CN116385434A - Intelligent detection method for precast beam cracks - Google Patents
Intelligent detection method for precast beam cracks Download PDFInfo
- Publication number
- CN116385434A CN116385434A CN202310643925.7A CN202310643925A CN116385434A CN 116385434 A CN116385434 A CN 116385434A CN 202310643925 A CN202310643925 A CN 202310643925A CN 116385434 A CN116385434 A CN 116385434A
- Authority
- CN
- China
- Prior art keywords
- target
- crack
- pixel point
- target area
- probability
- 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.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 21
- 238000003708 edge detection Methods 0.000 claims abstract description 23
- 238000000034 method Methods 0.000 claims description 29
- 239000013598 vector Substances 0.000 claims description 12
- 238000012163 sequencing technique Methods 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000012986 modification Methods 0.000 abstract description 5
- 230000004048 modification Effects 0.000 abstract description 5
- 238000012545 processing Methods 0.000 abstract description 3
- 238000005286 illumination Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 230000008646 thermal stress Effects 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- 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
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/761—Proximity, similarity or dissimilarity measures
-
- 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
- G06T2207/10024—Color image
-
- 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
- G06T2207/10048—Infrared image
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20076—Probabilistic image processing
-
- 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
- G06T2207/30132—Masonry; Concrete
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Abstract
The invention relates to the technical field of data processing and discloses an intelligent detection method for precast beam cracks, which comprises the steps of obtaining target pixel points in an RGB image and a first target area in a thermal infrared image, obtaining a second target area in the first target area, using the minimum threshold value of the updated target pixel points and the minimum threshold value of other pixel points in the RGB image as the minimum threshold value corresponding to each pixel point when the RGB image is subjected to Canny edge detection to obtain all edges, using the thermal edge probability of the first target area corresponding to the suspected cracks in the edges to update the corresponding crack probability of each suspected crack to obtain all cracks, and avoiding the problem of introducing a large number of noise points due to global modification of a threshold value scaling coefficient; a more accurate edge area and a complete slit edge are obtained.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to an intelligent detection method for a precast beam crack.
Background
In the construction engineering, prefabricating a precast beam in a factory, then transporting the precast beam to a construction site to install and fix the precast beam according to the position required by design, and detecting cracks on the precast beam, wherein the precast beam is large in size and difficult to accurately identify manually, so that high-altitude shooting is often adopted, and an image is analyzed to obtain a crack image on the precast beam;
at present, when a crack image in an image is detected, the crack image is detected through threshold segmentation, but the method has higher detection precision only when the crack is obvious, but the precast beam has larger volume, the field environment is complex, and shadows exist, so that some crack areas cannot be detected.
Disclosure of Invention
The invention is used for solving the problems that the existing crack detection precision is too low and the crack on a precast beam cannot be accurately detected, and provides an intelligent detection method for the quality of the building engineering, which is used for identifying unclear cracks on RGB images through a thermal infrared image and obtaining complete cracks, and comprises the following steps:
acquiring RGB images and thermal infrared images of the precast beams;
acquiring a first target area in a thermal infrared image; acquiring the hot edge probability of each first target area;
obtaining a double threshold value for Canny edge detection of an RGB image, extracting target pixel points corresponding to a gradient value smaller than a minimum threshold value in the double threshold value in the RGB image, obtaining pixel points corresponding to each target pixel point in a thermal infrared image, and obtaining a second target area corresponding to each target pixel point in a first target area according to Euclidean distance from the pixel point corresponding to each target pixel point to each first target area;
acquiring a threshold scaling coefficient of each target pixel point according to the hot edge probability of a second target area corresponding to each target pixel point, and updating the minimum threshold value of the target pixel point when double-threshold detection is performed according to the threshold scaling coefficient of each target pixel point to obtain the self-adaptive minimum threshold value of the target pixel point when the double-threshold detection is performed on the RGB image;
the updated self-adaptive minimum threshold value of the target pixel point and the minimum threshold values of other pixel points in the RGB image are used as the minimum threshold value corresponding to each pixel point when the channel edge detection is carried out on the RGB image, and all edges are obtained through the channel edge detection;
acquiring the crack probability that each edge belongs to a crack edge, and determining cracks and suspected cracks in all edges by using the crack probability of each edge;
updating the corresponding crack probability of the suspected cracks by using the hot edge probability of the first target area corresponding to the suspected cracks in the hot infrared image to obtain updated crack probability of each suspected crack, and determining the cracks in all the suspected cracks by using the updated crack probability of each suspected crack.
Further, the method for updating the corresponding crack probability of the suspected crack by using the hot edge probability of the first target area corresponding to the suspected crack in the thermal infrared image comprises the following steps:
obtaining the similarity between each suspected crack and each first target area;
obtaining a first target area matched with each suspected crack by using the obtained similarity;
and updating the crack probability after updating the crack probability of the edge corresponding to each suspected crack, the hot edge probability of the matched first target area and the similarity.
