CN116883444B - Automobile damage detection method based on machine vision and image scanning - Google Patents
Automobile damage detection method based on machine vision and image scanning Download PDFInfo
- Publication number
- CN116883444B CN116883444B CN202310966064.6A CN202310966064A CN116883444B CN 116883444 B CN116883444 B CN 116883444B CN 202310966064 A CN202310966064 A CN 202310966064A CN 116883444 B CN116883444 B CN 116883444B
- Authority
- CN
- China
- Prior art keywords
- scratch
- edge
- damage
- segmentation
- automobile
- 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
- 230000006378 damage Effects 0.000 title claims abstract description 331
- 238000001514 detection method Methods 0.000 title claims abstract description 23
- 230000011218 segmentation Effects 0.000 claims abstract description 186
- 239000011159 matrix material Substances 0.000 claims abstract description 60
- 238000004364 calculation method Methods 0.000 claims abstract description 38
- 238000000034 method Methods 0.000 claims abstract description 22
- 238000012545 processing Methods 0.000 claims abstract description 10
- 238000003708 edge detection Methods 0.000 claims abstract description 6
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims description 35
- 238000013528 artificial neural network Methods 0.000 claims description 17
- 238000006243 chemical reaction Methods 0.000 claims description 8
- 238000007781 pre-processing Methods 0.000 claims description 6
- 210000002364 input neuron Anatomy 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 5
- 208000027418 Wounds and injury Diseases 0.000 claims description 4
- 208000014674 injury Diseases 0.000 claims description 4
- 238000012549 training Methods 0.000 claims description 4
- 239000003973 paint Substances 0.000 description 18
- 230000006870 function Effects 0.000 description 12
- 238000011156 evaluation Methods 0.000 description 6
- 230000004913 activation Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 239000003086 colorant Substances 0.000 description 3
- 230000004927 fusion Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000003709 image segmentation Methods 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 230000008439 repair process Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000002591 computed tomography Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- JEIPFZHSYJVQDO-UHFFFAOYSA-N iron(III) oxide Inorganic materials O=[Fe]O[Fe]=O JEIPFZHSYJVQDO-UHFFFAOYSA-N 0.000 description 1
- 235000020061 kirsch Nutrition 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000000275 quality assurance Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
-
- 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
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- 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
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The application relates to the technical field of automobile damage detection, in particular to an automobile damage detection method based on machine vision and image scanning, which comprises the following steps: acquiring a first image of an automobile, and performing pretreatment and segmentation to obtain a third image; performing edge detection processing and clustering processing on the third image to obtain a scratch damage area; edge segmentation calculation is carried out on edge pixel points of the scratch damage area to obtain an edge segmentation coefficient and scratch saliency; calculating the edge segmentation floating index and the number of the edge segmentation orders of the scratch damage area based on the edge segmentation coefficient; and constructing an edge run matrix, calculating to obtain scratch damage degree and scratch disorder degree, determining scratch category based on the scratch saliency, the scratch damage degree and the scratch disorder degree, and evaluating an automobile damage detection result. The method can accurately detect and identify the significance, damage degree and texture complexity of the scratch to detect and classify the scratch damage of the automobile.
Description
Technical Field
The invention relates to the technical field of automobile damage detection, in particular to an automobile damage detection method based on machine vision and image scanning.
Background
Automobiles are capital and technology intensive products, today as a special consumer product, the loss of value in use of automobiles is a very complex process of related multifactorial action, and automobiles are subject to a variety of tangible and intangible losses from factory commissioning, with scratch damage being the most common. The vehicle paint is one of the paint, and for the automobile, the vehicle paint has the functions of rust prevention, stain resistance and corrosion resistance besides the attractive appearance of the automobile, so that the vehicle paint is not the same as the traditional furniture paint, and is more complex in process than the furniture paint, and the process is also more complex when repair is needed.
Therefore, how to accurately evaluate the scratches on the automobile provides more efficient, accurate, convenient and safe damage evaluation service for automobile maintenance and insurance, and has very important practical significance. The existing image segmentation technology can only segment scratch damage areas, cannot accurately classify the scratch damage areas, further cannot accurately judge the damage degree of vehicles, adopts a neural network to identify and classify scratches, needs to identify and classify images by learning a large amount of data, and can be inaccurate under complex scenes and changing light conditions.
Disclosure of Invention
In view of the above problems, the present application provides an automobile damage detection method based on machine vision and image scanning, which can accurately detect and identify the category of scratches and the degree of scratch damage, thereby rapidly and accurately giving evaluation comments.
The embodiment of the application provides an automobile damage detection method based on machine vision and image scanning, which comprises the following steps:
acquiring a first image of the automobile, preprocessing the first image to obtain a second image, and performing appearance part segmentation processing on the second image to obtain a third image of a plurality of automobile appearance parts;
performing edge detection processing on the third image of each automobile appearance part to obtain edge images of a plurality of automobile appearance parts; clustering the edge images of each automobile appearance part to obtain a plurality of scratch damage areas;
performing edge segmentation calculation on each edge pixel point of each scratch damage area to obtain an edge segmentation coefficient of each edge pixel point; calculating the scratch saliency of each scratch damage area based on the edge segmentation coefficient of the edge pixel point;
Calculating the edge segmentation floating index based on the edge segmentation coefficients of the edge pixel points for each scratch damage area, and calculating the number of edge segmentation orders of each scratch damage area based on the edge segmentation floating index;
for each scratch damage area, constructing an edge run matrix based on the number of the edge segmentation orders and the edge segmentation coefficients, calculating to obtain the scratch damage degree and the scratch disorder degree of the scratch damage area based on the edge run matrix, and determining the scratch category of the scratch damage area based on the scratch saliency, the scratch damage degree and the scratch disorder degree of the scratch damage area;
and evaluating a damage detection result of the automobile exterior part based on the scratch category and the scratch damage degree of the scratch damage area.
