Roberts crossover operator. It is clear that the Roberts first differential is not alongAxial direction andaxial differential, but taking rotationThe sum of the differential values in the two directions of degrees.
And carrying out non-maximum suppression on the gradient amplitude, and reserving the point with the maximum local gradient value. The above steps only result in a global gradient and are not sufficient to determine the edge, so to determine the edge, the point where the local gradient is maximal must be preserved, while suppressing non-maxima.
As shown in FIGS. 2 and 3, four sectors are numberedToCorrespond toFour possible combinations of neighborhoods. At each point, the central pixel of the neighborhoodIf compared to two pixels along the gradient lineThe gradient value of (a) is not larger than the gradient values of two adjacent pixels along the gradient line, then. Namely, it is
Edges are detected and connected using hysteresis thresholds. The choice of the threshold size affects the quality of the detected edge. This step is to convert the image into a non-black, i.e. white, edge map, defining two parameters:and。
if the amplitude of the point is lower than Low, the point is not an edge point, is removed and is set to be black.
If the amplitude of the point is higher than High, it is an edge point, and the point is set to white.
If the amplitude of the point is between Low and High, then:
a) if the point is directly adjacent to a point with amplitude greater than High or the point can be indirectly adjacent to a point with amplitude greater than High through a point also between Low and High, the point is considered to be on the edge and is an edge point, and is set to white.
b) Otherwise, the edge point is not considered to be the edge point, and the color is set to be black.
Through the processing, the final image is converted into an edge map which is not black or white and is suitable for computer processing, and the purpose of edge detection is achieved, namely the edge map of the current-carrying ring is obtained.
And S4, filling the picture by using a FloodFill filling method. The FloodFill method fills a connected region with a certain color, and achieves different filling effects by setting the upper limit and the lower limit of a connectable pixel and a connection mode. FloodFill is often used to mark or separate portions of an image for further processing or analysis, and may also be used to obtain areas of masks from an input image, where the masks speed up the process, or to process only pixels specified by the masks, and the result of the operation is always some continuous area.
The FloodFill algorithm accepts three parameters: the start node, the target color and the alternate color. The algorithm traverses all nodes to find nodes connected to the starting node (connected by a path of the target color) and then changes their color to the alternate color.
As shown in fig. 6, in this embodiment, a four-way FloodFill algorithm is adopted to find a pixel pointIf not, filling the four adjacent pixel points, and continuously searching the four connected pixels until the closed area is completely filled with new color.
The algorithm comprises the following steps:
(1) and finding a pixel point which is not dyed, dyeing the pixel point into the designated color, and finishing the algorithm if the pixel point is not dyed.
(2) Initializing an empty queue and inserting the pixel points of the first step into the queue.
(3) And continuously obtaining the value of the head element of the queue and popping up, and point-dyeing the uncolored pixel adjacent to the head element of the queue into the specified color and adding the uncolored pixel into the queue.
(4) And repeating the first step until all the pixel points are dyed, and finishing the algorithm.
Fill with white (255 ) from four points (0, 0), (0, 255), (255, 0) and (255 ) of the picture in sequence using the flodfill algorithm. If the current-carrying ring is broken, the whole picture is changed into white after filling; if the current-carrying ring is not broken, namely the current-carrying ring is annular, after filling, the middle of the current-carrying ring cannot be filled with white, and the part is black with background color.
And S5, calculating the number of white pixel values in the picture after each filling, calculating the ratio of the number of white pixels to the total number of pixels of the picture (65536), and if the ratio is more than or equal to 0.98 when any one time occurs in four times of filling, determining that the carrier ring has a fracture abnormity.
Although the embodiments of the present invention have been described with reference to the accompanying drawings, it is not intended to limit the scope of the present invention, and it should be understood by those skilled in the art that various modifications and variations can be made without inventive efforts by those skilled in the art based on the technical solution of the present invention.