Further, the method for obtaining the hot edge probability of each first target area comprises the following steps:
acquiring a midline of each first target area, and obtaining a perpendicular line of a tangent line of each pixel point on the midline;
extracting two intersection points of the vertical line corresponding to each pixel point on the middle line and the edge of the first target area;
respectively comparing the gray value of each pixel point on the middle line with the gray values of the two intersection points;
assigning values to the pixel points on the central line according to different comparison results;
and obtaining the hot edge probability of the first target area by using the assignment and the number of all pixel points on the central line of each first target area.
Further, the method for obtaining the second target area corresponding to each target pixel point in the first target area includes:
sequencing the first target areas according to the Euclidean distance from the pixel point corresponding to each target pixel point to each first target area, and taking the sequencing as a first target area sequence of each target pixel point;
acquiring a position of a first target area corresponding to the maximum hot edge probability in a first target area sequence of each target pixel point as a target position;
and taking all the first target areas in front of the target position and the first target areas positioned at the target position as second target areas of each target pixel point.
Further, the method for obtaining the threshold scaling factor of each target pixel point comprises the following steps:
the formula of the threshold scaling factor for each target pixel is as follows:
wherein:
expressed as a threshold scaling factor, ">
Representing a hot edge probability of an ith second target region; n represents the number of second target areas corresponding to each target pixel point; />
Representing the Euclidean distance from the pixel point corresponding to the target pixel point to the ith second target area; />
Expressed as the sum of the hot edge probabilities of all the second target areas corresponding to each target pixel point.
Further, the method for obtaining similarity of each suspected crack and each first target area comprises the following steps:
acquiring the minimum circumscribed rectangle of each suspected crack and each first target area;
decomposing the image matrix in each minimum circumscribed rectangle to obtain a plurality of feature vectors;
and matching the feature vector of the minimum circumscribed rectangle of each suspected crack with the feature vector of the minimum circumscribed rectangle of each first target area to obtain the similarity between each suspected crack and the first target area.
The beneficial effects of the invention are as follows: according to the invention, the target pixel point in the RGB image and the first target area in the thermal infrared image are obtained by prefabricating the Liang Dere infrared image and the RGB image, the Euclidean distance from each target pixel point to each first target area is utilized to obtain the second target area in the first target area, the threshold scaling factor of each target pixel point is obtained according to the thermal edge probability of the second target area corresponding to each target pixel point, the minimum threshold value of each target pixel point when the target pixel point is subjected to Canny edge detection is updated by utilizing the threshold scaling factor of each target pixel point, the minimum threshold value of each updated target pixel point and the minimum threshold value of other pixel points in the RGB image are utilized to carry out Canny edge detection on the minimum threshold value of each pixel point when the RGB image is subjected to Canny edge detection, then the update crack probability of each suspected crack is obtained by utilizing the thermal edge probability of the first target area corresponding to the suspected crack in the edge, and the update crack probability of each suspected crack is obtained, and the global scaling factor is avoided by calculating the threshold value of each pixel point, and the global scaling factor is avoided; by mutual verification and complementation of the thermal infrared edge and the canny edge, a more accurate edge area and a complete crack edge are obtained.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the invention, and that other drawings can be obtained according to these drawings without inventive faculty for a person skilled in the art.
FIG. 1 is a schematic flow chart of the method of the present invention;
FIG. 2 is a flow chart of a method for updating the probability of a suspected crack according to an embodiment of the present invention;
FIG. 3 is a flowchart illustrating a method for obtaining a second target area corresponding to each target pixel point in a first target area according to an embodiment of the present invention;
FIG. 4 is an image of a crack in an embodiment of the present invention;
fig. 5 is an image of the marking of the insignificant crack of fig. 4.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Example 1
The present invention as shown in fig. 1 provides an intelligent detection method for a crack of a precast beam, comprising:
the RGB image and the thermal infrared image of the precast beam are acquired through the multispectral camera, and when the multispectral camera is acquired, the multispectral camera can be moved to acquire all the precast Liang Tuxiang (RGB image and thermal infrared image), then the precast Liang Tuxiang is formed by splicing (the splicing method is common knowledge in the technical field and is not described in detail here), the subsequent processing is convenient, the single image can be processed, the positions of the acquired RGB image and the image in the thermal infrared image are required to be identical, and if the positions of the pixels in the RGB image and the image represented by the pixels in the same position of the thermal infrared image (the image is the recorded object) are identical.