In one possible implementation manner, the acquiring the first image of the automobile, and preprocessing the first image to obtain the second image includes:
image acquisition is carried out on the automobile by utilizing an image acquisition device at a specific angle to obtain a plurality of automobile projection images, wherein the automobile projection images are first images of the automobile;
Calculating a two-dimensional reconstruction image serving as an image to be registered according to a plurality of first images, and calculating a conversion relation between the image to be registered and a reference image based on characteristic point information of the image to be registered and pre-extracted reference image information;
and performing geometric transformation on the image to be registered to a reference image based on the conversion relation to obtain a second image of the automobile.
In one possible implementation manner, the performing edge segmentation calculation on each edge pixel point of each scratch damage area to obtain an edge segmentation coefficient of each edge pixel point includes:
constructing a neighborhood window for each edge pixel point of each scratch damage area, and calculating to obtain an edge segmentation coefficient of each edge pixel point, wherein the calculation formula of the edge segmentation coefficient is as follows:
wherein EDX (x) is an edge segmentation coefficient of the edge pixel point x,the average value of R channel values of edge pixel points and non-edge pixel points in the neighborhood window of the edge pixel point x is respectively; />The average value of the G channel values of the edge pixel points and the non-edge pixel points in the neighborhood window of the edge pixel point x is respectively; And the average values of the B channel values of the edge pixel points and the non-edge pixel points in the neighborhood window of the edge pixel point x are respectively.
In one possible implementation manner, the calculating, based on the edge segmentation coefficient of the edge pixel point, the scratch saliency of each scratch damage area includes:
calculating the scratch saliency of each scratch damage area according to the edge segmentation coefficient of the edge pixel point in each scratch damage area, wherein the calculation formula of the scratch saliency is as follows:
wherein HHO (k) For scratch saliency of the kth scratch damaged region, EDX (x i ) For the edge segmentation coefficient of the ith edge pixel point in the scratch damage area, n (k) Is the number of edge pixels in the kth scratch damage region.
In one possible implementation manner, the calculating, for each scratch damage area, an edge segmentation floating index based on the edge segmentation coefficient of the edge pixel point includes:
and calculating the edge segmentation floating index of each scratch damage area based on the edge segmentation coefficient of the edge pixel point, wherein the calculation formula of the edge segmentation floating index is as follows:
Therein, EFZ (k) The floating index is divided for the edge of the kth scratch damage region,and the variance of the edge segmentation coefficient of each edge pixel point in the kth scratch damage area is obtained.
In one possible implementation manner, the calculating, based on the edge segmentation floating index, the number of edge segmentation orders of each scratch damage region includes:
calculating the number of edge segmentation orders of each scratch damage area based on the edge segmentation floating index, wherein the calculation formula of the number of edge segmentation orders is as follows:
M (k) ＝round(EFZ (k) *M 0 )
wherein M is (k) The number of the edge segmentation orders of the kth scratch damage area; round () is a rounding function; EFZ (k) Splitting a floating index, M, for the edge of the kth scratch-damaged region 0 An initial value of the order is divided for the edge.
In one possible implementation manner, the constructing, for each scratch damage area, an edge run matrix based on the number of edge segmentation orders and the edge segmentation coefficients includes:
for each scratch damage area, dividing the edge segmentation coefficient of each edge pixel point in the scratch damage area into M in an average way (k) The number of edge segmentation orders, M (k) The number of the edge segmentation orders of the kth scratch damage region;
and constructing an edge run matrix according to the edge segmentation coefficient, wherein the number of lines of the edge run matrix is the number of the edge segmentation orders, the number of columns of the edge run matrix is the maximum value of the running length of the edge segmentation orders in a certain direction, and matrix elements EYA (i, j) of the edge run matrix represent the sum of the times of j edge segmentation orders i continuously appearing in all directions.
In one possible implementation manner, the calculating, based on the edge run matrix, the scratch damage degree and the scratch disorder degree of the scratch damage area includes:
and calculating the scratch damage degree of the scratch damage area based on the edge run matrix, wherein the calculation formula of the scratch damage degree is as follows:
wherein HSG (k) A scratch damage degree which is a kth scratch damage region; m is M (k) The number of rows of the edge run matrix which is the kth scratch damage area; n (N) (k) The number of columns of the edge run matrix that is the kth scratch damage region; EYA (i, j) represents the value of the ith row and jth column matrix elements of the edge run matrix;
and summing and normalizing matrix elements of the edge run matrix according to rows and columns respectively to obtain an edge segmentation order vector and an edge segmentation length vector, and calculating the scratch disorder degree of the scratch damage area based on the edge segmentation order vector and the edge segmentation length vector, wherein the scratch disorder degree has the following calculation formula:
Wherein, HWG (k) Representing the kth scratch damage regionScratch disorder degree, EDJV (i) EDLV for the ith element of the edge segmentation order vector (i) Dividing the ith element of the length vector for the edge, M (k) Dividing the length of the order vector EDJV for edges, N (k) The length of the length vector EDLV is segmented for edges.
In one possible implementation, determining the scratch category of the scratch damage region based on the scratch saliency, the scratch damage degree, and the scratch disorder degree of the scratch damage region includes:
the method comprises the steps that training classification is conducted on scratch damage areas by using a neural network, input neurons of the neural network comprise scratch saliency, scratch damage degree and scratch disorder degree of the scratch damage areas, and final output comprises four categories of light scratch, medium scratch, deep scratch and scratch of automobile scratch damage.
In one possible implementation manner, the evaluating the damage detection result of the automobile exterior part based on the scratch category and the scratch damage degree of the scratch damage area includes:
giving different category weights to the mild scratch, the moderate scratch, the deep scratch and the scratch respectively, normalizing the scratch damage degree to obtain damage degree weight, and obtaining a scratch damage value of the automobile based on the category weight and the damage degree weight, wherein the calculation formula of the scratch damage value is as follows:
Wherein HDV represents scratch damage value, wl of the automobile (k) Class weight for kth scratch damage region, ws (k) The damage degree weight of the kth scratch damage area is given, and N represents the number of the scratch damage areas in the automobile;
and determining the damage level of the automobile based on the scratch damage assessment value through a manual expert system, and giving an assessment opinion.
