CN112232243B - Locking plate fault identification method based on 3D image information - Google Patents
Locking plate fault identification method based on 3D image information Download PDFInfo
- Publication number
- CN112232243B CN112232243B CN202011132380.6A CN202011132380A CN112232243B CN 112232243 B CN112232243 B CN 112232243B CN 202011132380 A CN202011132380 A CN 202011132380A CN 112232243 B CN112232243 B CN 112232243B
- Authority
- CN
- China
- Prior art keywords
- locking plate
- image
- side edge
- edge
- line segment
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/20—Scenes; Scene-specific elements in augmented reality scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/24—Aligning, centring, orientation detection or correction of the image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/28—Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
- G06T2207/10012—Stereo images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
Abstract
A locking plate fault identification method based on 3D image information relates to the field of image processing. The invention aims to solve the problem that the displacement fault of a locking plate on a truck is not accurately identified at present. Sequentially carrying out binarization processing on the acquired 3D height information image and correcting all pixels in the image to the same plane by adopting an image correction method to obtain a corrected 3D binarization image; then distinguishing a locking plate image from the corrected 3D binarization image; then, capturing a locking plate component image from the locking plate image; and finally, extracting line segments meeting set conditions from the locking plate component image so as to judge whether the locking plate is shifted. The device is used for identifying whether the locking plate on the train is displaced or not.
Description
Technical Field
The invention relates to a locking plate displacement fault image identification method. Belonging to the field of image processing.
Background
The prior TFDS-3 equipment uses a linear array camera to acquire projection information of a vehicle in the direction vertical to the optical axis of the camera, so that stains, water stains and the like which appear in the running process of the vehicle are easy to cause false alarm. A freight car operation fault dynamic image detection system (TFDS-3D) is characterized in that 3D height information acquisition equipment is added on the basis of the existing TFDS-3 equipment, and height information images of a car are introduced, so that the alarm accuracy rate is improved, and the safety precaution level of freight car operation is enhanced.
The TFDS-3D system is different from a traditional detection method which mainly uses dynamic detection personnel to check each train image and provide suspected abnormality, and has the core idea that the images passing through the train are detected through three-dimensional images and three-dimensional coordinate information, the collected truck images can be automatically analyzed and fault recognized, abnormal positions in the images are subjected to alarm prompt, interference of non-fault factors such as dust, water stain, scratch, chalk mark, stain, paint removal and the like can be eliminated, the number of false alarms is obviously reduced, and the fault recognition accuracy rate is improved. The multi-angle detection and identification can be carried out when the key part is covered by ice and snow or other foreign matters. The safety monitoring measures of the railway wagon are more perfect, the development requirements of the railway can be better met, and the method has wide popularization and application prospects. Although the TFDS-3D system can recognize the 3D image, it cannot accurately recognize the abnormality for the oblique image, and thus, a method of accurately recognizing the image of the shift failure of the locking plate on the truck is now lacking.
Disclosure of Invention
The invention aims to solve the problem that the displacement fault of a locking plate on a truck is not accurately identified at present. A locking plate fault identification method based on 3D image information is now provided.
A locking plate fault identification method based on 3D image information comprises the following steps:
step 1, acquiring a 3D height information image of a locking plate area on a train;
step 2, carrying out binarization processing on the 3D height information image to obtain a gray level image, and correcting all pixels in the gray level image to the same plane by adopting an image correction method to obtain a corrected 3D binarization image;
3, identifying a locking plate image from the corrected 3D binarization image;
step 4, intercepting a locking plate component image from the locking plate image;
step 5, extracting edge line segments of the locking plate component image, wherein the edge line segments comprise at least one locking plate upper side edge inclined line segment, at least one locking plate vertical edge vertical line segment and at least one locking plate lower side edge horizontal line segment;
step 6, judging whether a locking plate upper side edge inclined line segment, a locking plate vertical edge vertical line segment and a locking plate lower edge horizontal line segment which meet preset conditions can be selected from at least one locking plate upper side edge inclined line segment, at least one locking plate vertical edge vertical line segment and at least one locking plate lower edge horizontal line segment respectively, if not, indicating that the locking plate is shifted, and if so, executing the step 7;
and 7, judging whether the distance between the adjacent end points of two adjacent line segments in the selected three line segments is greater than a set pixel value, if so, indicating that the locking plate shifts, otherwise, judging whether the included angle of two adjacent line segments in the selected three edge lines is within a preset included angle range, if so, indicating that the locking plate does not shift, and if not, indicating that the locking plate shifts.