The temperature of the precast beam is increased due to illumination in the use process, when the temperature is increased, the thermal expansion coefficient of the concrete surface is larger, and the thermal expansion coefficients of the concrete at two sides of the crack are different, so that the change of the temperature gradient in the concrete can be caused, and the thermal stress effect is generated, and therefore, the temperature at two sides of the crack is higher than the temperature at the crack under the normal condition, and the embodiment acquires the precast Liang Dere infrared image to identify the unclear crack in the RGB image;
because the temperature of the crack in the thermal infrared image is higher than that of other positions, a high-temperature region, namely a first target region recorded in the embodiment, needs to be acquired in the thermal infrared image in order to obtain an accurate crack; obtaining a hot edge probability of the first target area according to the gray value of the first target area while obtaining the first target area, wherein the hot edge probability is used as the probability of cracks in the thermal infrared image, and the first target area with the hot edge probability lower than 0.7 (probability threshold) is removed to obtain all the removed first target areas as final first target areas in the thermal infrared image;
since a larger error exists when the RGB image carries out edge detection through a canny, as shown in fig. 4 and 5, fig. 4 is a crack original image, fig. 5 is an image obtained by marking cracks in fig. 4 unobvious, in order to reduce false edges in the canny edge detection, a double-threshold method is adopted to reserve edge pixel points, the edge pixel points are reserved more than a high threshold, and whether high-threshold pixel points exist in neighborhood pixel points of a pixel point between the high threshold and the low threshold is reserved; deleting pixels smaller than the low threshold; the marked area in fig. 5 may be between the high and low thresholds, and the pixels with the high threshold may be deleted because of the larger neighborhood; the marked area may also be less than a low threshold, resulting in direct deletion; in order to avoid the occurrence of the situations, a marking area is reserved, and the relation between the pixel points and the edge in the thermal infrared image is used as a reference factor for pixel point reservation in the embodiment, so that the pixel points in the marking area can be reserved;
specifically, a double threshold value for Canny edge detection of an RGB image is obtained, a high threshold value in the double threshold value adopted when Canny edge detection is carried out on the RGB image is represented by K1, and a low threshold value is represented by K2, because a gradient of a certain edge is smaller when Canny edge detection is carried out, namely a gradient value is smaller than K2, pixel points are easy to reject when detection is carried out, and the pixel points can be crack areas, in order to ensure that the pixel points can be stored, when edge detection is carried out on the pixel points, the low threshold value K2 is required to be adaptively adjusted to obtain the self-adaptive low threshold value K2 so as to ensure that the pixel points can not be subjected to edge rejection when edge detection is carried out on the pixel points; when a pixel point which is easy to be removed is obtained, extracting a pixel point of which the gradient value is smaller than a minimum threshold value K2 in the double threshold values in the RGB image, and taking the pixel point as a target pixel point; because the thermal infrared image is required to be cited to identify the crack, after the target pixel point is acquired, the pixel point at the position of the target pixel point in the thermal infrared image is required to be acquired, and the pixel point at the position of the rain target pixel point in the thermal infrared image is taken as the corresponding pixel point in the thermal infrared image;
because of the distance relation between the corresponding pixel point in the target pixel point thermal infrared image and the adjacent high-temperature region, if a certain characteristic unobvious region is close to the Euclidean distance of one high-probability high-temperature region, the threshold requirement of the region can be lowered as a basis, so that the region is reserved, and in view of the requirement, second target regions meeting the conditions in all first target regions are screened out;
the Euclidean distance from each pixel point corresponding to each target pixel point to each first target area is mainly utilized to obtain a second target area corresponding to each target pixel point in the first target area;
and then obtaining the threshold scaling factor of each target pixel point according to the hot edge probability of the second target area corresponding to each target pixel point.
The threshold scaling factor is obtained by subtracting an inverse threshold scaling factor from 1, wherein the inverse threshold scaling factor is that the result value is opposite to the threshold scaling factor, the smaller the Euclidean distance between the second target region and the pixel point corresponding to the target pixel point is, the larger the thermal edge probability of the second target region is, the larger the corresponding inverse threshold scaling factor is, namely a larger value is obtained, but because the value is taken as a coefficient of a low threshold value, the larger the value is, the larger the inverse threshold scaling factor is subtracted from 1 after the low threshold value is corrected, and the threshold scaling factor is obtained; each point has a threshold scaling factor, so that the introduction of a large number of noise points caused by global modification of the threshold scaling factor is avoided;
updating the minimum threshold value of the target pixel point when the double-threshold detection is carried out according to the threshold value scaling coefficient of each target pixel point to obtain the self-adaptive minimum threshold value of the target pixel point when the double-threshold detection is carried out on the RGB image; specifically, taking the product of the threshold scaling coefficient of the target pixel point and k2 as an updated low threshold corresponding to the target pixel point as an adaptive minimum threshold;
then, using the updated self-adaptive minimum threshold value of the target pixel point and the minimum threshold values of other pixel points in the RGB image as the minimum threshold value corresponding to each pixel point when the channel edge detection is carried out on the RGB image, and carrying out the channel edge detection to obtain all edges;
normalizing by using the ratio of the average gradient of all points on each edge to a low threshold value through the maximum ratio to obtain the ratio of each edge, wherein the ratio is used as the edge probability of each edge, namely the crack probability of each edge belonging to the crack edge, and determining the cracks and suspected cracks in all edges by using the crack probability of each edge; taking the cracks in all the edges as recognized cracks, taking the edges as cracks when the crack probability of the edges is larger than 0.7 and taking the edges as suspected cracks when the crack probability of the edges is smaller than 0.7 when judging whether the edges are cracks or not; then judging whether the suspected crack is a crack or not;
the method specifically comprises the steps of updating the corresponding crack probability of the suspected cracks by using the hot edge probability of the first target area corresponding to the suspected cracks in the thermal infrared image to obtain updated crack probability of each suspected crack, determining the cracks in all the suspected cracks by using the updated crack probability of each suspected crack, wherein when the updated crack probability value of the suspected cracks is greater than 0.7, the suspected cracks are cracks, and when the updated crack probability value of the suspected cracks is less than 0.7, the suspected cracks are not cracks, so that the identification of all the cracks in the RGB image is completed, the accuracy of crack identification is improved, and accurate parameters are provided for the subsequent judgment of the quality of the precast beams.