The beneficial effects of this application lie in: the machine vision and image scanning method can accurately detect and identify the characteristics of the saliency, the damage degree, the texture complexity and the like of scratches, different scratch areas are segmented, the scratch saliency, the scratch damage degree and the scratch disorder degree of the scratch damage areas are synthesized according to the differences of scratch damage areas of different scratch categories and paint layers and the changes of colors and shapes, the scratch damage areas of different scratches are classified into four categories of light scratches, medium scratches, deep scratches and scratch scratches, and the damage detection results of all appearance parts are subjected to fusion evaluation according to the scratch category and the scratch damage degree of each scratch damage area, so that the problem that the scratch damage areas cannot be classified and evaluated according to the current image segmentation technology is solved.
Drawings
Fig. 1 is a flowchart of steps of a method for detecting damage to an automobile based on machine vision and image scanning according to an embodiment of the present application.
Detailed Description
So that the manner in which the above recited objects, features and advantages of the present application can be understood in detail, a more particular description of the invention, briefly summarized above, may be had by reference to the appended drawings and detailed description, which follow. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present application are within the scope of the protection of the present application.
The terminology used in the description section of the present application is for the purpose of describing particular embodiments of the present application only and is not intended to be limiting of the present application.
It should be noted that references to "one" or "a plurality" in this application are intended to be illustrative rather than limiting, and those of ordinary skill in the art will appreciate that "one or more" is intended to be interpreted as "one or more" unless the context clearly indicates otherwise.
Embodiments of the present application are described below with reference to the accompanying drawings. As one of ordinary skill in the art can appreciate, with the development of technology and the appearance of new scenes, the technical solutions provided in the embodiments of the present application are applicable to similar technical problems.
Referring to fig. 1, an embodiment of the application discloses an automobile damage detection method based on machine vision and image scanning, which includes:
step S11, a first image of the automobile is obtained, the first image is preprocessed to obtain a second image, appearance component segmentation processing is carried out on the second image, and a third image of a plurality of automobile appearance components is obtained;
step S12, performing edge detection processing on the third image of each automobile appearance part to obtain edge images of a plurality of automobile appearance parts; clustering the edge images of each automobile appearance part to obtain a plurality of scratch damage areas;
step S13, performing edge segmentation calculation on each edge pixel point of each scratch damage area to obtain an edge segmentation coefficient of each edge pixel point; calculating the scratch saliency of each scratch damage area based on the edge segmentation coefficient of the edge pixel point;
step S14, calculating an edge segmentation floating index based on the edge segmentation coefficient of the edge pixel point for each scratch damage area, and calculating the number of edge segmentation orders of each scratch damage area based on the edge segmentation floating index;
Step S15, for each scratch damage area, constructing an edge run matrix based on the number of the edge segmentation orders and the edge segmentation coefficients, calculating the scratch damage degree and the scratch disorder degree of the scratch damage area based on the edge run matrix, and determining the scratch category of the scratch damage area based on the scratch saliency, the scratch damage degree and the scratch disorder degree of the scratch damage area;
and step S16, evaluating a damage detection result of the automobile appearance part based on the scratch category and the scratch damage degree of the scratch damage area.
In the steps of the embodiment, first, a first image of an automobile is acquired, the first image is preprocessed to obtain a second image, the omnidirectional image of the automobile is acquired and processed, appearance part segmentation processing is carried out on the second image to obtain a third image of a plurality of automobile appearance parts, appearance part segmentation is carried out on an automobile damage detection photo, and the position and the type of each appearance part are determined so as to further detect damage to each appearance part.
Then, in order to obtain a scratch damage area of each appearance part image, edge detection processing is performed on the third image of each automobile appearance part by using a Canny operator (an operator for edge detection can use a Prewitt operator, a Sobel operator, a Kirsch operator or the like besides the Canny operator, and is not particularly limited here), so as to obtain a binary image, wherein a pixel point with a median value of 1 in the binary image is an edge pixel point, and thus edge images of a plurality of automobile appearance parts are obtained; and clustering the edge images of each automobile appearance part to obtain a plurality of scratch damage areas. The automobile paint surface is mainly provided with three layers, which are sequentially from inside to outside: primer layer, colored paint layer and clear paint layer. The scratch damage is classified into a slight scratch, a moderate scratch, a deep scratch and a scratch according to the damage to the paint surface. When the scratch is generated, the primer layer generally touching the surface of the automobile is mainly composed of a plurality of linear concave-convex traces, and thus the scratch area may be composed of a plurality of edges. In order to divide and identify each scratch among scratch scratches as a whole, a DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering algorithm is used to cluster the closer edges into one cluster, namely a scratch damage area, the maximum radius Epsilon of the cluster is set to be 10, and the minimum point minPts is set to be 1. The clustering algorithm may use K-MEANS, K-MEDOIDS, clara, clarans, etc. besides DBSCAN, and is not particularly limited herein; the various parameters of the clustering algorithm can be flexibly adjusted according to the actual situation, and the method is not particularly limited herein.
Then, after each scratch damage area is obtained according to the clustering algorithm, the scratch saliency of each scratch damage area is different due to different scratch types, and the scratch saliency can be obtained according to the difference degree of three channel values of the upper edge pixel point of each edge and the surrounding pixel point R, G, B, and the saliency is measured. Therefore, edge segmentation calculation is carried out on each edge pixel point of each scratch damage area, and an edge segmentation coefficient of each edge pixel point is obtained; and calculating the scratch saliency of each scratch damage area based on the edge segmentation coefficient of the edge pixel point.
Then, considering that the damage degree and the texture complexity of each scratch damage area are different due to different scratch types; therefore, for each of the scratch damage regions, an edge division floating index is calculated based on the edge division coefficient of the edge pixel point, and the number of edge division orders of each of the scratch damage regions is calculated based on the edge division floating index.
Then, considering that the scratch damage degree is different according to the different scratch damage grades, the scratch damage degree can be used as a judging standard for classifying the scratch damage areas, compared with light scratch, medium scratch and deep scratch, the scratch texture is complex, and the disorder degree of the scratch can be used as a judging standard for classifying the scratch damage areas; therefore, for each of the scratch damage areas, an edge run matrix is constructed based on the number of the edge division orders and the edge division coefficient, the scratch damage degree and the scratch disorder degree of the scratch damage area are calculated based on the edge run matrix, and the scratch category of the scratch damage area is determined based on the scratch saliency, the scratch damage degree and the scratch disorder degree of the scratch damage area.