Preferably, in step 1, the step of obtaining a 3D height information image of a truck upper lock plate area comprises the following specific steps:
a dynamic image detection system for truck operation faults is adopted to collect a truck passing image, and a 3D height information image of a locking plate area is intercepted from the truck passing image.
Preferably, in step 2, all pixels in the grayscale image are corrected to the same plane by using an image correction method, so as to obtain a corrected 3D image binary image, which specifically comprises the following steps:
let each pixel position in a gray image of M rows and N columns be expressed as (x, y), and then the gray value of the corresponding pixel point is expressed as f (x, y), where x represents the height range of the gray image, y represents the width range of the gray image,
let the offset value of each pixel point be Δ offset, expressed as:
Δ offset — y × tan (θ) formula 1,
wherein, theta is an arc value, (alpha x pi)/180, alpha is a locking plate inclination angle value,
equation 1 is modified as:
Δ offset ═ y × tan ((α × pi)/180) formula 2,
assuming that the corrected 3D binarized image is represented as g (x, y), f (x, y) is corrected by the pixel offset value to obtain:
preferably, in step 3, the locking plate image is distinguished from the corrected 3D binarized image, and the specific process is as follows:
step 31, classifying the regions with the same pixel value in the corrected 3D binary image into a sub-image;
step 32, calculating the proportion of each sub-image in the corrected 3D binary image, and intercepting the sub-images with the proportion of a preset proportion;
step 33, setting a preset binarization threshold range according to the pixel value of the intercepted sub-image;
and step 34, intercepting pixel points of the corrected 3D binarization image within the binarization threshold range to form a locking plate image.
Preferably, in step 33, the lowest threshold value and the highest threshold value of the binarization threshold value range are respectively expressed as:
low Threshold-n Threshold-a equation 4,
high Threshold + a equation 5,
in the formula, Low _ Threshold represents the lowest Threshold, High _ Threshold represents the highest Threshold, nThreshold is the pixel value of the locking plate image, and a is 2000.
Preferably, in step 4, the locking plate component image is captured from the locking plate image, and the specific process is as follows:
step 41, obtaining the sum of the number of each row of white pixel points in the locking plate image, obtaining the height value of each row according to the sum of the number of each row of white pixel points, taking the height value larger than the set height value as the length of the left edge of the locking plate, and obtaining the position of the left edge of the locking plate in the locking plate image;
step 42, obtaining the sum of the number of white pixel points of each line in the locking plate image, obtaining the width value of each line according to the sum of the number of white pixel points of each line, taking the width value larger than the set width value as the length of the lower side edge of the locking plate, and obtaining the position of the lower side edge of the locking plate in the locking plate image;
and 43, obtaining a locking plate component image according to an area surrounded by the position of the left side edge of the locking plate in the locking plate image, the position of the lower side edge of the locking plate in the locking plate image, the length of the left side edge of the locking plate and the length of the lower side edge of the locking plate.
Preferably, in step 41, the height value is set to 0.5M, where M is the height of the grayscale image;
in step 42, the width value is set to 0.5N, where N is the width of the grayscale image.
Preferably, the specific process of step 5 is:
step 51, obtaining a locking plate part image from a locking plate part image;
step 52, performing edge detection on the partial image of the locking plate by adopting an edge detection algorithm to obtain an edge image of the locking plate;
and 53, performing line detection on the edge image by adopting a Hough line detection method to obtain at least one inclined line segment on the upper side edge of the locking plate, a vertical line segment on the vertical edge of the locking plate and a horizontal line segment on the lower edge of the locking plate.
Preferably, in step 51, the locking plate part image is obtained from the locking plate part image, and the specific process is as follows:
step 511, intercepting the area 1/2 on the right side of the locking plate component image as a locking plate displacement judgment image;
and 512, removing white pixel point noise in the locking plate displacement judgment image by adopting a seed filling algorithm to obtain a locking plate partial image.