Example 2
Based on embodiment 1, due to the heat transfer on the thermal infrared image, the boundary of the crack is blurred, so that whether the suspected crack is a crack cannot be accurately judged by the probability of the crack, and in the embodiment, in order to accurately identify the crack in the suspected crack, the crack region is obtained by matching the RGB edge with the thermal infrared edge, and meanwhile, the high-precision crack edge is obtained.
The larger the crack probability of the edge area is, the larger the hot edge probability of the corresponding matched second target area is, namely the edge probability of the edge area is larger, namely the original edge probability is smaller, but the probability of the corresponding matched high-temperature area is larger, and the probability of the edges is larger if the edge shapes are similar;
in order to ensure that the edge probability of the suspected crack is more accurate, as shown in fig. 2, the method for updating the corresponding crack probability of the suspected crack by using the thermal edge probability of the first target area corresponding to the suspected crack in the thermal infrared image includes:
obtaining the similarity between each suspected crack and each first target area;
when the similarity is acquired, firstly acquiring the minimum circumscribed rectangle of each suspected crack and each first target area;
performing K-SVD decomposition on the image matrix in each minimum circumscribed rectangle to obtain a plurality of feature vectors and feature values corresponding to each feature vector, keeping the maximum ten feature values of the feature vectors obtained by each minimum circumscribed rectangle and the corresponding feature vectors, taking the feature vector obtained by the minimum circumscribed rectangle of the suspected crack as a node on one side, taking the feature vector obtained by the minimum circumscribed rectangle of the first target area as a node on the other side, calculating by adopting a maximum matching principle in KM matching to obtain a matching result of each suspected crack area and each first target area, and taking the edge value of the matching result as the similarity of each suspected crack and each first target area, wherein the KM matching in the embodiment belongs to a conventional KM matching method in the field, and the specific process of the KM matching method is not specifically explained herein;
when the similarity between the suspected crack and the first target area is greater than or equal to 0.7, the first target area is used as a matching target area of the suspected crack; obtaining a first target area matched with all suspected cracks;
updating the crack probability of the edge corresponding to each suspected crack, the hot edge probability of the matched first target area and the similarity to obtain updated crack probability;
the formula for updating the crack probability is as follows:
；
wherein:
indicating the probability of updating cracks after suspected crack updating, < ->
Representing a hot edge probability of a first target area matched with the suspected crack; />
Representing a fracture probability before the suspected fracture is updated; />
Representing the similarity of the suspected crack to the first target area to which it is matched, from which the formula can be seen +.>
The larger the probability that the suspected crack is a crack edge, the larger; similarity->
The larger the probability that the suspected crack is a crack edge, the greater.