And finally, carrying out fusion evaluation on damage detection results of all the appearance parts according to scratch categories and scratch damage degrees of all the scratch damage areas, and evaluating the damage detection results of the automobile appearance parts based on the scratch categories and the scratch damage degrees of the scratch damage areas.
In an optional embodiment of the present application, the acquiring a first image of the automobile, and preprocessing the first image to obtain a second image, includes:
image acquisition is carried out on the automobile by utilizing an image acquisition device at a specific angle to obtain a plurality of automobile projection images, wherein the automobile projection images are first images of the automobile;
calculating a two-dimensional reconstruction image serving as an image to be registered according to a plurality of first images, and calculating a conversion relation between the image to be registered and a reference image based on characteristic point information of the image to be registered and pre-extracted reference image information;
and performing geometric transformation on the image to be registered to a reference image based on the conversion relation to obtain a second image of the automobile.
In the steps of the embodiment, an image acquisition device (for example, mixed Reality (MR) glasses and a CT scanning frame) are utilized to acquire images of the automobile at a specific angle (for example, every 0.1 degree of rotation), so as to obtain a plurality of automobile projection images (for example, 3600 automobile projection images), wherein the automobile projection images are first images of the automobile; according to the method, a two-dimensional reconstructed image is obtained through calculation according to a plurality of first images to serve as an image to be registered, the omnibearing 3D scanning of the automobile is achieved, the conversion relation between the image to be registered and a reference image is obtained through calculation based on feature point information of the image to be registered and pre-extracted reference image information, the image to be registered is subjected to geometric transformation to the reference image based on the conversion relation, and a second image of the automobile is obtained, so that the geometric transformation from the image to be registered to the reference image is achieved. The preprocessing of the first image in the steps of the embodiment enables the specific object in the image to be registered, which is acquired in real time, to be kept in the position of the reference image, thereby realizing accurate registration of virtual and real scenes and reducing the problems of position deviation, image scale distortion and the like caused by movement of the image acquisition equipment.
In an optional embodiment of the present application, performing edge segmentation calculation on each edge pixel point of each scratch damage area to obtain an edge segmentation coefficient of each edge pixel point includes:
constructing a neighborhood window for each edge pixel point of each scratch damage area, and calculating to obtain an edge segmentation coefficient of each edge pixel point, wherein the calculation formula of the edge segmentation coefficient is as follows:
wherein EDX (x) is an edge segmentation coefficient of the edge pixel point x,the average value of R channel values of edge pixel points and non-edge pixel points in the neighborhood window of the edge pixel point x is respectively; />The average value of the G channel values of the edge pixel points and the non-edge pixel points in the neighborhood window of the edge pixel point x is respectively;and the average values of the B channel values of the edge pixel points and the non-edge pixel points in the neighborhood window of the edge pixel point x are respectively.
No matter what kind of damage is generated on the surface of the automobile, most of the damage damages the colored paint layer, the colored paint layer is colored paint, and the visual display of the five colors and six colors is determined by the colored paint layer and is positioned in the middle layer of the paint surface. When the scratch is damaged into the paint layer, the values of the R, G, B three channels of the scratch area and the surrounding lossless area are different. In the steps of the embodiment, a neighborhood window is constructed for each edge pixel point of each scratch damage area, and an edge segmentation coefficient of each edge pixel point is calculated. It should be noted that, when the degree of dissimilarity between the three channel values of the edge pixel point and the non-edge pixel point R, G, B in the neighborhood window of the pixel point x is higher, the edge division coefficient of the pixel point x is larger, and the edge division coefficient of the non-edge pixel point is 0.
In an optional embodiment of the present application, the calculating, based on the edge segmentation coefficient of the edge pixel point, a scratch saliency of each scratch damage area includes:
calculating the scratch saliency of each scratch damage area according to the edge segmentation coefficient of the edge pixel point in each scratch damage area, wherein the calculation formula of the scratch saliency is as follows:
wherein HHO (k) For scratch saliency of the kth scratch damaged region, EDX (x i ) For the edge segmentation coefficient of the ith edge pixel point in the scratch damage area, n (k) Is the number of edge pixels in the kth scratch damage region.
Since the scratches are usually linear, the damage degree and texture complexity of the scratches are closely related to the edge segmentation coefficients of the edge pixels, and after each damaged area is obtained, the sum of the edge segmentation coefficients of each pixel in each damaged area can be approximated to the scratch significance. In the foregoing embodiment, the scratch saliency of each scratch damage region is calculated according to the edge segmentation coefficient of the edge pixel point in each scratch damage region. When the edge division coefficient of each edge pixel point is larger, the degree of dissimilarity between the edge pixel point and the surrounding area is higher, and therefore the scratch saliency of the scratch damage area is larger.
In an optional embodiment of the present application, the calculating, for each of the scratch damaged areas, an edge segmentation floating index based on the edge segmentation coefficient of the edge pixel point includes:
and calculating the edge segmentation floating index of each scratch damage area based on the edge segmentation coefficient of the edge pixel point, wherein the calculation formula of the edge segmentation floating index is as follows:
therein, EFZ (k) The floating index is divided for the edge of the kth scratch damage region,and e is a natural constant, which is the variance of the edge segmentation coefficient of each edge pixel point in the kth scratch damage region.
The scratches are often linear, so the degree of damage and texture complexity of the scratches depend on the edge segmentation factor of the edge pixels and their extension distribution. For each scratch damage area, according to the edge segmentation coefficient of each edge pixel point in the scratch damage area, an edge run-length matrix is constructed, so that the scratch damage degree and scratch disorder degree of different scratch damage areas can be calculated, and the damage degree and texture complexity of the scratch damage areas are measured. In order to facilitate calculation and reduce the calculation amount, the edge segmentation coefficient needs to be divided into a plurality of edge segmentation orders, and the number of the edge segmentation orders is determined by the range of the edge segmentation coefficient of each scratch damage area. When scratch scratches are generated, the difference of the edge segmentation coefficients of all the edge pixel points is larger due to different scratch degrees, and more edge segmentation orders are needed to be divided to measure the damage condition of the scratch damage area; for light, moderate and deep scratches, the damage degree of the same edge at different positions is consistent, the difference of edge segmentation coefficients of pixel points at each edge is smaller, and the damage condition of the scratch damage area can be measured by dividing fewer edge segmentation orders.
