CN-122023342-A - Unmanned aerial vehicle-based automatic detection method for corrosion-resistant coating quality of steel member

CN122023342ACN 122023342 ACN122023342 ACN 122023342ACN-122023342-A

Abstract

The application relates to the technical field of image processing, in particular to an unmanned aerial vehicle-based automatic detection method for the corrosion-resistant coating quality of a steel member, which comprises the steps of acquiring an image of the steel member based on an unmanned aerial vehicle, and acquiring each closed connected domain in the image of the steel member; determining a first abnormal characteristic value of each edge pixel point of each closed connected domain, determining the random degree of peak distribution in the curvature of all edge pixel points of each closed connected domain, determining the edge fluctuation degree of each closed connected domain by combining the texture characteristic difference of each edge pixel point on each closed connected domain and the pixel points in the local neighborhood of each edge pixel point, acquiring a second abnormal characteristic value of each closed connected domain, and identifying the corrosion falling-off region in the steel member image based on the second abnormal characteristic value. The application improves the anti-corrosion coating quality detection effect of the steel member.

Inventors

LIU DUO
ZHANG JIANDONG
YANG YANG
NI YA
XIE JUNXIAN
SU XINHUA

Assignees

苏交科集团股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260130

Claims (10)

1. The automatic detection method for the corrosion-resistant coating quality of the steel member based on the unmanned aerial vehicle is characterized by comprising the following steps of: acquiring each closed connected domain in the steel member image based on the unmanned aerial vehicle acquisition steel member image; Analyzing gradient differences of all edge pixel points on all closed connected domains and pixel points in local adjacent domains of the edge pixel points, and determining first abnormal characteristic values of all edge pixel points of all closed connected domains; Determining the random degree of peak distribution in the curvature of all edge pixel points of each closed connected domain, and determining the edge fluctuation degree of each closed connected domain by combining the texture characteristic difference of each edge pixel point on each closed connected domain and the pixel points in the local neighborhood of each edge pixel point; Combining the edge fluctuation degree of each closed communication domain and the first abnormal characteristic values of all edge pixel points on each closed communication domain to obtain a second abnormal characteristic value of each closed communication domain; and identifying the corrosion falling area in the steel member image based on the second abnormal characteristic value.
2. The automated inspection method for the quality of corrosion protection coating of a steel member based on an unmanned aerial vehicle according to claim 1, wherein the determination of the first abnormal characteristic value comprises: Calculating the average value of the differences of the gradient magnitudes of each edge pixel point on each closed connected domain and all the pixel points in the local neighborhood of the closed connected domain, marking the differences as first differences based on the average value of each edge pixel point on each closed connected domain and the edge pixel points in the local neighborhood of the closed connected domain, and determining the first abnormal characteristic value based on the differences of the gradient angles of each edge pixel point on each closed connected domain and the pixel points in the local neighborhood of the closed connected domain.
3. The automated steel member anti-corrosion coating quality detection method based on the unmanned aerial vehicle as claimed in claim 2, wherein the determining process of the first abnormal characteristic value is as follows: And calculating the difference value of the gradient angle between each edge pixel point on each closed connected domain and each pixel point in the local neighborhood of each edge pixel point, and calculating the square of the cosine value of the difference value, wherein the first abnormal characteristic value and the square of the cosine value are in negative correlation, and the first abnormal characteristic value and the first difference are in positive correlation.
4. The automated steel member corrosion prevention coating quality detection method based on the unmanned aerial vehicle as set forth in claim 3, wherein the expression of the first abnormal characteristic value is: In the formula (I), in the formula (II), For the first abnormal characteristic value of the u-th edge pixel point of each closed connected domain, sig () is a sigmoid normalization function, M is the number of pixels in the local neighborhood of the u-th edge pixel point of each closed connected domain, The average value of the pixel points at the ith edge of each closed connected domain, For the mean value of the a-th pixel point in the local neighborhood of the u-th edge pixel point of each closed connected domain, norm () is a normalization function, For the gradient angle of the pixel point at the u-th edge of each closed connected domain, For the gradient angle of the a-th pixel point in the local neighborhood of the u-th edge pixel point of each closed connected domain, cos () is a triangular cosine function, And presetting a value larger than 0, wherein when the a pixel point in the local neighborhood of the u-th edge pixel point is a non-edge pixel point, the average value of the a pixel point is 0.
5. The automated inspection method for the corrosion-resistant coating quality of the steel member based on the unmanned aerial vehicle according to claim 1, wherein the determining the degree of randomness of the peak distribution in the curvature of all the edge pixel points of each closed connected domain comprises: The curvatures of all the edge pixel points of each closed connected domain are uniformly and sequentially arranged, the peak values of the curvatures of all the edge pixel points of each closed connected domain after arrangement are obtained, and the random degree of peak value distribution in the curvatures of all the edge pixel points of each closed connected domain is determined based on the number of the curvatures between adjacent peak values.
6. The automated inspection method for corrosion protection coating quality of steel members based on unmanned aerial vehicle according to claim 5, wherein if the number of curvatures between any peak and its adjacent previous peak is equal to the number of curvatures between any peak and its adjacent next peak, marking any peak as 0, otherwise marking any peak as 1, and accumulating the sum of the marked values of all peaks of curvatures of all edge pixels of each closed connected domain as the random degree of peak distribution in the curvatures of all edge pixels of each closed connected domain.
7. The automated inspection method for the quality of corrosion protection coating of steel components based on unmanned aerial vehicle according to claim 1, wherein the determining the edge waviness of each closed communication domain comprises: Calculating the difference of LBP values of each edge pixel point on each closed connected domain and the pixel points in the local neighborhood of the edge pixel points, marking the difference as a second difference, and obtaining a fusion result of the second difference of all the edge pixel points on each closed connected domain, wherein the edge fluctuation degree, the randomness degree and the fusion result are positively correlated.
8. The automated inspection method of corrosion protection coating quality for steel components based on unmanned aerial vehicle according to claim 7, wherein the edge waviness of each closed connected domain is the product of the randomness and the fusion result.
9. The automated inspection method for the quality of corrosion protection coating of steel members based on unmanned aerial vehicle according to claim 1, wherein the determination of the second abnormal characteristic value of each closed-loop communication domain comprises: And calculating the average value of the first abnormal characteristic values of all the edge pixel points on each closed connected domain, and multiplying the average value by the normalized value of the edge fluctuation degree of each closed connected domain to obtain the second abnormal characteristic value of each closed connected domain.
10. The automated inspection method for the corrosion protection coating quality of the steel member based on the unmanned aerial vehicle according to claim 1, wherein the identifying the corrosion falling area in the steel member image based on the second abnormal characteristic value comprises: Threshold segmentation is carried out on second abnormal characteristic values of all closed contours in the steel member image, if the closed contours with the second abnormal characteristic values larger than the segmentation threshold value exist in the steel member image, the corrosion shedding area exists in the steel member image, otherwise, the corrosion shedding area does not exist in the steel member image.