Example 3
Each first target area (area with higher temperature) is not necessarily a crack area, and can be reflected by illumination of a certain area due to different illumination conditions, so that the temperature of a certain non-crack area is higher due to higher light intensity; the difference is made by the nature of the high temperature areas of the crack, when the temperature is raised, the thermal expansion coefficient of the concrete surface is larger, and the thermal expansion coefficients of the concrete at two sides of the crack are different, so that the change of the internal temperature gradient of the concrete is caused, and the thermal stress effect is generated, therefore, the temperature at two sides of the crack is higher than the temperature at the crack in general, and in view of the embodiment, whether the first target area is the crack area is judged by the hot edge of each first target area; therefore, judging whether the first target area is a crack area or not, and obtaining the hot edge probability of each first target area;
thus, on the basis of embodiment 1, the present embodiment provides a method for hot edge probability of each first target area as shown in fig. 2, including:
extracting a skeleton from each first target area to obtain a central line of each first target area, obtaining all pixel points on each central line, and making a vertical line through a tangent line of each pixel point on the central line;
acquiring two intersection points of the vertical line and the corresponding edge of the first target area; thus obtaining two intersection points corresponding to all pixel points on the central line;
respectively comparing the gray value of each pixel point on the middle line with the gray values of the two corresponding intersection points;
assigning values to the pixel points on the central line according to different comparison results; specifically, when the gray value of the pixel point on the middle line is smaller than the pixel value of the two corresponding intersection points, the pixel point on the middle line is assigned 2, when the gray value of the pixel point on the middle line is smaller than the pixel value of one of the two intersection points, the pixel point on the middle line is assigned 1, and when the gray value of the pixel point on the middle line is larger than the pixel value of the two corresponding intersection points, the pixel point on the middle line is assigned 0, and the assignment of all the pixel points on the middle line is completed according to the assignment;
the hot edge probability of the first target area is obtained by using the assignment and the quantity of all pixel points on the central line of each first target area; the specific expression is:
Example 4
The double-threshold screening in the Canny edge detection process may have larger errors, if the gradient of a certain edge is smaller, the edge is often screened out directly, but the edge may actually belong to a crack area, and the characteristics are not obvious. Therefore, according to the embodiment, firstly, through the distance relation between the corresponding pixel point of the target pixel point in the thermal infrared image and the adjacent high-temperature region, if the Euclidean distance between a certain feature unobvious region and a high-probability high-temperature region is similar, the threshold requirement of the region can be reduced as a basis, so that the region is reserved;
therefore, the method for acquiring the second target area corresponding to each target pixel point in the first target area according to the embodiment shown in fig. 3 includes:
sequencing the first target areas according to the Euclidean distance from the pixel point corresponding to each target pixel point to each first target area, and taking the sequencing as a first target area sequence of each target pixel point;
acquiring a position of a first target area corresponding to the maximum hot edge probability in a first target area sequence of each target pixel point as a target position;
taking all the first target areas in front of the target position and the first target areas positioned at the target position as second target areas of each target pixel point;
examples: such as a first target area: b1, B2, B3, B4, B5, B6, B7, B8, B9, B10;
one target pixel point A of each target pixel point is obtained, and the Euclidean distance between the pixel point corresponding to the target pixel point A and each first target area is as follows:
sequencing the target pixel points according to the Euclidean distance, wherein the sequencing is performed according to the sequence from small to large, and the sequence from the target pixel point A to the Euclidean distance of each target pixel point is as follows:
if the hot edge probability of B7 is the largest among the hot edge probabilities of the first target areas B1, B2, B3, B4, B5, B6, B7, B8, B9, and B10, the screening of the second target areas corresponding to the target pixel points a is completed according to the method that all the first target areas before the target position and the first target areas located at the target position are used as the second target areas of each target pixel point, wherein the obtained second target areas are B3, B5, B1, B8, B2, and B7;
after a second target area corresponding to the target pixel point is obtained, the method for obtaining the threshold scaling factor of each target pixel point is as follows:
the formula of the threshold scaling factor for each target pixel is as follows:
wherein:
expressed as a threshold scaling factor, ">
Representing a hot edge probability of an ith second target region; n represents the number of second target areas corresponding to each target pixel point; />
Representing the Euclidean distance from the pixel point corresponding to the target pixel point to the ith second target area; />
Expressed as the sum of the hot edge probabilities of all the second target areas corresponding to each target pixel point.
For each target pixel point, obtaining an inverse threshold scaling factor of each point by a distance weighting method of the second target area:
the inverse threshold scaling factor is that the result value is opposite to the wanted threshold scaling factor, the smaller the Euclidean distance between the second target area and the target pixel point is, the larger the hot edge probability of the second target area is, the larger the corresponding inverse threshold scaling factor is, namely, a larger value is obtained, but because the value is a coefficient used as a low threshold, the larger the value is, the larger the low threshold is after correction, and therefore, the inverse threshold scaling factor needs to be subtracted by 1 to obtain the threshold scaling factor; each point has a threshold scaling factor, so that the introduction of a large number of noise points caused by global modification of the threshold scaling factor is avoided;
and calculating to obtain a threshold scaling coefficient of each point smaller than the low threshold K2, taking the product of the threshold scaling coefficient and the low threshold K2 as an updated low threshold corresponding to the point, and completing edge detection by utilizing other processes of canny edge detection by using the updated low threshold and the original high threshold of each target pixel point.