In the above embodiment, the edge division floating index of the scratch damage region is calculated for each of the scratch damage regions based on the edge division coefficient of the edge pixel point. When the variance of the edge segmentation coefficient of each edge pixel point is larger, the segmentation degree of the edge pixel point is different, and the calculated edge segmentation floating index of the scratch damage area is larger; in order to better measure the damage condition of the scratch damage area, the number of edge segmentation orders should be larger.
In an optional embodiment of the present application, the calculating, based on the edge segmentation floating index, the number of edge segmentation orders of each scratch damage region includes:
calculating the number of edge segmentation orders of each scratch damage area based on the edge segmentation floating index, wherein the calculation formula of the number of edge segmentation orders is as follows:
M (k) ＝round(EFZ (k) *M 0 )
wherein M is (k) The number of the edge segmentation orders of the kth scratch damage area; round () is a rounding function; EFZ (k) Splitting a floating index, M, for the edge of the kth scratch-damaged region 0 An initial value of the order is divided for the edge.
In the above embodiment, the number of edge segmentation orders of each scratch damage region is calculated based on the edge segmentation floating index. The rounding function is used for ensuring that the number of the edge segmentation orders is an integer, the edge segmentation floating index of the scratch damage area is inversely proportional to the consistency of the segmentation degree of each edge pixel point of the scratch damage area, and when the edge segmentation floating index is larger, the segmentation degree of each edge pixel point of the scratch damage area is different, and the number of the edge segmentation orders is required to be larger; initial value M of edge segmentation order 0 The value 10 may be empirically taken.
In an optional embodiment of the present application, the constructing, for each of the scratch damaged areas, an edge run matrix based on the number of edge segmentation orders and the edge segmentation coefficients includes:
for each scratch damage area, dividing the edge segmentation coefficient of each edge pixel point in the scratch damage area into M in an average way (k) The number of edge segmentation orders, M (k) The number of the edge segmentation orders of the kth scratch damage region;
and constructing an edge run matrix according to the edge segmentation coefficient, wherein the number of lines of the edge run matrix is the number of the edge segmentation orders, the number of columns of the edge run matrix is the maximum value of the running length of the edge segmentation orders in a certain direction, and matrix elements EYA (i, j) of the edge run matrix represent the sum of the times of j edge segmentation orders i continuously appearing in all directions.
In the above embodiment, for each scratch damage region, the edge division coefficient of each edge pixel point in the scratch damage region is divided into M (k) The edge segmentation orders are used for constructing an edge run matrix according to the edge segmentation coefficients, obtaining the occurrence times of scratches with different extension lengths and depths, wherein the size of the edge run matrix is M (k) ×N (k) The number of rows of the matrix is the number of edge segmentation orders, and the number of columns is the maximum value of the running length of the edge segmentation orders in a certain direction. In order to reduce the calculation amount while taking into consideration the connected information in all directions as much as possible, the edge run matrix may be calculated every 45 degrees in actual calculation. The edge run matrix element EYA (i, j) represents the sum of the number of times of consecutively appearing j edge division orders i in each direction, corresponding to the number of times of occurrence of scratches of the same extension length, depth.
In an optional embodiment of the present application, the calculating, based on the edge run matrix, the scratch damage degree and the scratch disorder degree of the scratch damage area includes:
and calculating the scratch damage degree of the scratch damage area based on the edge run matrix, wherein the calculation formula of the scratch damage degree is as follows:
wherein HSG (k) A scratch damage degree which is a kth scratch damage region; m is M (k) The number of rows of the edge run matrix which is the kth scratch damage area; n (N) (k) The number of columns of the edge run matrix that is the kth scratch damage region; EYA (i, j) represents the value of the ith row and jth column matrix elements of the edge run matrix;
and summing and normalizing matrix elements of the edge run matrix according to rows and columns respectively to obtain an edge segmentation order vector and an edge segmentation length vector, and calculating the scratch disorder degree of the scratch damage area based on the edge segmentation order vector and the edge segmentation length vector, wherein the scratch disorder degree has the following calculation formula:
Wherein, HWG (k) Indicating the scratch disorder degree of the kth scratch damage region, EDJV (i) EDLV for the ith element of the edge segmentation order vector (i) Dividing the ith element of the length vector for the edge, M (k) Dividing the length of the order vector EDJV for edges, N (k) The length of the length vector EDLV is segmented for edges.
In the foregoing embodiment, the scratch damage degree of the scratch damage area is calculated based on the edge run matrix. In the calculation formula of the scratch damage degree, i is the edge segmentation order, the depth of the scratch is represented, and the larger the order is, the higher the degree of dissimilarity between the edge and the surrounding area is, the deeper the scratch is, the higher the scratch damage degree is, and the damage level is higher; j is the number of continuous occurrences of the same edge segmentation order, and represents the length of the scratch, the more the number is, the longer the edge is, the longer the scratch is, the higher the scratch damage degree is, and the damage level is higher; EYA (i, j) is the sum of the number of times of j edge dividing orders i continuously appearing in each direction, and represents the occurrence number of scratches of the same extension length and depth, and the higher the occurrence number of scratches is, the higher the scratch damage degree is and the damage level is.