Claims (9)
1. A high-speed rail contact net current-carrying ring fracture abnormity detection method based on FloodFill is characterized by comprising the following steps:
firstly, preprocessing an input current-carrying ring picture, and zooming the picture so as to remove the influence of illumination and noise in the picture;
secondly, thresholding is carried out on the preprocessed picture, the processed picture only has two kinds of pixels, namely black pixels and white pixels, the black pixels represent the background, and the white pixels represent the foreground, namely a current-carrying ring;
extracting the contour of the current carrying ring by using a Canny algorithm;
filling white from the upper left, upper right, lower left and lower right corners of the picture by using a FloodFill filling method in sequence;
and fifthly, calculating the number of the white pixels after filling and the ratio of the white pixels to all pixels of the picture, and judging whether the current-carrying ring has fracture abnormity according to the ratio.
2. The method for detecting the abnormal breakage of the current-carrying ring of the high-speed rail contact network based on the FloodFill as claimed in claim 1, wherein in the first step, histogram equalization is adopted to remove the influence of illumination, the histogram of an original graph is converted into a uniform distribution form, and the dynamic range of pixel gray values is increased.
3. The FloodFill-based high-speed rail contact net current-carrying ring fracture abnormality detection method according to claim 2, wherein in the first step, image noise is removed by adopting Gaussian filtering.
4. The FloodFill-based high-speed rail contact net current-carrying ring fracture abnormality detection method according to claim 2, wherein in the first step, picture noise is removed by adopting bilateral filtering.
5. The FloodFill-based high-speed rail contact net current-carrying ring fracture abnormality detection method according to claim 1, wherein in the second step, the picture is subjected to fixed threshold segmentation, and an original image is set,Is a threshold value, the following formula is satisfied when the image is divided:
all the gray values of the original image pixels which are larger than or equal to T are set to be 255, and the current-carrying ring pixels are represented; all the others are set to 0, indicating background.
6. The method for detecting the abnormal breakage of the current-carrying ring of the high-speed rail contact network based on the FloodFill according to the claim 1, wherein the Canny algorithm is used for extracting the outline of the current-carrying ring in the third step, and the method comprises the following steps:
firstly, calculating the gradient amplitude direction of an image by using first-order partial derivative finite difference, searching a Roberts operator of an edge by using a local difference operator, wherein the sharpness of the edge is determined by the gradient of the gray level of the image;
then, carrying out non-maximum suppression on the gradient amplitude, and reserving a point with the maximum local gradient value;
finally, edges are detected and connected by using a hysteresis threshold, and the contour of the current-carrying ring is extracted.
7. The FloodFill-based high-speed rail contact net current-carrying ring fracture abnormality detection method according to claim 1, wherein a four-way connection FloodFill algorithm is adopted in the fourth step, four adjacent pixel points, namely upper, lower, left and right, of the pixel points are searched, if the four adjacent pixel points are not filled, the four adjacent pixel points are filled, and the four-way connection pixel points are continuously searched until a closed area is completely filled with a new color.
8. The FloodFill-based high-speed rail contact net current-carrying ring fracture abnormality detection method according to claim 1, wherein an eight-connected FloodFill algorithm is used in the fourth step, adjacent pixels of pixel points, namely upper, lower, left, right, upper left, lower left, upper right and lower right, are calculated, and eight-connected pixel filling of the pixels is searched recursively until the area is completely filled with new colors.