And calculating the ratio of the average gradient of all points on each edge to the low threshold value for each edge obtained by canny edge detection, and normalizing the ratio by the maximum ratio to obtain the ratio of each edge as the edge probability of each edge, wherein the edge probability is the probability that each edge belongs to a crack edge, so that the problem of introducing a large number of noise points caused by global modification of a threshold scaling factor is avoided, and the identification precision of the crack is improved.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (6)
1. An intelligent detection method for a precast beam crack is characterized by comprising the following steps:
acquiring RGB images and thermal infrared images of the precast beams;
acquiring a first target area in a thermal infrared image; acquiring the hot edge probability of each first target area;
obtaining a double threshold value for Canny edge detection of an RGB image, extracting target pixel points corresponding to a gradient value smaller than a minimum threshold value in the double threshold value in the RGB image, obtaining pixel points corresponding to each target pixel point in a thermal infrared image, and obtaining a second target area corresponding to each target pixel point in a first target area according to Euclidean distance from the pixel point corresponding to each target pixel point to each first target area;
acquiring a threshold scaling coefficient of each target pixel point according to the hot edge probability of a second target area corresponding to each target pixel point, and updating the minimum threshold value of the target pixel point when double-threshold detection is performed according to the threshold scaling coefficient of each target pixel point to obtain the self-adaptive minimum threshold value of the target pixel point when the double-threshold detection is performed on the RGB image;
the updated self-adaptive minimum threshold value of the target pixel point and the minimum threshold values of other pixel points in the RGB image are used as the minimum threshold value corresponding to each pixel point when the channel edge detection is carried out on the RGB image, and all edges are obtained through the channel edge detection;
acquiring the crack probability that each edge belongs to a crack edge, and determining cracks and suspected cracks in all edges by using the crack probability of each edge;
updating the corresponding crack probability of the suspected cracks by using the hot edge probability of the first target area corresponding to the suspected cracks in the hot infrared image to obtain updated crack probability of each suspected crack, and determining the cracks in all the suspected cracks by using the updated crack probability of each suspected crack.
2. The intelligent detection method for a precast beam crack according to claim 1, wherein the method for updating the corresponding crack probability of the suspected crack by using the thermal edge probability of the first target area corresponding to the suspected crack in the thermal infrared image comprises:
obtaining the similarity between each suspected crack and each first target area;
obtaining a first target area matched with each suspected crack by using the obtained similarity;
and updating the crack probability after updating the crack probability of the edge corresponding to each suspected crack, the hot edge probability of the matched first target area and the similarity.
3. The intelligent detection method for precast beam cracks according to claim 1, wherein the method for obtaining the hot edge probability of each first target area comprises:
acquiring a midline of each first target area, and obtaining a perpendicular line of a tangent line of each pixel point on the midline;
extracting two intersection points of the vertical line corresponding to each pixel point on the middle line and the edge of the first target area;
respectively comparing the gray value of each pixel point on the middle line with the gray values of the two intersection points;
assigning values to the pixel points on the central line according to different comparison results;
and obtaining the hot edge probability of the first target area by using the assignment and the number of all pixel points on the central line of each first target area.
4. The intelligent detection method for a precast beam crack according to claim 1, wherein the method for obtaining the second target area corresponding to each target pixel point in the first target area comprises the following steps:
sequencing the first target areas according to the Euclidean distance from the pixel point corresponding to each target pixel point to each first target area, and taking the sequencing as a first target area sequence of each target pixel point;
acquiring a position of a first target area corresponding to the maximum hot edge probability in a first target area sequence of each target pixel point as a target position;
and taking all the first target areas in front of the target position and the first target areas positioned at the target position as second target areas of each target pixel point.
5. The intelligent detection method for precast beam cracks according to claim 1 or 4, wherein the method for obtaining the threshold scaling factor of each target pixel point comprises:
the formula of the threshold scaling factor for each target pixel is as follows:
wherein:
expressed as a threshold scaling factor, ">