Then, according to the distribution of the edge segmentation order and the edge length in the kth scratch damage area, the matrix elements of the edge run matrix are summed and normalized according to rows and columns respectively to obtain a length M (k) Edge segmentation order vector EDJV and length N (k) Edge segmentation length vector EDLV, and calculating scratch disorder degree HWG of the kth scratch damage area based on the edge segmentation order vector EDJV and the edge segmentation length vector EDLV (k) . In the calculation formula of scratch disorder, EDJV (i) The i-th element of the edge segmentation order vector represents the frequency of occurrence of the pixel point with the edge segmentation order of i, when the edge segmentation order is single, the frequency of occurrence of one edge segmentation order is closer to 1, the rest edge segmentation orders are closer to 0,the larger the value, the gradually approaching to 1, and the scratch disorder degree HWG (k) The lower the less likely scratch scratches; EDLV (i) For the ith element of the edge-divided length vector, similarly when the edge-divided length is more single,the greater the value, the scratch disorder HWG (k) The lower the less likely a scratch.
In an optional embodiment of the present application, determining the scratch category of the scratch damage region based on the scratch saliency, the scratch damage degree, and the scratch disorder degree of the scratch damage region includes:
the method comprises the steps that training classification is conducted on scratch damage areas by using a neural network, input neurons of the neural network comprise scratch saliency, scratch damage degree and scratch disorder degree of the scratch damage areas, and final output comprises four categories of light scratch, medium scratch, deep scratch and scratch of automobile scratch damage.
In the steps of the embodiment, according to the scratch damage of the automobile, the difference of the damaged paint layer and the change of the color and the shape, the scratch saliency, the scratch damage degree and the scratch disorder degree are synthesized, the scratch category of the scratch damage area is determined by adopting a neural network, and the scratch damage area is divided into a slight scratch, a moderate scratch, a deep scratch and a scratch. Training and classifying the scratch damage area by using a fully-connected neural network, wherein the fully-connected neural network comprises three input neurons, the three input neurons are the scratch significance, the scratch damage and the scratch disorder of the scratch damage area, a hidden layer of the fully-connected neural network can use a ReLU activation function, the final output can be a fully-connected layer, the number of the neurons of the fully-connected layer can be 4, and the four categories of mild scratch, medium scratch, deep scratch and scratch of automobile scratch damage are finally output; the output layer may use a Softmax function (also known as a normalized exponential function) for classifying the scratch categories of the scratch damaged areas. A full-Layer permission (MLP) is an artificial neural network structure with a simpler connection mode, belongs to one type of feedforward neural network, and mainly comprises an input Layer, a hidden Layer and an output Layer, and a plurality of neurons can be arranged in each hidden Layer; in addition to the fully connected neural network, the neural network may also be a feedforward neural network, a Convolutional Neural Network (CNN), a recurrent neural network, a Transformers, a generation countermeasure network (GAN), or the like, which is not particularly limited herein. The ReLU full name Rectified Linear Unit, the ReLU activation function is a common nerve activation function, and the activation function of the hidden layer may use a Sigmoid function, a Tanh function, or the like, besides the ReLU, and is not particularly limited herein.
In an optional embodiment of the present application, the evaluating the damage detection result of the automotive exterior part based on the scratch category and the scratch damage degree of the scratch damage area includes:
giving different category weights to the mild scratch, the moderate scratch, the deep scratch and the scratch respectively, normalizing the scratch damage degree to obtain damage degree weight, and obtaining a scratch damage value of the automobile based on the category weight and the damage degree weight, wherein the calculation formula of the scratch damage value is as follows:
wherein HDV represents scratch damage value, wl of the automobile (k) Class weight for kth scratch damage region, ws (k) For the kth scratch damaged regionA damage degree weight, N represents the number of scratch damage areas in the automobile;
and determining the damage level of the automobile based on the scratch damage assessment value through a manual expert system, and giving an assessment opinion.
In the above-described example steps, the damage detection results of all the exterior parts were subjected to fusion evaluation according to the scratch categories and scratch damage degrees of the respective scratch damage regions, and different category weights wl were respectively given to the light scratch, the medium scratch, the deep scratch, and the scratch (k) The empirical values may be 0.1, 0.2, 0.3, 0.4 (other values may be selected according to the actual situation, and are not specifically limited herein), and normalizing the scratch damage degree corresponds to the damage degree weights ws thereof (k) Based on the category weights wl (k) And the injury-degree weight ws (k) Obtaining a scratch damage value HDV of the automobile; the scratch damage assessment value of the automobile corresponds to the damage degree of the automobile, and the damage degree of the automobile is higher as the scratch damage assessment value is larger, so that the damage level of the automobile is determined based on the scratch damage assessment value through a manual expert system, and evaluation comments are given, such as repair suggestions, quality assurance and the like. Note that, in the calculation formula of the scratch damage value, when the category weight wl (k) The larger the scratch type of the damaged area is, the more serious the influence is caused, and the larger the scratch damage value is; when the injury degree weight ws (k) The larger the scratch damage degree of the region, the larger the scratch damage value.
In this specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, so that the same or similar parts between the embodiments are referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. The software modules may be disposed in Random Access Memory (RAM), memory, read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
The foregoing has described in detail a method for detecting damage to an automobile based on machine vision and image scanning, and specific examples are applied herein to illustrate the principles and embodiments of the present application, and the above description of the examples is only for aiding in understanding the method and core ideas of the present application; meanwhile, as those skilled in the art will vary in the specific embodiments and application scope according to the ideas of the present application, the contents of the present specification should not be construed as limiting the present application in summary.