Preferably, the specific process of step 6 is:
step 61, obtaining the angle formed by each line segment and the horizontal line obtained in the step 53, the length of the vertical edge and the horizontal line segment of the vertical edge of the locking plate, the length of the horizontal line segment of the lower edge of the locking plate, the central position coordinates of the vertical edge and the horizontal line segment of the lower edge of the locking plate;
step 62, judging whether a hough line detection method can be adopted to select one line segment from the inclined line segments on the upper side edges of the plurality of locking plates, the vertical line segments on the vertical edges of the locking plates and the horizontal line segments on the lower edges of the locking plates respectively, so that the three selected line segments simultaneously meet the preset conditions: the method comprises the following steps that the length of one of a plurality of locking plate upper side edge inclined line segments, a locking plate vertical side edge vertical line segment and a locking plate lower side edge horizontal line segment is the longest, the angle formed by the three line segments and a horizontal line respectively meets the respective preset angle range, the center line coordinate of the locking plate upper side edge inclined line segment in the three line segments is smaller than the center line coordinate of the locking plate lower side edge horizontal line segment in the locking plate vertical line segment and the angle formed by the three line segments and the horizontal line meet the respective preset angle range, if not, the locking plate is shifted, and if yes, the step 7 is executed.
The invention has the beneficial effects that:
the method is based on the 3D height information image acquired by the TFDS-3D image acquisition system, false alarm caused by various interference factors is reduced, binarization processing is sequentially carried out on the acquired 3D height information image, all pixels in the image are corrected to the same plane by adopting an image correction method, and a corrected 3D binarization image is obtained; then distinguishing a locking plate image from the corrected 3D binarization image by adopting a pixel value statistical method; then, intercepting a locking plate component image from the locking plate image by adopting a pixel point accumulation sum method; and finally, extracting the edge line segment of the locking plate component image so as to judge whether the locking plate is shifted.
Drawings
Fig. 1 is a flowchart of a locking plate failure identification method based on 3D image information according to embodiment 1;
FIG. 2 is a schematic view of a latch plate area on a truck;
FIG. 3 is a 3D height information image of a locking plate area;
FIG. 4 is a corrected 3D binarized image;
FIG. 5 is a schematic diagram of an image correction method;
FIG. 6 is the result of distinguishing the locking plate image from the corrected 3D binarized image;
FIG. 7 is a view of a locking plate component image taken from the locking plate image;
fig. 8(a) is a locking plate displacement judgment image;
FIG. 8(b) is an edge image of the locking plate;
FIG. 8(c) is a line segment selected from each of an upper side edge oblique line, a vertical edge vertical line and a lower side edge horizontal line of the plurality of locking plates;
fig. 9 is an overall flowchart of a locking plate failure identification method based on 3D image information.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of real-time embodiments of the invention, and not all embodiments. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
It should be noted that, in the present application, the embodiments may be combined with each other without conflict.
The invention is further described with reference to the following figures and specific examples.
Example 1:
as shown in fig. 1 to 4, the method for identifying the fault of the locking plate based on the 3D image information includes:
step 1, acquiring a 3D height information image of a locking plate area on a train;
step 2, carrying out binarization processing on the 3D height information image to obtain a gray level image, and correcting all pixels in the gray level image to the same plane by adopting an image correction method to obtain a corrected 3D binarization image;
3, identifying a locking plate image from the corrected 3D binarization image;
step 4, intercepting a locking plate component image from the locking plate image;
step 5, extracting edge line segments of the locking plate component image, wherein the edge line segments comprise at least one locking plate upper side edge inclined line segment, at least one locking plate vertical edge vertical line segment and at least one locking plate lower side edge horizontal line segment;
step 6, judging whether a locking plate upper side edge inclined line segment, a locking plate vertical edge vertical line segment and a locking plate lower edge horizontal line segment which meet preset conditions can be selected from at least one locking plate upper side edge inclined line segment, at least one locking plate vertical edge vertical line segment and at least one locking plate lower edge horizontal line segment respectively, if not, indicating that the locking plate is shifted, and if so, executing the step 7;
and 7, judging whether the distance between the adjacent end points of two adjacent line segments in the selected three line segments is greater than a set pixel value, if so, indicating that the locking plate shifts, otherwise, judging whether the included angle of two adjacent line segments in the selected three edge lines is within a preset included angle range, if so, indicating that the locking plate does not shift, and if not, indicating that the locking plate shifts.
Specifically, in step 7, the pixel value is set to 10 pixels.