Description

Unmanned aerial vehicle-based automatic detection method for corrosion-resistant coating quality of steel member Technical Field The application relates to the technical field of image processing, in particular to an unmanned aerial vehicle-based automatic detection method for the corrosion-resistant coating quality of a steel member. Background Along with the continuous promotion of the high-speed development and the industrialized development of the national economic construction, the steel structure is widely applied in various fields at present due to the advantages of high strength, light dead weight, good anti-seismic performance, short construction period and the like. The steel structure is commonly used in various engineering structures with large span, large height, large load and large power action. At present, the quality assurance and service life of the steel structure are greatly focused and valued at the same time of wide application, and the anti-corrosion coating is a key link for prolonging the service life of the steel structure. The steel structure anti-corrosion coating is a key for guaranteeing the durability and safety of the steel structure member, and directly influences the service life of the steel member. In corrosive environments such as humidity, acid rain, salt fog, industrial pollution and the like, steel structures which are not protected by the coating can be rapidly corroded, the bearing capacity of the steel structures is seriously influenced, and even safety accidents can be possibly caused. Through using anticorrosive coating to the steel component to protect, guarantee the anticorrosive application quality of steel component simultaneously, can effectively keep apart moisture, oxygen and corrosive medium, effectively reduce the possibility of steel component corrosion, improve the life of steel component. At present, a machine vision technology is generally adopted in the prior art to detect corrosion-falling areas in a steel member image, so that the corrosion-resistant coating quality detection of the steel member is finished, however, the characteristic differences of the corrosion-falling areas such as the shape, the size, the color, the texture and the like are large, the initial corrosion can be only tiny spots, the serious corrosion can cause large-area falling, different types of corrosion areas are different in expression form, the existing detection algorithm easily causes that some corrosion areas are ignored or mistakenly identified, the identification precision of the corrosion-falling areas is influenced, and the corrosion-resistant coating quality detection effect of the steel member is reduced. Disclosure of Invention In order to solve the technical problems, the application provides an unmanned aerial vehicle-based automatic detection method for the corrosion-resistant coating quality of a steel member, which aims to solve the existing problems. The application discloses an unmanned aerial vehicle-based automatic detection method for the corrosion-resistant coating quality of a steel member, which adopts the following technical scheme: the embodiment of the application provides an unmanned aerial vehicle-based automatic detection method for the corrosion-resistant coating quality of a steel member, which comprises the following steps: acquiring each closed connected domain in the steel member image based on the unmanned aerial vehicle acquisition steel member image; Analyzing gradient differences of all edge pixel points on all closed connected domains and pixel points in local adjacent domains of the edge pixel points, and determining first abnormal characteristic values of all edge pixel points of all closed connected domains; Determining the random degree of peak distribution in the curvature of all edge pixel points of each closed connected domain, and determining the edge fluctuation degree of each closed connected domain by combining the texture characteristic difference of each edge pixel point on each closed connected domain and the pixel points in the local neighborhood of each edge pixel point; Combining the edge fluctuation degree of each closed communication domain and the first abnormal characteristic values of all edge pixel points on each closed communication domain to obtain a second abnormal characteristic value of each closed communication domain; and identifying the corrosion falling area in the steel member image based on the second abnormal characteristic value. In one embodiment, the determining of the first outlier includes: Calculating the average value of the differences of the gradient magnitudes of each edge pixel point on each closed connected domain and all the pixel points in the local neighborhood of the closed connected domain, marking the differences as first differences based on the average value of each edge pixel point on each closed connected domain and the edge pixel points in the local neighborhood of the closed