Claims (8)
1. The automobile damage detection method based on machine vision and image scanning is characterized by comprising the following steps of:
acquiring a first image of the automobile, preprocessing the first image to obtain a second image, and performing appearance part segmentation processing on the second image to obtain a third image of a plurality of automobile appearance parts;
Performing edge detection processing on the third image of each automobile appearance part to obtain edge images of a plurality of automobile appearance parts; clustering the edge images of each automobile appearance part to obtain a plurality of scratch damage areas;
performing edge segmentation calculation on each edge pixel point of each scratch damage area to obtain an edge segmentation coefficient of each edge pixel point; calculating the scratch saliency of each scratch damage area based on the edge segmentation coefficient of the edge pixel point;
calculating the edge segmentation floating index based on the edge segmentation coefficients of the edge pixel points for each scratch damage area, and calculating the number of edge segmentation orders of each scratch damage area based on the edge segmentation floating index;
for each scratch damage area, constructing an edge run matrix based on the number of the edge segmentation orders and the edge segmentation coefficients, calculating to obtain the scratch damage degree and the scratch disorder degree of the scratch damage area based on the edge run matrix, and determining the scratch category of the scratch damage area based on the scratch saliency, the scratch damage degree and the scratch disorder degree of the scratch damage area;
Evaluating a damage detection result of the automobile exterior part based on the scratch category and the scratch damage degree of the scratch damage area;
the calculating the scratch damage degree and the scratch disorder degree of the scratch damage area based on the edge run matrix comprises the following steps:
and calculating the scratch damage degree of the scratch damage area based on the edge run matrix, wherein the calculation formula of the scratch damage degree is as follows:
wherein HSG (k) A scratch damage degree which is a kth scratch damage region; m is M (k) The number of rows of the edge run matrix which is the kth scratch damage area; n (N) (k) The number of columns of the edge run matrix that is the kth scratch damage region; EYA (i, j) represents the value of the ith row and jth column matrix elements of the edge run matrix;
and summing and normalizing matrix elements of the edge run matrix according to rows and columns respectively to obtain an edge segmentation order vector and an edge segmentation length vector, and calculating the scratch disorder degree of the scratch damage area based on the edge segmentation order vector and the edge segmentation length vector, wherein the scratch disorder degree has the following calculation formula:
wherein, HWG (k) Indicating the scratch disorder degree of the kth scratch damage region, EDJV (i) EDLV for the ith element of the edge segmentation order vector (i) Dividing the ith element of the length vector for the edge, M (k) Dividing the length of the order vector EDJV for edges, N (k) Dividing the length of the length vector EDLV for edges;
the evaluating the damage detection result of the automobile exterior part based on the scratch category and the scratch damage degree of the scratch damage area includes:
giving different category weights to the mild scratch, the moderate scratch, the deep scratch and the scratch respectively, normalizing the scratch damage degree to obtain damage degree weight, and obtaining a scratch damage value of the automobile based on the category weight and the damage degree weight, wherein the calculation formula of the scratch damage value is as follows:
wherein HDV represents scratch damage value, wl of the automobile (k) Class weight for kth scratch damage region, ws (k) The damage degree weight of the kth scratch damage area is given, and N represents the number of the scratch damage areas in the automobile;
and determining the damage level of the automobile based on the scratch damage assessment value through a manual expert system, and giving an assessment opinion.
2. The method for detecting damage to an automobile based on machine vision and image scanning of claim 1, wherein the acquiring a first image of the automobile, preprocessing the first image to obtain a second image, comprises:
Image acquisition is carried out on the automobile by utilizing an image acquisition device at a specific angle to obtain a plurality of automobile projection images, wherein the automobile projection images are first images of the automobile;
calculating a two-dimensional reconstruction image serving as an image to be registered according to a plurality of first images, and calculating a conversion relation between the image to be registered and a reference image based on characteristic point information of the image to be registered and pre-extracted reference image information;
and performing geometric transformation on the image to be registered to a reference image based on the conversion relation to obtain a second image of the automobile.
3. The method for detecting the damage to the automobile based on the machine vision and the image scanning according to claim 1, wherein the performing an edge segmentation calculation on each edge pixel point of each scratch damage area to obtain an edge segmentation coefficient of each edge pixel point comprises:
constructing a neighborhood window for each edge pixel point of each scratch damage area, and calculating to obtain an edge segmentation coefficient of each edge pixel point, wherein the calculation formula of the edge segmentation coefficient is as follows:
wherein EDX (x) is an edge segmentation coefficient of the edge pixel point x, The average value of R channel values of edge pixel points and non-edge pixel points in the neighborhood window of the edge pixel point x is respectively; />The average value of the G channel values of the edge pixel points and the non-edge pixel points in the neighborhood window of the edge pixel point x is respectively;and the average values of the B channel values of the edge pixel points and the non-edge pixel points in the neighborhood window of the edge pixel point x are respectively.
4. The method for detecting the damage to the automobile based on the machine vision and the image scanning according to claim 1, wherein the calculating the scratch saliency of each scratch damage area based on the edge segmentation coefficient of the edge pixel point comprises:
calculating the scratch saliency of each scratch damage area according to the edge segmentation coefficient of the edge pixel point in each scratch damage area, wherein the calculation formula of the scratch saliency is as follows:
wherein HHO (k) For scratch saliency of the kth scratch damaged region, EDX (x i ) For the edge segmentation coefficient of the ith edge pixel point in the scratch damage area, n (k) Is the number of edge pixels in the kth scratch damage region.
5. The method for detecting an automobile injury based on machine vision and image scanning according to claim 1, wherein said calculating an edge segmentation floating index for each of said scratch injury regions based on said edge segmentation coefficients of said edge pixels comprises:
And calculating the edge segmentation floating index of each scratch damage area based on the edge segmentation coefficient of the edge pixel point, wherein the calculation formula of the edge segmentation floating index is as follows:
therein, EFZ (k) The floating index is divided for the edge of the kth scratch damage region,for the kth divisionAnd the variance of the edge segmentation coefficient of each edge pixel point in the trace damage area, wherein e is a natural constant.
6. The method for detecting the damage to the automobile based on the machine vision and the image scanning according to claim 1, wherein the calculating the number of the edge segmentation orders of each scratch damage region based on the edge segmentation floating index comprises:
calculating the number of edge segmentation orders of each scratch damage area based on the edge segmentation floating index, wherein the calculation formula of the number of edge segmentation orders is as follows:
M (k) ＝round(EFZ (k) *M 0 )
wherein M is (k) The number of the edge segmentation orders of the kth scratch damage area; round () is a rounding function; EFZ (k) Splitting a floating index, M, for the edge of the kth scratch-damaged region 0 An initial value of the order is divided for the edge.