In a preferred embodiment of the present invention, as shown in fig. 2, in step 1, a 3D height information image of a locking plate area on a truck is obtained, and the specific process is as follows:
a dynamic image detection system for truck operation faults is adopted to collect a truck passing image, and a 3D height information image of a locking plate area is intercepted from the truck passing image.
Specifically, fig. 2 shows a captured image of the passing vehicle, from which a 3D height information image of the locking plate area can be seen.
In a preferred embodiment of the present invention, as shown in fig. 4, in step 2, the grayscale image is corrected to the same plane by using an image correction method, so as to obtain a corrected 3D image binary image, and the specific process is as follows:
let each pixel position in a gray image of M rows and N columns be expressed as (x, y), and then the gray value of the corresponding pixel point is expressed as f (x, y), where x represents the height range of the gray image, y represents the width range of the gray image,
let the offset value of each pixel point be Δ offset, expressed as:
Δ offset — y × tan (θ) formula 1,
wherein, theta is an arc value, (alpha x pi)/180, alpha is a locking plate inclination angle value,
equation 1 is modified as:
Δ offset ═ y × tan ((α × pi)/180) formula 2,
assuming that the corrected 3D binarized image is represented as g (x, y), f (x, y) is corrected by the pixel offset value to obtain:
specifically, a certain included angle exists between the locking plate device and the rail direction, as shown in fig. 2, the locking plate shot by the camera perpendicular to the rail direction also presents an inclined state, so that the 3D image collected by the TFDS-3D image collecting system also presents an inclined state, the height information of the 3D image is the distance between the reaction object and the camera, the closer the locking plate is to the camera, the smaller the pixel value is, otherwise, the farther the locking plate is from the camera, the larger the pixel value is, and due to the inclined state of the locking plate, the shot 3D height information image is as shown in fig. 3. Based on the problem, an image correction method is provided, the 3D height image of the inclined locking plate is corrected to the same plane, the locking plate area segmentation and fault judgment are facilitated, and the result of the corrected 3D image after binarization is shown in FIG. 4.
Since the 3D image height information is extracted, and the pixel value is larger as the object is farther from the camera, the pixel value of the long-distance point B is larger than the pixel value of the short-distance point a in fig. 5, and therefore, the offset Δ offset needs to be subtracted when the point B is corrected in the direction of the point a in the present invention.
Through formulas 1 to 3, the pixel points at the corresponding positions are subjected to weighted calculation of the offset, and finally the corrected image g (x, y) is obtained. The g (x, y) corrected image is shown in fig. 4.
In a preferred embodiment of the present invention, as shown in fig. 6, in step 3, the locking plate image is distinguished from the corrected 3D binarized image, and the specific process is as follows:
step 31, classifying the regions with the same pixel value in the corrected 3D binary image into a sub-image;
step 32, calculating the proportion of each sub-image in the corrected 3D binary image, and intercepting the sub-images with the proportion of a preset proportion;
step 33, setting a preset binarization threshold range according to the pixel value of the intercepted sub-image;
and step 34, intercepting pixel points of the corrected 3D binarization image within the binarization threshold range to form a locking plate image.
Specifically, after the 3D height information image is corrected, the locking plate area is completely displayed in the image, at this time, a binarization image is usually performed to extract the locking plate area, but due to differences of rail wagons, the captured locking plate area image may have differences, and a binarization threshold value may also have differences correspondingly, and it is not ensured that the locking plate information is completely retained by using a fixed threshold value to binarize the image, so a mode of automatically calculating the threshold value is adopted herein to determine the binarization threshold value. The corrected 3D binarized image pixel value range is (0, 65535).
The locking plate image is a white area in fig. 6, and accounts for more than 80% of the image, so that the pixel value of 80% is calculated, namely, the pixel value of the locking plate area is regarded as the pixel value, and the value is expressed by nThreshold.
In a preferred embodiment of the present invention, in step 31, the predetermined ratio is 80%.
In a preferred embodiment of the present invention, in step 33, the lowest threshold and the highest threshold of the binarization threshold range are respectively expressed as:
low Threshold-n Threshold-a equation 4,
high Threshold + a equation 5,
in the formula, Low _ Threshold represents the lowest Threshold, High _ Threshold represents the highest Threshold, nThreshold is the pixel value of the locking plate image, and a is 2000.