Representing a hot edge probability of an ith second target region; n represents the number of second target areas corresponding to each target pixel point; />
Representing the Euclidean distance from the pixel point corresponding to the target pixel point to the ith second target area; />
Expressed as the sum of the hot edge probabilities of all the second target areas corresponding to each target pixel point.
6. The intelligent detection method for precast beam cracks according to claim 2, wherein the method for obtaining similarity of each suspected crack to each first target area comprises:
acquiring the minimum circumscribed rectangle of each suspected crack and each first target area;
decomposing the image matrix in each minimum circumscribed rectangle to obtain a plurality of feature vectors;
and matching the feature vector of the minimum circumscribed rectangle of each suspected crack with the feature vector of the minimum circumscribed rectangle of each first target area to obtain the similarity between each suspected crack and the first target area.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310643925.7A CN116385434B (en) | 2023-06-02 | 2023-06-02 | Intelligent detection method for precast beam cracks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310643925.7A CN116385434B (en) | 2023-06-02 | 2023-06-02 | Intelligent detection method for precast beam cracks |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116385434A true CN116385434A (en) | 2023-07-04 |
| CN116385434B CN116385434B (en) | 2023-08-08 |
Family
ID=86971424
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310643925.7A Active CN116385434B (en) | 2023-06-02 | 2023-06-02 | Intelligent detection method for precast beam cracks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116385434B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116703907A (en) * | 2023-08-04 | 2023-09-05 | 合肥亚明汽车部件有限公司 | Machine vision-based method for detecting surface defects of automobile castings |
| CN116958182A (en) * | 2023-09-20 | 2023-10-27 | 广东华宸建设工程质量检测有限公司 | Quick concrete crack detection method based on image data |
Citations (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016107408A1 (en) * | 2014-12-30 | 2016-07-07 | 江苏科技大学 | Infrared vision sensing detection method and device for narrow-gap weld seam deviation |
| WO2016127883A1 (en) * | 2015-02-12 | 2016-08-18 | 阿里巴巴集团控股有限公司 | Image area detection method and device |
| JP2017053819A (en) * | 2015-09-11 | 2017-03-16 | 国立大学法人富山大学 | Crack detection method and detection program of concrete |
| WO2017119154A1 (en) * | 2016-01-07 | 2017-07-13 | 三菱電機株式会社 | Detection device and detection method |
| US20170301078A1 (en) * | 2016-04-15 | 2017-10-19 | ecoATM, Inc. | Methods and systems for detecting cracks in electronic devices |
| GB201719239D0 (en) * | 2017-11-20 | 2018-01-03 | Gkn Aerospace Services Ltd | Defect detection and correction |
| WO2018008370A1 (en) * | 2016-07-06 | 2018-01-11 | キヤノン株式会社 | Information processing device, information processing method and program |
| CN109544546A (en) * | 2018-11-30 | 2019-03-29 | 电子科技大学 | A kind of thermal-induced imagery defect characteristic extracting method based on multiple-objection optimization |
| CN109886939A (en) * | 2019-01-30 | 2019-06-14 | 西安电子科技大学 | Bridge Crack detection method based on Tensor Voting |
| CN110136196A (en) * | 2019-01-30 | 2019-08-16 | 西安电子科技大学 | A kind of Bridge Crack width method for automatic measurement |
| AU2020101011A4 (en) * | 2019-06-26 | 2020-07-23 | Zhejiang University | Method for identifying concrete cracks based on yolov3 deep learning model |
| CN111784677A (en) * | 2020-07-03 | 2020-10-16 | 安徽炬视科技有限公司 | Single-camera knife switch arm opening angle identification method based on scene analysis |
| US20210082098A1 (en) * | 2019-09-18 | 2021-03-18 | Rupesh Wasudevrao KUMBHARE | Method and system for detection of anomalies in surfaces |
| CN112529880A (en) * | 2020-12-16 | 2021-03-19 | 武汉巨合科技有限公司 | Dam image crack detection device and method based on binocular vision |
| WO2021135372A1 (en) * | 2019-12-30 | 2021-07-08 | 歌尔股份有限公司 | Product defect detection method, device and system |
| CN113436168A (en) * | 2021-06-25 | 2021-09-24 | 重庆邮电大学 | Crack defect detection method for industrial weld X-ray image |
| CN113643245A (en) * | 2021-07-26 | 2021-11-12 | 深圳市鑫信腾科技股份有限公司 | Screen defect measuring method and device and computer readable storage medium |
| CN114419025A (en) * | 2022-01-27 | 2022-04-29 | 江苏泰和木业有限公司 | Fiberboard quality evaluation method based on image processing |
| CN114723681A (en) * | 2022-03-22 | 2022-07-08 | 江苏禹润智能科技有限公司 | Concrete crack defect detection method based on machine vision |
| CN115965623A (en) * | 2023-03-16 | 2023-04-14 | 深圳市岑科实业有限公司 | Surface flaw detection method and system in transformer production |
-
2023
- 2023-06-02 CN CN202310643925.7A patent/CN116385434B/en active Active
Patent Citations (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016107408A1 (en) * | 2014-12-30 | 2016-07-07 | 江苏科技大学 | Infrared vision sensing detection method and device for narrow-gap weld seam deviation |
| WO2016127883A1 (en) * | 2015-02-12 | 2016-08-18 | 阿里巴巴集团控股有限公司 | Image area detection method and device |
| JP2017053819A (en) * | 2015-09-11 | 2017-03-16 | 国立大学法人富山大学 | Crack detection method and detection program of concrete |
| WO2017119154A1 (en) * | 2016-01-07 | 2017-07-13 | 三菱電機株式会社 | Detection device and detection method |
| US20170301078A1 (en) * | 2016-04-15 | 2017-10-19 | ecoATM, Inc. | Methods and systems for detecting cracks in electronic devices |