7. The method for detecting damage to an automobile based on machine vision and image scanning of claim 1, wherein the constructing an edge run matrix for each of the scratch damaged areas based on the number of edge segmentation orders and the edge segmentation coefficients comprises:
For each scratch damage area, dividing the edge segmentation coefficient of each edge pixel point in the scratch damage area into M in an average way (k) The number of edge segmentation orders, M (k) The number of the edge segmentation orders of the kth scratch damage region;
and constructing an edge run matrix according to the edge segmentation coefficient, wherein the number of lines of the edge run matrix is the number of the edge segmentation orders, the number of columns of the edge run matrix is the maximum value of the running length of the edge segmentation orders in a certain direction, and matrix elements EYA (i, j) of the edge run matrix represent the sum of the times of j edge segmentation orders i continuously appearing in all directions.
8. The machine vision and image scanning based automotive damage detection method of claim 1, wherein determining a scratch category of the scratch damaged area based on the scratch saliency, the scratch damage degree, and the scratch disorder degree of the scratch damaged area comprises:
the method comprises the steps that training classification is conducted on scratch damage areas by using a neural network, input neurons of the neural network comprise scratch saliency, scratch damage degree and scratch disorder degree of the scratch damage areas, and final output comprises four categories of light scratch, medium scratch, deep scratch and scratch of automobile scratch damage.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310966064.6A CN116883444B (en) | 2023-08-02 | 2023-08-02 | Automobile damage detection method based on machine vision and image scanning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310966064.6A CN116883444B (en) | 2023-08-02 | 2023-08-02 | Automobile damage detection method based on machine vision and image scanning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116883444A CN116883444A (en) | 2023-10-13 |
| CN116883444B true CN116883444B (en) | 2024-01-12 |
Family
ID=88266257
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310966064.6A Active CN116883444B (en) | 2023-08-02 | 2023-08-02 | Automobile damage detection method based on machine vision and image scanning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116883444B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106127747A (en) * | 2016-06-17 | 2016-11-16 | 史方 | Car surface damage classifying method and device based on degree of depth study |
| CN112098419A (en) * | 2020-09-11 | 2020-12-18 | 江苏理工学院 | System and method for detecting surface defects of automobile outer covering part |
| CN113096085A (en) * | 2021-04-01 | 2021-07-09 | 武汉理工大学 | Container surface damage detection method based on two-stage convolutional neural network |
| WO2023040142A1 (en) * | 2021-09-15 | 2023-03-23 | 平安科技（深圳）有限公司 | Vehicle damage detection method and apparatus, and device and storage medium |
| CN116519699A (en) * | 2023-02-23 | 2023-08-01 | 中国民用航空飞行学院 | Structural damage detection system and structural damage detection method |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115269629A (en) * | 2022-06-30 | 2022-11-01 | 启明信息技术股份有限公司 | Data query method and system supporting multiple data sources |
| CN115374116A (en) * | 2022-08-22 | 2022-11-22 | 中国铁建重工集团股份有限公司 | High-allocable multi-data-source instant heterogeneous fusion method and system |
| CN115905740A (en) * | 2022-12-15 | 2023-04-04 | 中电万维信息技术有限责任公司 | Multi-data source service engine interface interconnection method |
-
2023
- 2023-08-02 CN CN202310966064.6A patent/CN116883444B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106127747A (en) * | 2016-06-17 | 2016-11-16 | 史方 | Car surface damage classifying method and device based on degree of depth study |
| CN112098419A (en) * | 2020-09-11 | 2020-12-18 | 江苏理工学院 | System and method for detecting surface defects of automobile outer covering part |
| CN113096085A (en) * | 2021-04-01 | 2021-07-09 | 武汉理工大学 | Container surface damage detection method based on two-stage convolutional neural network |
| WO2023040142A1 (en) * | 2021-09-15 | 2023-03-23 | 平安科技（深圳）有限公司 | Vehicle damage detection method and apparatus, and device and storage medium |
| CN116519699A (en) * | 2023-02-23 | 2023-08-01 | 中国民用航空飞行学院 | Structural damage detection system and structural damage detection method |
Non-Patent Citations (1)
| Title |
|---|
| 基于3D成像和激光轮廓传感技术的共享汽车非事故损伤检测系统设计与应用;顾赫，李泓宣，杨佳男，王怡，冯美宁;《机电信息》(第28期);4 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116883444A (en) | 2023-10-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110738697B (en) | Monocular depth estimation method based on deep learning | |
| Yin et al. | Hot region selection based on selective search and modified fuzzy C-means in remote sensing images | |
| CN109596634B (en) | Cable defect detection method and device, storage medium and processor | |
| CN103593670B (en) | A kind of copper plate/strip detection method of surface flaw based on online limit of sequence learning machine | |
| Yu et al. | A new edge detection approach based on image context analysis | |
| CN111611874B (en) | Face mask wearing detection method based on ResNet and Canny | |
| CN110286124A (en) | Refractory brick measuring system based on machine vision | |
| CN109034184B (en) | Grading ring detection and identification method based on deep learning | |
| CN111507426A (en) | No-reference image quality grading evaluation method and device based on visual fusion characteristics | |
| CN113505865B (en) | Sheet surface defect image recognition processing method based on convolutional neural network | |
| CN114742799B (en) | Industrial scene unknown type defect segmentation method based on self-supervision heterogeneous network | |
| CN113221881B (en) | Multi-level smart phone screen defect detection method | |
| CN115909256B (en) | Road disease detection method based on road visual image | |
| CN116542982A (en) | Departure judgment device defect detection method and device based on machine vision | |
| CN116883408B (en) | Integrating instrument shell defect detection method based on artificial intelligence | |
| CN110348307B (en) | Path edge identification method and system for crane metal structure climbing robot | |
| CN116279592A (en) | Method for dividing travelable area of unmanned logistics vehicle | |
| CN115937518A (en) | Pavement disease identification method and system based on multi-source image fusion | |
| CN116934725A (en) | Method for detecting sealing performance of aluminum foil seal based on unsupervised learning | |
| CN111814852A (en) | Image detection method, image detection device, electronic equipment and computer-readable storage medium | |
| CN115034997A (en) | Image processing method and device | |
| CN116721096B (en) | New energy harness quality online detection method based on artificial intelligence | |
| CN116883444B (en) | Automobile damage detection method based on machine vision and image scanning | |
| CN116542963A (en) | Float glass defect detection system and detection method based on machine learning | |
| CN116342519A (en) | Image processing method based on machine learning |
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 |