In a preferred embodiment of the present invention, as shown in fig. 7, in step 4, the locking plate component image is captured from the locking plate image, and the specific process is as follows:
step 41, obtaining the sum of the number of each row of white pixel points in the locking plate image, obtaining the height value of each row according to the sum of the number of each row of white pixel points, taking the height value larger than the set height value as the length of the left edge of the locking plate, and obtaining the position of the left edge of the locking plate in the locking plate image;
step 42, obtaining the sum of the number of white pixel points of each line in the locking plate image, obtaining the width value of each line according to the sum of the number of white pixel points of each line, taking the width value larger than the set width value as the length of the lower side edge of the locking plate, and obtaining the position of the lower side edge of the locking plate in the locking plate image;
and 43, obtaining a locking plate component image according to an area surrounded by the position of the left side edge of the locking plate in the locking plate image, the position of the lower side edge of the locking plate in the locking plate image, the length of the left side edge of the locking plate and the length of the lower side edge of the locking plate.
Specifically, in the preferred embodiment, the binary image is continuously and finely positioned, and the coordinates of the row and column of the locking plate, namely the left edge and the lower edge, are accurately positioned. The fine positioning steps are as follows:
1) traversing the binary image, traversing from the left side to the right side of the image, counting the sum of each row of white pixels, and recording the row number as col when the sum is greater than 0.5M (M is the image height value), wherein the row col is the left edge of the locking plate, as shown in col in FIG. 6.
2) Traversing the binary image, traversing from the lower side to the upper side of the image, counting the sum of white pixel points in each line, and recording the line number row when the sum is greater than 0.5N (N is the image width value), wherein the line row is the lower edge of the locking plate, as shown in row in fig. 6.
3) The accurate locking plate image can be captured by combining the positioning results col column and row of the last two steps, as shown by the rectangular area in fig. 6.
In a preferred embodiment of the present invention, in step 41, the height value is set to 0.5M, where M is the height of the gray image;
in step 42, the width value is set to 0.5N, where N is the width of the grayscale image.
In a preferred embodiment of the present invention, the specific process of step 5 is:
step 51, obtaining a locking plate part image from a locking plate part image;
step 52, performing edge detection on the partial image of the locking plate by adopting an edge detection algorithm to obtain an edge image of the locking plate;
and 53, performing line detection on the edge image by adopting a Hough line detection method to obtain at least one inclined line segment on the upper side edge of the locking plate, a vertical line segment on the vertical edge of the locking plate and a horizontal line segment on the lower edge of the locking plate.
In a preferred embodiment of the present invention, in step 51, a locking plate part image is obtained from a locking plate part image, and the specific process is as follows:
step 511, intercepting the area 1/2 on the right side of the locking plate component image as a locking plate displacement judgment image;
and 512, removing white pixel point noise in the locking plate displacement judgment image by adopting a seed filling algorithm to obtain a locking plate partial image.
In a preferred embodiment of the present invention, as shown in fig. 8, the specific process of step 6 is:
step 61, obtaining the angle formed by each line segment and the horizontal line obtained in the step 53, the length of the vertical edge and the horizontal line segment of the vertical edge of the locking plate, the length of the horizontal line segment of the lower edge of the locking plate, the central position coordinates of the vertical edge and the horizontal line segment of the lower edge of the locking plate;
step 62, judging whether a hough line detection method can be adopted to select one line segment from the inclined line segments on the upper side edges of the plurality of locking plates, the vertical line segments on the vertical edges of the locking plates and the horizontal line segments on the lower edges of the locking plates respectively, so that the three selected line segments simultaneously meet the preset conditions: the method comprises the following steps that the length of one of a plurality of locking plate upper side edge inclined line segments, a locking plate vertical side edge vertical line segment and a locking plate lower side edge horizontal line segment is the longest, the angle formed by the three line segments and a horizontal line respectively meets the respective preset angle range, the center line coordinate of the locking plate upper side edge inclined line segment in the three line segments is smaller than the center line coordinate of the locking plate lower side edge horizontal line segment in the locking plate vertical line segment and the angle formed by the three line segments and the horizontal line meet the respective preset angle range, if not, the locking plate is shifted, and if yes, the step 7 is executed.
Specifically, after fine positioning, an accurate locking plate binary sub-image is obtained, as shown in the rectangular area in fig. 7, the area 1/2 on the right side of the rectangular area is cut out to be used as a locking plate displacement judgment image, and the cut-out sub-image is shown in fig. 8 (a).