| WO2018008370A1 (en) * | 2016-07-06 | 2018-01-11 | キヤノン株式会社 | Information processing device, information processing method and program |
| GB201719239D0 (en) * | 2017-11-20 | 2018-01-03 | Gkn Aerospace Services Ltd | Defect detection and correction |
| CN109544546A (en) * | 2018-11-30 | 2019-03-29 | 电子科技大学 | A kind of thermal-induced imagery defect characteristic extracting method based on multiple-objection optimization |
| CN109886939A (en) * | 2019-01-30 | 2019-06-14 | 西安电子科技大学 | Bridge Crack detection method based on Tensor Voting |
| CN110136196A (en) * | 2019-01-30 | 2019-08-16 | 西安电子科技大学 | A kind of Bridge Crack width method for automatic measurement |
| AU2020101011A4 (en) * | 2019-06-26 | 2020-07-23 | Zhejiang University | Method for identifying concrete cracks based on yolov3 deep learning model |
| US20210082098A1 (en) * | 2019-09-18 | 2021-03-18 | Rupesh Wasudevrao KUMBHARE | Method and system for detection of anomalies in surfaces |
| WO2021135372A1 (en) * | 2019-12-30 | 2021-07-08 | 歌尔股份有限公司 | Product defect detection method, device and system |
| CN111784677A (en) * | 2020-07-03 | 2020-10-16 | 安徽炬视科技有限公司 | Single-camera knife switch arm opening angle identification method based on scene analysis |
| CN112529880A (en) * | 2020-12-16 | 2021-03-19 | 武汉巨合科技有限公司 | Dam image crack detection device and method based on binocular vision |
| CN113436168A (en) * | 2021-06-25 | 2021-09-24 | 重庆邮电大学 | Crack defect detection method for industrial weld X-ray image |
| CN113643245A (en) * | 2021-07-26 | 2021-11-12 | 深圳市鑫信腾科技股份有限公司 | Screen defect measuring method and device and computer readable storage medium |
| CN114419025A (en) * | 2022-01-27 | 2022-04-29 | 江苏泰和木业有限公司 | Fiberboard quality evaluation method based on image processing |
| CN114723681A (en) * | 2022-03-22 | 2022-07-08 | 江苏禹润智能科技有限公司 | Concrete crack defect detection method based on machine vision |
| CN115965623A (en) * | 2023-03-16 | 2023-04-14 | 深圳市岑科实业有限公司 | Surface flaw detection method and system in transformer production |
Non-Patent Citations (3)
| Title |
|---|
| ZHANG L 等: "Road crack detection using deep convolutional neural network", 《IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING》, pages 3708 - 3712 * |
| 严世强 等: "基于渗流和区域生长联合分析的红外图像裂缝病害检测方法", 《现代电子技术》, vol. 43, no. 18, pages 6 - 10 * |
| 王道峰;周传林;: "图像分析技术在混凝土桥梁裂缝识别与计算中的应用", 筑路机械与施工机械化, no. 11, pages 79 - 84 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116703907A (en) * | 2023-08-04 | 2023-09-05 | 合肥亚明汽车部件有限公司 | Machine vision-based method for detecting surface defects of automobile castings |
| CN116703907B (en) * | 2023-08-04 | 2023-10-27 | 合肥亚明汽车部件有限公司 | Machine vision-based method for detecting surface defects of automobile castings |
| CN116958182A (en) * | 2023-09-20 | 2023-10-27 | 广东华宸建设工程质量检测有限公司 | Quick concrete crack detection method based on image data |
| CN116958182B (en) * | 2023-09-20 | 2023-12-08 | 广东华宸建设工程质量检测有限公司 | Quick concrete crack detection method based on image data |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116385434B (en) | 2023-08-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116385434B (en) | Intelligent detection method for precast beam cracks | |
| CN108364280B (en) | Method and equipment for automatically describing structural crack and accurately measuring width | |
| CN111179243A (en) | Small-size chip crack detection method and system based on computer vision | |
| CN111292305A (en) | Improved YOLO-V3 metal processing surface defect detection method | |
| CN116596936B (en) | Solar wing sailboard quality detection method based on image data | |
| JP6099479B2 (en) | Crack detection method | |
| CN113658132A (en) | Computer vision-based structural part weld joint detection method | |
| CN115082466B (en) | PCB surface welding spot defect detection method and system | |
| CN115511889A (en) | Method for detecting welding defects on surface of solar cell panel bracket | |
| CN115063423B (en) | Self-adaptive identification method for cold and hot cracks of mechanical castings based on computer vision | |
| CN112132206A (en) | Image recognition method, training method of related model, related device and equipment | |
| JP2013238449A (en) | Crack detection method | |
| CN113034474A (en) | Test method for wafer map of OLED display | |
| CN116883408B (en) | Integrating instrument shell defect detection method based on artificial intelligence | |
| CN115131354A (en) | Laboratory plastic film defect detection method based on optical means | |
| CN112508857A (en) | Aluminum product surface defect detection method based on improved Cascade R-CNN | |
| CN115639248A (en) | System and method for detecting quality of building outer wall | |
| CN117058147A (en) | Environment-friendly plastic product defect detection method based on computer vision | |
| CN114004858A (en) | Method and device for identifying aviation cable surface code based on machine vision | |
| CN116862910B (en) | Visual detection method based on automatic cutting production | |
| CN110751623A (en) | Joint feature-based defect detection method, device, equipment and storage medium | |
| CN114943738A (en) | Sensor packaging curing adhesive defect identification method based on visual identification | |
| CN113643245A (en) | Screen defect measuring method and device and computer readable storage medium | |
| CN114581901A (en) | Method for extracting edges of ancient building wall contaminated inscription character images | |
| CN114066993A (en) | Power distribution cabinet control panel segmentation method based on machine vision |
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 |