9. The method for detecting the abnormal breakage of the current-carrying ring of the high-speed rail contact net based on the FloodFill according to the claim 7 or 8, wherein the fifth step is to calculate the number of white pixel values in the picture after each filling, calculate the ratio of the number of the white pixels to the total number of the pixels in the picture, and determine that the abnormal breakage of the current-carrying ring occurs when the ratio is larger than or equal to a set value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110827688.0A CN113284076A (en) | 2021-07-22 | 2021-07-22 | FloodFill-based high-speed rail contact net current-carrying ring fracture abnormity detection method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110827688.0A CN113284076A (en) | 2021-07-22 | 2021-07-22 | FloodFill-based high-speed rail contact net current-carrying ring fracture abnormity detection method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113284076A true CN113284076A (en) | 2021-08-20 |
Family
ID=77286925
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110827688.0A Pending CN113284076A (en) | 2021-07-22 | 2021-07-22 | FloodFill-based high-speed rail contact net current-carrying ring fracture abnormity detection method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113284076A (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20180027145A1 (en) * | 2016-07-25 | 2018-01-25 | Iteris, Inc. | Image-based field boundary detection and identification |
CN109840909A (en) * | 2019-01-18 | 2019-06-04 | 西安科技大学 | A kind of crucible bubble counting device and method of counting |
CN111402214A (en) * | 2020-03-07 | 2020-07-10 | 西南交通大学 | Neural network-based automatic detection method for breakage defect of catenary dropper current-carrying ring |
CN111899268A (en) * | 2020-08-17 | 2020-11-06 | 上海商汤智能科技有限公司 | Image segmentation method and device, electronic equipment and storage medium |
US20200363408A1 (en) * | 2018-01-11 | 2020-11-19 | Essenlix Corporation | Homogenous assay (ii) |
-
2021
- 2021-07-22 CN CN202110827688.0A patent/CN113284076A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20180027145A1 (en) * | 2016-07-25 | 2018-01-25 | Iteris, Inc. | Image-based field boundary detection and identification |
US20200363408A1 (en) * | 2018-01-11 | 2020-11-19 | Essenlix Corporation | Homogenous assay (ii) |
CN109840909A (en) * | 2019-01-18 | 2019-06-04 | 西安科技大学 | A kind of crucible bubble counting device and method of counting |
CN111402214A (en) * | 2020-03-07 | 2020-07-10 | 西南交通大学 | Neural network-based automatic detection method for breakage defect of catenary dropper current-carrying ring |
CN111899268A (en) * | 2020-08-17 | 2020-11-06 | 上海商汤智能科技有限公司 | Image segmentation method and device, electronic equipment and storage medium |
Non-Patent Citations (1)
Title |
---|
杨帆: "《数字图像处理与分析》", 31 January 2019, 北京航空航天大学出版社 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109741356B (en) | Sub-pixel edge detection method and system | |
CN102782706B (en) | Text enhancement of a textual image undergoing optical character recognition | |
CN108022233A (en) | A kind of edge of work extracting method based on modified Canny operators | |
CN110264459A (en) | A kind of interstices of soil characteristics information extraction method | |
CN109377450B (en) | Edge protection denoising method | |
US20100008576A1 (en) | System and method for segmentation of an image into tuned multi-scaled regions | |
CN105913396A (en) | Noise estimation-based image edge preservation mixed de-noising method | |
CN107220988A (en) | Based on the parts image edge extraction method for improving canny operators | |
KR20150097367A (en) | Method and apparatus for adjusting image | |
CN109961416B (en) | Business license information extraction method based on morphological gradient multi-scale fusion | |
CN105894491A (en) | Image high-frequency information positioning method and device | |
CN103942756B (en) | A kind of method of depth map post processing and filtering | |
CN107798670A (en) | A kind of dark primary prior image defogging method using image wave filter | |
CN109472788A (en) | A kind of scar detection method on airplane riveting surface | |
CN115330645A (en) | Welding image enhancement method | |
CN115131351A (en) | Engine oil radiator detection method based on infrared image | |
CN113537037A (en) | Pavement disease identification method, system, electronic device and storage medium | |
CN110807406B (en) | Foggy day detection method and device | |
CN105787912A (en) | Classification-based step type edge sub pixel localization method | |
CN115660990A (en) | Endoscope image mirror reflection detection and restoration method based on brightness classification | |
CN109101985A (en) | It is a kind of based on adaptive neighborhood test image mismatch point to elimination method | |
CN117094975A (en) | Method and device for detecting surface defects of steel and electronic equipment | |
CN107292897A (en) | Image edge extraction method, device and terminal for YUV domains | |
CN114943744A (en) | Edge detection method based on local Otsu thresholding | |
JP2004030188A (en) | Method, apparatus and program for dividing image into areas |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210820 |
|
RJ01 | Rejection of invention patent application after publication |