Step 52 specifically includes: searching the locking plate displacement by adopting a seed filling algorithm to judge the maximum image communication area, eliminating noise interference and only leaving the locking plate area; the maximum connected region is like the white region of the image in fig. 8(a), white pixel noise can exist in the image judged by shifting the locking plate under the general condition, but the area is not larger than the white region of the locking plate, the maximum connected region, namely the white region with the maximum area, is calculated to obtain the region of the locking plate, and other white regions with smaller areas (namely the white noise) are eliminated.
Specifically, the detected alarm information is directly uploaded to a vehicle detection platform, whether the locking plate shifting condition exists or not is further confirmed by a person, and the whole flow chart of the locking plate shifting fault detection method based on the 3D image is shown in fig. 9.
In a preferred embodiment of the present invention, in step 7, the predetermined included angle range is more than 20 degrees above the normal included angle range,
the normal included angle range specifically refers to: the normal included angle between the oblique line segment at the upper side edge of the locking plate and the vertical line segment at the vertical edge of the locking plate ranges from 130 degrees to 160 degrees,
the normal included angle between the vertical line segment of the vertical edge of the locking plate and the horizontal line segment of the lower edge of the locking plate ranges from 80 degrees to 100 degrees.
Specifically, in step 62, the three line segments satisfy respective preset angle ranges, specifically:
the angle range of the inclined line segment at the upper side edge of the preset locking plate is 105 degrees to 166 degrees,
presetting the angle range of the vertical line segment at the vertical edge of the locking plate to be 85-103 degrees,
the angle range of the lower edge of the preset locking plate along the horizontal line segment is 155 degrees to 180 degrees.
Claims (8)
1. The locking plate fault identification method based on the 3D image information is characterized by comprising the following steps:
step 1, acquiring a 3D height information image of a locking plate area on a train;
step 2, carrying out binarization processing on the 3D height information image to obtain a gray level image, and correcting all pixels in the gray level image to the same plane by adopting an image correction method to obtain a corrected 3D binarization image;
3, identifying a locking plate image from the corrected 3D binarization image;
step 4, intercepting a locking plate component image from the locking plate image;
step 5, extracting edge line segments of the locking plate component image, wherein the edge line segments comprise at least one locking plate upper side edge inclined line segment, at least one locking plate vertical edge vertical line segment and at least one locking plate lower side edge horizontal line segment;
step 6, judging whether a locking plate upper side edge inclined line segment, a locking plate vertical edge vertical line segment and a locking plate lower edge horizontal line segment which meet preset conditions can be selected from at least one locking plate upper side edge inclined line segment, at least one locking plate vertical edge vertical line segment and at least one locking plate lower edge horizontal line segment respectively, if not, indicating that the locking plate is shifted, and if so, indicating that the locking plate is not shifted;
in step 2, correcting all pixels in the gray level image to the same plane by adopting an image correction method to obtain a corrected 3D image binarization image, and the specific process is as follows:
let each pixel position in a gray image of M rows and N columns be expressed as (x, y), and then the gray value of the corresponding pixel point is expressed as f (x, y), where x represents the height range of the gray image, y represents the width range of the gray image,
let the offset value of each pixel point be Δ offset, expressed as:
Δ offset — y × tan (θ) formula 1,
wherein, theta is an arc value, (alpha x pi)/180, alpha is a locking plate inclination angle value,
equation 1 is modified as:
Δ offset ═ y × tan ((α × pi)/180) formula 2,
assuming that the corrected 3D binarized image is represented as g (x, y), f (x, y) is corrected by the pixel offset value to obtain:
the specific process of the step 6 is as follows:
step 61, obtaining the angle formed by each line segment and a horizontal line, the upper side edge oblique line segments of the plurality of locking plates, the length of the vertical edge vertical line segments and the lower side edge horizontal line segments of the locking plates, the upper side edge oblique line segments of the plurality of locking plates, and the central position coordinates of the vertical edge vertical line segments and the lower side edge horizontal line segments of the locking plates obtained in the step 5;
step 62, judging whether a hough line detection method can be adopted to select one line segment from the inclined line segments on the upper side edges of the plurality of locking plates, the vertical line segments on the vertical edges of the locking plates and the horizontal line segments on the lower edges of the locking plates respectively, so that the three selected line segments simultaneously meet the preset conditions: the locking plate is characterized by comprising a plurality of locking plate upper side edge inclined line segments, a locking plate vertical side edge vertical line segment and a locking plate lower side edge horizontal line segment which are the longest in length, wherein the locking plate upper side edge inclined line segments, the locking plate vertical side edge vertical line segments and the locking plate lower side edge horizontal line segments, and the angles formed by the three line segments and a horizontal line respectively meet the respective preset angle ranges, the center line coordinate of the locking plate upper side edge inclined line segment in the three line segments is smaller than the center line coordinate of the locking plate lower side edge horizontal line segment in the locking plate horizontal line segment, if not, the locking plate is indicated to be displaced, and if yes, the locking plate is indicated to be not displaced.
2. The method for identifying the fault of the locking plate based on the 3D image information as claimed in claim 1, wherein in the step 1, the 3D height information image of the locking plate area on the truck is obtained, and the specific process is as follows:
a dynamic image detection system for truck operation faults is adopted to collect a truck passing image, and a 3D height information image of a locking plate area is intercepted from the truck passing image.
3. The locking plate fault identification method based on 3D image information as claimed in claim 1, wherein in step 3, the locking plate image is identified from the corrected 3D binarized image by the specific process:
step 31, classifying the regions with the same pixel value in the corrected 3D binary image into a sub-image;
step 32, calculating the proportion of each sub-image in the corrected 3D binary image, and intercepting the sub-images with the proportion of a preset proportion;
step 33, setting a preset binarization threshold range according to the pixel value of the intercepted sub-image;
and step 34, intercepting pixel points of the corrected 3D binarization image within a preset binarization threshold range to form a locking plate image.
4. The locking plate fault identification method based on 3D image information as claimed in claim 3, wherein in step 33, the lowest threshold and the highest threshold of the preset binarization threshold range are respectively expressed as:
low Threshold-n Threshold-a equation 4,
high Threshold + a equation 5,
in the formula, Low _ Threshold represents the lowest Threshold, High _ Threshold represents the highest Threshold, nThreshold is the pixel value of the locking plate image, and a is 2000.
5. The locking plate fault identification method based on 3D image information as claimed in claim 1, wherein in step 4, the locking plate component image is intercepted from the locking plate image, and the specific process is as follows:
step 41, obtaining the sum of the number of each row of white pixel points in the locking plate image, obtaining the height value of each row according to the sum of the number of each row of white pixel points, taking the height value larger than the set height value as the length of the left edge of the locking plate, and obtaining the position of the left edge of the locking plate in the locking plate image;
step 42, obtaining the sum of the number of white pixel points of each line in the locking plate image, obtaining the width value of each line according to the sum of the number of white pixel points of each line, taking the width value larger than the set width value as the length of the lower side edge of the locking plate, and obtaining the position of the lower side edge of the locking plate in the locking plate image;
and 43, obtaining a locking plate component image according to an area surrounded by the position of the left side edge of the locking plate in the locking plate image, the position of the lower side edge of the locking plate in the locking plate image, the length of the left side edge of the locking plate and the length of the lower side edge of the locking plate.
6. The locking plate fault identification method based on 3D image information as claimed in claim 5, wherein in step 41, a height value is set to be 0.5M, wherein M is the height of a gray image;
in step 42, the width value is set to 0.5N, where N is the width of the grayscale image.
7. The locking plate fault identification method based on 3D image information according to claim 1, characterized in that the specific process of step 5 is as follows:
step 51, obtaining a locking plate part image from a locking plate part image;
step 52, performing edge detection on the partial image of the locking plate by adopting an edge detection algorithm to obtain an edge image of the locking plate;
and 53, performing line detection on the edge image by adopting a Hough line detection method to obtain at least one inclined line segment on the upper side edge of the locking plate, a vertical line segment on the vertical edge of the locking plate and a horizontal line segment on the lower edge of the locking plate.
8. The method for identifying the locking plate fault based on the 3D image information as claimed in claim 7, wherein in the step 51, the locking plate part image is obtained from the locking plate part image by the following specific processes:
step 511, intercepting the area 1/2 on the right side of the locking plate component image as a locking plate displacement judgment image;
and 512, removing white pixel point noise in the locking plate displacement judgment image by adopting a seed filling algorithm to obtain a locking plate partial image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011132380.6A CN112232243B (en) | 2020-10-21 | 2020-10-21 | Locking plate fault identification method based on 3D image information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011132380.6A CN112232243B (en) | 2020-10-21 | 2020-10-21 | Locking plate fault identification method based on 3D image information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112232243A CN112232243A (en) | 2021-01-15 |
| CN112232243B true CN112232243B (en) | 2021-05-07 |
Family
ID=74118417
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011132380.6A Active CN112232243B (en) | 2020-10-21 | 2020-10-21 | Locking plate fault identification method based on 3D image information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112232243B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103500328A (en) * | 2013-10-16 | 2014-01-08 | 北京航空航天大学 | Method for automatically detecting deflection fault of railway wagon locking plate |
| CN104259258A (en) * | 2014-09-28 | 2015-01-07 | 王松 | Simple automobile body metal plate stretching correction device |
| EP3388601A1 (en) * | 2017-04-12 | 2018-10-17 | ALSTOM Transport Technologies | A system for locking a door for access to a space, in particular a toilet of a railway vehicle |
| CN108898574A (en) * | 2018-05-09 | 2018-11-27 | 江苏大学 | Train bogie cross-braced device head bolts looseness fault automatic testing method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7708509B1 (en) * | 2009-05-08 | 2010-05-04 | Bennett Bruce A | Locking retainer for threaded fasteners |
-
2020
- 2020-10-21 CN CN202011132380.6A patent/CN112232243B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103500328A (en) * | 2013-10-16 | 2014-01-08 | 北京航空航天大学 | Method for automatically detecting deflection fault of railway wagon locking plate |
| CN104259258A (en) * | 2014-09-28 | 2015-01-07 | 王松 | Simple automobile body metal plate stretching correction device |
| EP3388601A1 (en) * | 2017-04-12 | 2018-10-17 | ALSTOM Transport Technologies | A system for locking a door for access to a space, in particular a toilet of a railway vehicle |
| CN108898574A (en) * | 2018-05-09 | 2018-11-27 | 江苏大学 | Train bogie cross-braced device head bolts looseness fault automatic testing method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112232243A (en) | 2021-01-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103077526B (en) | There is train method for detecting abnormality and the system of depth detection function | |
| CN108647572B (en) | Lane departure early warning method based on Hough transform | |
| CN102323070B (en) | Method and system for detecting abnormality of train | |
| CN111289261B (en) | Detection method for in-warehouse bottom piece | |
| CN109489724B (en) | Tunnel train safe operation environment comprehensive detection device and detection method | |
| CN104966049B (en) | Lorry detection method based on image | |
| CN105203552A (en) | 360-degree tread image detecting system and method | |
| CN106548147A (en) | A kind of quick noise robustness image foreign matter detection method and TEDS systems | |
| CN113516629A (en) | Intelligent detection system for TFDS passing operation | |
| CN111832571B (en) | Automatic detection method for truck brake beam strut fault | |
| CN111080650A (en) | Method for detecting looseness and loss faults of small part bearing blocking key nut of railway wagon | |
| CN110610516A (en) | Railway fastener nut center positioning method | |
| CN110415221B (en) | Automatic detection method for preventing container truck from being lifted based on image feature point matching | |
| CN115527170A (en) | Method and system for identifying closing fault of door stopper handle of automatic freight car derailing brake device | |
| CN112001908B (en) | Railway freight car sleeper beam hole carried foreign matter detection method | |
| CN112232243B (en) | Locking plate fault identification method based on 3D image information | |
| CN113469974A (en) | Method and system for monitoring grate plate state of pellet chain grate | |
| CN112037207A (en) | Method for detecting closing fault of automatic brake valve plug handle during railway wagon derailment | |
| JP4224328B2 (en) | Car number recognition device | |
| CN114581447B (en) | Conveying belt deviation identification method and device based on machine vision | |
| CN109978879B (en) | Box corner in-groove state detection method based on railway wagon loading video monitoring | |
| CN112132824B (en) | Automatic detection method for failure of freight car axle box spring | |
| JPH02242382A (en) | Defect checking method | |
| CN112949644B (en) | Fault image identification method for lower pull rod of truck | |
| Yang et al. | Research of computer vision fault recognition algorithm of center plate bolts of train |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |