CN116740383A - Vehicle paint formula accurate identification method and system based on image identification - Google Patents
Vehicle paint formula accurate identification method and system based on image identification Download PDFInfo
- Publication number
- CN116740383A CN116740383A CN202310731844.2A CN202310731844A CN116740383A CN 116740383 A CN116740383 A CN 116740383A CN 202310731844 A CN202310731844 A CN 202310731844A CN 116740383 A CN116740383 A CN 116740383A
- Authority
- CN
- China
- Prior art keywords
- vehicle body
- color number
- color
- sub
- image data
- 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
- 239000003973 paint Substances 0.000 title claims abstract description 133
- 238000000034 method Methods 0.000 title claims abstract description 35
- 239000013598 vector Substances 0.000 claims abstract description 224
- 239000000203 mixture Substances 0.000 claims description 38
- 238000009472 formulation Methods 0.000 claims description 29
- 238000001914 filtration Methods 0.000 claims description 20
- 239000004595 color masterbatch Substances 0.000 claims description 6
- 230000011218 segmentation Effects 0.000 claims description 5
- 239000011159 matrix material Substances 0.000 claims description 4
- 241000251511 Holothuroidea Species 0.000 claims description 3
- 239000000654 additive Substances 0.000 claims description 3
- 230000000996 additive effect Effects 0.000 claims description 3
- 238000005315 distribution function Methods 0.000 claims description 3
- 238000009499 grossing Methods 0.000 claims description 3
- 239000004922 lacquer Substances 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 3
- 230000008569 process Effects 0.000 description 10
- 230000006870 function Effects 0.000 description 7
- 230000008439 repair process Effects 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 239000003086 colorant Substances 0.000 description 4
- 230000004044 response Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 238000005299 abrasion Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000012887 quadratic function Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- 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
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
-
- 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)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (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 discloses a vehicle paint formula accurate identification method and system based on image identification, comprising the following steps: according to the SIFT feature vector of the vehicle body, the HSI feature vector of the vehicle body, the SIFT feature vector of the color number and the HSI feature vector of the color number, the matching degree of each part of sub-vehicle body image data and each color number in the color chart is calculated in sequence, the color number corresponding to the maximum value of the matching degree is used as the target color number of the sub-vehicle body image data when the maximum value of the matching degree is larger than or equal to the first point number threshold value, and the vehicle paint formula of the sub-vehicle body image data is identified according to the target color number. By adopting the method, the corresponding feature vector is obtained by identifying the vehicle paint on the surface of the vehicle by utilizing the image identification technology, the target color number and the corresponding vehicle paint formula are selected according to the matching degree after the matching degree is calculated according to the feature vector, the corresponding vehicle paint formula is accurately identified in the database, and the accuracy and the efficiency of vehicle paint formula identification are ensured.
Description
Technical Field
The application relates to the technical field of image data processing, in particular to a vehicle paint formula accurate identification method and system based on image identification.
Background
With the dramatic increase in automobile production, the use of automotive surface paints has increased. On the one hand, after automobile manufacturers finish producing automobiles, the surfaces of the automobiles need to be treated by using paint with specific colors, on the other hand, as automobiles are increased continuously, traffic accidents are increased suddenly, the damage to the surfaces of the automobiles is unavoidable, and a large number of automobile repair factories need to treat the automobiles by using the paint with the specific colors.
At present, when repairing the surface of an automobile, if a user selects to repair the surface of the automobile at each production factory, the production factory only can repaint the paint according to the paint formulation adopted when the automobile leaves the factory, the influence of the using time of the automobile on the automobile paint is not considered, the original formulation is continuously adopted for refreshing, the color and the tone of each area of the automobile are inconsistent, and if the user selects to repair the surface of the automobile, the repair factory generally adopts a painter with a certain experience technology to manually prepare the required color, but the manual experience of a paint repairing master is completely relied on, and color difference of the automobile often occurs after the paint repairing.
Disclosure of Invention
The embodiment of the application provides a vehicle paint formula accurate identification method and system based on image identification, which are used for identifying the change of vehicle paint on the surface of a vehicle by utilizing an image identification technology, accurately identifying the corresponding vehicle paint formula in a database and ensuring the accuracy and efficiency of vehicle paint formula selection.
The first aspect of the embodiment of the application provides a vehicle paint formula accurate identification method based on image identification, which comprises the following steps:
acquiring a factory paint formula and a factory color number according to the vehicle type information; the factory color number refers to the color number of the factory vehicle paint formula corresponding to the color in a color chart, and each color number corresponds to one color and one vehicle paint formula in the color chart;
acquiring vehicle body image data and dividing the vehicle body image data into different sub-vehicle body image data;
carrying out Gaussian filtering of two different preset parameters on each part of sub-vehicle body image data, and obtaining a corresponding sub-vehicle body SIFT feature map after subtracting the two filtering results;
in each sub-vehicle body SIFT feature map, SIFT feature vectors and HSI feature vectors of each extreme point are calculated and used as vehicle body SIFT feature vectors and vehicle body HSI feature vectors of corresponding sub-vehicle body image data;
for each color number in the color card table, randomly selecting one pixel point in a color number area, and sequentially calculating SIFT feature vectors and HSI feature vectors as color number SIFT feature vectors and color number HSI feature vectors;
and according to the SIFT feature vector of the automobile body, the HSI feature vector of the automobile body, the SIFT feature vector of the color number and the HSI feature vector of the color number, calculating the matching degree of each part of sub-automobile body image data and each color number in a color card table in sequence, taking the color number corresponding to the maximum value of the matching degree which is larger than or equal to a first point number threshold value as a target color number of the sub-automobile body image data, and identifying the automobile paint formula of the sub-automobile body image data according to the target color number.
In a possible implementation manner of the first aspect, the acquiring the vehicle body image data and dividing the vehicle body image data into different sub-vehicle body image data specifically is:
obtaining a main view image, a first side view image, a second side view image, a rear view image and a top view image according to vehicle body image data, and taking vehicle edge pixel points in the main view image, the first side view image, the second side view image and the rear view image as vehicle contour lines;
and dividing the vehicle body image data according to the vehicle contour line to obtain front sub-vehicle body image data, rear sub-vehicle body image data, first side sub-vehicle body image data, second side sub-vehicle body image data and upper sub-vehicle body image data.
In one possible implementation manner of the first aspect, the performing gaussian filtering of two different preset parameters on each part of sub-vehicle body image data, and subtracting the two filtering results to obtain a corresponding sub-vehicle body SIFT feature map specifically includes:
sequentially passing the sub-vehicle body image data through two Gaussian low-pass filters with different preset parameters and a representation function of a forward distribution function to obtain two Gaussian filtered images;
and subtracting the two Gaussian filtered images to obtain a sub-vehicle SIFT feature map.
In a possible implementation manner of the first aspect, before the calculating, using the extremum point in each sub-vehicle body SIFT feature map as a feature point of the sub-vehicle body image data, a SIFT feature vector and an HSI feature vector of each extremum point as a vehicle body SIFT feature vector and a vehicle body HSI feature vector of the corresponding sub-vehicle body image data, the method further includes:
calculating three-element second-order Taylor expansion at each extreme point in each sub-vehicle body SIFT feature map to obtain the contrast of each extreme point;
for each extreme point, if the contrast of the extreme point is smaller than a preset contrast threshold value, eliminating the extreme point; calculating a local curvature difference through the sea cucumber matrix, and eliminating the extreme points if the local curvature difference of the extreme points is larger than a preset curvature threshold.
In a possible implementation manner of the first aspect, the calculating a SIFT feature vector and an HSI feature vector of each extreme point as a vehicle body SIFT feature vector and a vehicle body HSI feature vector of corresponding sub-vehicle body image data specifically includes:
according to the gradient direction distribution characteristics of the extreme point neighborhood pixels, specifying a direction parameter for each extreme point;
according to the direction parameters, the positions and the scales of the extreme points, SIFT feature vectors of the extreme points are obtained and used as the SIFT feature vectors of the vehicle bodies corresponding to the sub-vehicle body image data;
and calculating an HSI characteristic vector of the extreme point to be used as an HSI characteristic vector of the vehicle body according to the R component value, the G component value and the B component value of the extreme point.
In a possible implementation manner of the first aspect, the specifying, for each extremum point, a direction parameter according to a gradient direction distribution characteristic of the extremum point neighborhood pixel specifically includes:
taking the extreme point as the center to sample a preset neighborhood window, and counting the gradient direction of the neighborhood pixels by using a histogram;
smoothing the histogram by using a preset Gaussian function;
and designating the direction parameter of the extreme point according to the smoothed histogram peak.
In a possible implementation manner of the first aspect, each color number in the color card table, a pixel point is selected at will in a color number area, and a SIFT feature vector and an HS I feature vector are sequentially calculated as a color number SIFT feature vector and a color number HSI feature vector, which specifically include:
according to the gradient direction distribution characteristics of the neighborhood pixels of the pixel points, direction parameters are designated for each extreme point;
according to the direction parameters, the positions and the scales of the pixel points, obtaining SI FT feature vectors of the pixel points as color number SIFT feature vectors of corresponding color numbers;
and calculating the HSI characteristic vector of the pixel point to be used as a color number HSI characteristic vector of the corresponding color number according to the R component value, the G component value and the B component value of the pixel point.
In a possible implementation manner of the first aspect, the calculating, according to the vehicle body SI FT feature vector, the vehicle body HSI feature vector, the color number SIFT feature vector, and the color number HS I feature vector, a matching degree between each part of sub-vehicle body image data and each color number in the color card table includes:
normalizing the SIFT feature vector of the vehicle body, the HS I feature vector of the vehicle body, the SIFT feature vector of the color number and the HS I feature vector of the color number;
if the inner product of the SIFT feature vector of the vehicle body and the SIFT feature vector of the color number is larger than or equal to a matching threshold value and is smaller than or equal to 1, the SIFT feature vector of the vehicle body and the SIFT feature vector of the color number are matched;
if the inner product of the vehicle body HSI feature vector and the color number HSI feature vector is larger than or equal to a matching threshold value and smaller than or equal to 1, the vehicle body HSI feature vector and the color number HS I feature vector are matched;
when the SIFT feature vector of the automobile body is matched with the SIFT feature vector of the color number and the HS I feature vector of the automobile body is matched with the HSI feature vector of the color number, the extreme point in the SIFT feature map of the sub automobile body is matched with the corresponding color number;
and counting the number of extreme points matched with the corresponding color number to serve as the matching degree of the sub-body image data and the color number.
In a possible implementation manner of the first aspect, after the taking the color number corresponding to the maximum value of the matching degree being greater than or equal to the first point number threshold as the target color number of the sub-vehicle body image data and identifying the vehicle paint formula of the sub-vehicle body image data according to the target color number, the method further includes:
if the maximum value of the matching degree is smaller than the first point threshold value, taking the corresponding color number when the matching degree is the maximum value as the target color number of the sub-vehicle body image data and recording all color numbers with the matching degree larger than the second point threshold value;
identifying a target paint formula corresponding to the target color number according to the color chart;
identifying a different vehicle paint formula corresponding to each color number with the matching degree larger than a second point threshold according to the color chart;
comparing the color master composition of each differential paint formulation with the color master composition of the target paint formulation, and counting the color master existing in the differential paint formulation and not in the target paint formulation as an additive color master;
and adding the added color masterbatch into the target paint formula, adjusting the proportion of the added color masterbatch to form a new color until the matching degree of sub-vehicle body image data and the target color number is greater than or equal to a first point threshold value, and recording the target paint formula at the moment as a paint formula identification result.
A second aspect of the embodiment of the present application provides an image recognition-based vehicle paint formulation accurate recognition system, including:
the obtaining module is used for obtaining a factory vehicle paint formula and a factory color number according to the vehicle type information; the factory color number refers to the color number of the factory vehicle paint formula corresponding to the color in a color chart, and each color number corresponds to one color and one vehicle paint formula in the color chart;
the segmentation module is used for acquiring the vehicle body image data and segmenting the vehicle body image data into different sub-vehicle body image data;
the filtering module is used for carrying out Gaussian filtering of two different preset parameters on each part of sub-vehicle body image data, and obtaining a corresponding sub-vehicle body SIFT feature map after subtracting the two filtering results;
the extreme value vector module is used for calculating SIFT feature vectors and HSI feature vectors of each extreme point in each sub-vehicle body SIFT feature map as vehicle body SIFT feature vectors and vehicle body HSI feature vectors of corresponding sub-vehicle body image data;
the color number vector module is used for selecting one pixel point in the color number area at will for each color number in the color card table, and sequentially calculating SIFT feature vectors and HSI feature vectors as color number SIFT feature vectors and color number HSI feature vectors;
the identification module is used for sequentially calculating the matching degree of each part of sub-vehicle body image data and each color number in the color card table according to the vehicle body SIFT feature vector, the vehicle body HSI feature vector, the color number SIFT feature vector and the color number HSI feature vector, taking the color number corresponding to the maximum value of the matching degree which is larger than or equal to a first point number threshold value as a target color number of the sub-vehicle body image data, and identifying the vehicle paint formula of the sub-vehicle body image data according to the target color number.
Compared with the prior art, the embodiment of the application provides a vehicle paint formula accurate identification method and system based on image identification, which are used for obtaining the number of matching points between a sub-vehicle body SIFT feature image and each color number in a color card table according to SIFT feature vectors and HSI feature vectors, further obtaining the matching degree between corresponding sub-vehicle body image data and each color number in the color card table, selecting a target color number and a corresponding vehicle paint formula according to the matching degree, and ensuring the accuracy of vehicle paint formula identification. The whole process does not need manual participation, avoids errors caused by manual participation and saves the time consumed in manually adjusting the paint formula.
In addition, if the target color number in the database does not meet the matching degree condition, the application can also increase the composition components of the vehicle paint formula corresponding to the color number of the target color number in the vehicle paint formula corresponding to the target color number according to the difference of the vehicle paint formulas between the target color number and the factory color number, and further adjust the target color number until the target color number meets the matching degree condition. The matching degree condition is met, namely, the color and the tone of each region can be kept consistent by the vehicle paint formula corresponding to the target color number after adjustment.
Drawings
FIG. 1 is a schematic flow chart of a vehicle paint formulation accurate identification method based on image identification according to an embodiment of the application;
FIG. 2 is a schematic diagram of segmentation of body image data in accordance with an embodiment of the present application;
fig. 3 is a schematic structural diagram of an accurate vehicle paint formulation recognition system based on image recognition according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application 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 application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
Referring to fig. 1, an embodiment of the application provides a vehicle paint formula accurate identification method based on image identification, which includes:
s10, acquiring a factory vehicle paint formula and a factory color number according to vehicle type information; the factory color number refers to the color number of the factory vehicle paint formula corresponding to the color in a color chart, and each color number corresponds to one color and one vehicle paint formula in the color chart.
S11, acquiring vehicle body image data and dividing the vehicle body image data into different sub-vehicle body image data.
S12, gaussian filtering of two different preset parameters is carried out on each part of sub-vehicle body image data, and a corresponding sub-vehicle body SIFT feature diagram is obtained after subtraction of the two filtering results.
S13, in each sub-vehicle body SIFT feature map, SIFT feature vectors and HSI feature vectors of each extreme point are calculated and used as vehicle body SIFT feature vectors and vehicle body HSI feature vectors of corresponding sub-vehicle body image data.
S14, selecting one pixel point in the color number area at will for each color number in the color card table, and sequentially calculating the S IFT feature vector and the HSI feature vector as the color number S IFT feature vector and the color number HS I feature vector.
And S15, calculating the matching degree of each part of sub-vehicle body image data and each color number in a color card table in sequence according to the vehicle body S IFT feature vector, the vehicle body HSI feature vector, the color number SIFT feature vector and the color number HS I feature vector, taking the color number corresponding to the maximum value of the matching degree which is larger than or equal to a first point number threshold value as a target color number of the sub-vehicle body image data, and identifying the vehicle paint formula of the sub-vehicle body image data according to the target color number.
The core of the embodiment is that the matching degree of each part of sub-vehicle body image data and each color number in the color card table is calculated through the vehicle body SIFT feature vector, the vehicle body HS I feature vector, the color number SIFT feature vector and the color number HSI feature vector, and the color number corresponding to the maximum matching degree is selected as the target color number for each part of sub-vehicle body image data. The SIFT feature vector and the HSI feature vector are adopted as the basis for calculating the matching degree, and the basis is as follows: the SIFT feature vector has stronger robustness to operations such as rotation, scaling, blurring, JPEG recompression and the like, and the similarity between a vehicle image (a vehicle paint image) and a fixed pixel (a color corresponding to a color number) is calculated by adopting the S IFT feature vector, so that the SIFT algorithm needs to be firstly converted into a gray image and then the features are extracted for a color image, and most of color information can be lost. The color information of the image can overcome the problem of mismatching caused by consistent gradient information and inconsistent color information when the S IFT features are matched, so that the HSI color features of the SIFT feature are extracted on the basis of extracting the SIFT feature from the feature points. The HSI color model overcomes the defects that the RGB color model only considers the image brightness information and ignores the image color information, and comprehensively considers the brightness, color and other information of the color image. Wherein H represents hue, i.e. human perception of different colors; s represents saturation, i.e. purity of the color; i represents the intensity, i.e. the brightness of the color.
It should be noted that, the color chart adopted in this embodiment is based on the international color chart comparison table, and a vehicle paint formula is associated with each color number, and each vehicle paint formula is composed of a plurality of color masters, so that the vehicle paint formula includes the color master numbers, glossiness, net weights and proportions of the respective color masters in the interior.
Illustratively, S11 is specifically:
obtaining a main view image, a first side view image, a second side view image, a rear view image and a top view image according to vehicle body image data, and taking vehicle edge pixel points in the main view image, the first side view image, the second side view image and the rear view image as vehicle contour lines;
and dividing the vehicle body image data according to the vehicle contour line to obtain front sub-vehicle body image data, rear sub-vehicle body image data, first side sub-vehicle body image data, second side sub-vehicle body image data and upper sub-vehicle body image data.
Referring to fig. 2, the present embodiment divides the vehicle body image data to obtain five sub-vehicle body image data, i.e., front sub-vehicle body image data, rear sub-vehicle body image data, first side sub-vehicle body image data, second side sub-vehicle body image data, and upper sub-vehicle body image data. The reason for image segmentation is that the abrasion degrees of external conditions experienced by different areas of the vehicle body are different, the vehicle body image data are segmented, then a vehicle paint formula is identified for the vehicle body area corresponding to each sub-vehicle body image data, and the different vehicle body areas are subjected to paint repair by using different vehicle paint formulas, so that the vehicle paint formulas are better matched with the vehicle body areas, and the color and tone after paint repair are easier to be consistent in overall vision.
Illustratively, S12 specifically includes:
sequentially passing the sub-vehicle body image data through two Gaussian low-pass filters with different preset parameters and a representation function of a forward distribution function to obtain two Gaussian filtered images;
and subtracting the two Gaussian filtered images to obtain a sub-vehicle SIFT feature map.
Illustratively, prior to S13, further comprising:
and (3) calculating a ternary second-order Taylor expansion at each extreme point in each sub-vehicle body SIFT feature map, and obtaining the contrast of each extreme point.
For each extreme point, if the contrast of the extreme point is smaller than a preset contrast threshold value, eliminating the extreme point; calculating a local curvature difference through the sea cucumber matrix, and eliminating the extreme points if the local curvature difference of the extreme points is larger than a preset curvature threshold.
In this embodiment, the extreme points of local curvature asymmetry are essentially removed. The position and scale of the key points are precisely determined by fitting a three-dimensional quadratic function (reaching sub-pixel precision), and meanwhile, the key points with low contrast and unstable edge response points (because a sub-vehicle body SIFT feature map can generate stronger edge response in S12) are removed, so that matching stability is enhanced, noise resistance is improved, and an approximate HarrisCorner detector is selected to be used in the embodiment.
(1) The spatial scale function taylor expansion is as follows:deriving the above formula and making it 0 to obtain the accurate position, obtain +.>Where D (x) represents the contrast function.
(2) Among the extreme points that have been detected, the extreme points of low contrast and the extreme points of unstable edge response are to be removed. Points of low contrast are removed: substituting the formula (2) into the formula (1), namely taking the value of D (x) at the extreme point of the sub-body SIFT feature map, and obtaining only the first two terms:
as an example, the contrast threshold value in this embodiment is 0.03, ifThe extreme point remains, otherwise it is discarded.
(3) And (3) removing extreme points of edge response:
an extreme point of a sub-body SIFT feature map, defined as bad, has a larger principal curvature across the edge and a smaller principal curvature in the direction perpendicular to the edge. In this example, the principal curvature is determined by a 2×2 Hessian matrix H:the derivative is estimated from the sample point adjacent difference. The principal curvature of D is proportional to the eigenvalue of H.
Illustratively, S13 specifically includes:
according to the gradient direction distribution characteristics of the extreme point neighborhood pixels, specifying a direction parameter for each extreme point;
according to the direction parameters, the positions and the scales of the extreme points, SIFT feature vectors of the extreme points are obtained and used as the SIFT feature vectors of the vehicle bodies corresponding to the sub-vehicle body image data;
and calculating an HSI characteristic vector of the extreme point to be used as an HSI characteristic vector of the vehicle body according to the R component value, the G component value and the B component value of the extreme point.
According to the embodiment, a direction is calculated for each extreme point, further calculation is performed according to the direction, and the gradient direction distribution characteristics of the extreme point neighborhood pixels are utilized to assign direction parameters for each extreme point, so that the sub-vehicle body SIFT feature map has rotation invariance. Each extreme point has three pieces of information: the position of the extreme point, the scale where the extreme point is located and the direction of the extreme point, thereby determining a SIFT feature area.
Illustratively, the specifying the direction parameter for each extreme point according to the gradient direction distribution characteristic of the extreme point neighborhood pixel specifically includes:
taking the extreme point as the center to sample a preset neighborhood window, and counting the gradient direction of the neighborhood pixels by using a histogram;
smoothing the histogram by using a preset Gaussian function;
and designating the direction parameter of the extreme point according to the smoothed histogram peak.
The gradient histogram ranges from 0 to 360 degrees, with one bin per 10 degrees, for a total of 36 bins. The contribution to the histogram also decreases with the area farther from the center point. The present embodiment uses a gaussian function to smooth the histogram in order to reduce the effects of abrupt changes. In actual calculation, sampling is carried out in a neighborhood window taking an extreme point as a center, and the gradient direction of the neighborhood pixels is counted by using a histogram. The gradient histogram ranges from 0 to 360 degrees with one bin every 45 degrees, a total of 8 bins, or a total of 36 bins every 10 degrees. The peak of the histogram then represents the main direction of the neighborhood gradient at the extreme point, i.e. the direction that is the extreme point.
Illustratively, S14 specifically includes:
according to the gradient direction distribution characteristics of the neighborhood pixels of the pixel points, direction parameters are designated for each extreme point;
according to the direction parameters, the positions and the scales of the pixel points, obtaining SI FT feature vectors of the pixel points as color number SIFT feature vectors of corresponding color numbers;
and calculating the HSI characteristic vector of the pixel point to be used as a color number HSI characteristic vector of the corresponding color number according to the R component value, the G component value and the B component value of the pixel point.
The calculation process of the SIFT feature vector and the HSI feature vector of the pixel point is basically identical to the calculation process of the S IFT feature vector and the HS I feature vector of the extreme point in the above embodiment, and will not be described herein.
The calculating, according to the vehicle body SIFT feature vector, the vehicle body HS I feature vector, the color number SIFT feature vector, and the color number HSI feature vector, the matching degree between each part of sub-vehicle body image data and each color number in the color card table in turn specifically includes:
normalizing the SIFT feature vector of the vehicle body, the HS I feature vector of the vehicle body, the SIFT feature vector of the color number and the HS I feature vector of the color number;
if the inner product of the SIFT feature vector of the vehicle body and the SIFT feature vector of the color number is larger than or equal to a matching threshold value and is smaller than or equal to 1, the SIFT feature vector of the vehicle body and the SIFT feature vector of the color number are matched;
if the inner product of the vehicle body HSI feature vector and the color number HSI feature vector is larger than or equal to a matching threshold value and smaller than or equal to 1, the vehicle body HSI feature vector and the color number HS I feature vector are matched;
when the SIFT feature vector of the automobile body is matched with the SIFT feature vector of the color number and the HS I feature vector of the automobile body is matched with the HSI feature vector of the color number, the extreme point in the SIFT feature map of the sub automobile body is matched with the corresponding color number;
and counting the number of extreme points matched with the corresponding color number to serve as the matching degree of the sub-body image data and the color number.
After SIFT and color feature vectors of extreme points (located in a sub-vehicle body SIFT feature map) and pixel points (located in colors corresponding to color numbers of color cards) are respectively constructed, feature matching is carried out on any two extreme points and pixel points, and whether a copy-paste tampered area exists in an image is judged. The process of feature matching is as follows:
to reduce the effects of uneven lighting conditions, the SIFT feature and HSI color feature vectors are first normalized separately. After the feature extraction is carried out on the extreme point, 128-dimensional SIFT feature vectors and 3-dimensional HSI color feature vectors can be obtained. Taking an extreme point as an example, setting a SIFT feature vector of the extreme point i as car_sift (i), wherein the normalization process is shown in a formula (3); the HSI color feature vector is car_hsi (i), and the normalization process is shown in the formula (4).
Where car_sift_nor (i) is normalized car_sift (i); the i car_sift (i) is the modulus of the vector car_sift (i).
Wherein, car_hsi_nor (i) is normalized car_hsi (i); the i car_hsi (i) is the modulus of the vector feat_hsi (i).
For an extreme point i and a pixel point j, calculating the inner products of normalized SIFT feature vectors and color feature vectors, namely in_sift and in_hsi, as shown in formulas (5) and (6):
in_sift＝car_sift_nor(i)*car_sift_nor(j)′ (5)
in_hsi＝car_hsi_nor(i)*car_hsi_nor(j)′ (6)
if the extreme point i and the pixel point j match, both their SIFT feature vector and HSI color feature vector should be equal, i.e., feature_sift_nor (i) =feature_sift_nor (j), and feature_hsi_nor (i) =feature_hsi_nor (j), i.e., the inner products of SIFT feature and HSI feature of the two feature points are 1, respectively. If the inner products are equal, the similarity judgment between the two images is severe, so that a matching threshold thr is introduced to balance the calculation complexity and the mismatching rate of the algorithm, in other words, if the inner products are in the interval [ thr,1], the extreme point i and the pixel point j are considered to be matched. The matching degree is substantially equal to the number of matching extreme points of the pixel point j in a certain color area in the color card table in the sub-vehicle body SIFT feature map.
The method for identifying the paint formula of the sub-vehicle body image data according to the target color number further includes:
if the maximum value of the matching degree is smaller than the first point threshold value, taking the corresponding color number when the matching degree is the maximum value as the target color number of the sub-vehicle body image data and recording all color numbers with the matching degree larger than the second point threshold value;
identifying a target paint formula corresponding to the target color number according to the color chart;
identifying a different vehicle paint formula corresponding to each color number with the matching degree larger than a second point threshold according to the color chart;
comparing the color master composition of each differential paint formulation with the color master composition of the target paint formulation, and counting the color master existing in the differential paint formulation and not in the target paint formulation as an additive color master;
and adding the added color masterbatch into the target paint formula, adjusting the proportion of the added color masterbatch to form a new color until the matching degree of sub-vehicle body image data and the target color number is greater than or equal to a first point threshold value, and recording the target paint formula at the moment as a paint formula identification result.
The differential paint formulation is a paint formulation having a color closest to the current actual paint color on the vehicle surface, in addition to the target paint formulation. The addition of the color master is a cause of the difference between the different vehicle paint formulas and the target vehicle paint formulas, and the color of the target vehicle paint formulas can be adjusted by adding the addition color master into the target vehicle paint formulas, so that the color of the target vehicle paint formulas gradually approaches to the color of the actual vehicle paint on the surface of the vehicle.
Compared with the prior art, the embodiment of the application provides a vehicle paint formula accurate identification method based on image identification, which obtains the number of matching points between the sub-vehicle body S IFT feature map and each color number in the color card table according to SIFT feature vectors and HSI feature vectors, further obtains the matching degree between corresponding sub-vehicle body image data and each color number in the color card table, selects a target color number and a corresponding vehicle paint formula according to the matching degree, and ensures the accuracy of vehicle paint formula identification. The whole process does not need manual participation, avoids errors caused by manual participation and saves the time consumed in manually adjusting the paint formula.
In addition, if the target color number in the database does not meet the matching degree condition, the application can also increase the composition components of the vehicle paint formula corresponding to the color number of the target color number in the vehicle paint formula corresponding to the target color number according to the difference of the vehicle paint formulas between the target color number and the factory color number, and further adjust the target color number until the target color number meets the matching degree condition. The matching degree condition is met, namely, the color and the tone of each region can be kept consistent by the vehicle paint formula corresponding to the target color number after adjustment.
Referring to fig. 3, an embodiment of the present application provides a vehicle paint formulation accurate recognition system based on image recognition, including: the device comprises an acquisition module 20, a segmentation module 21, a filtering module 22, an extremum vector module 23, a color number vector module 24 and an identification module 25.
The obtaining module 20 is used for obtaining a factory paint formula and a factory color number according to the vehicle type information; the factory color number refers to the color number of the factory vehicle paint formula corresponding to the color in a color chart, and each color number corresponds to one color and one vehicle paint formula in the color chart.
The segmentation module 21 is used for acquiring the vehicle body image data and carrying out sub-vehicle body image data with different vehicle body image data.
The filtering module 22 is configured to perform gaussian filtering of two different preset parameters on each part of sub-vehicle body image data, and obtain a corresponding sub-vehicle body SIFT feature map after subtracting the two filtering results.
The extremum vector module 23 is configured to calculate, in each sub-vehicle body SIFT feature map, a SIFT feature vector and an HSI feature vector of each extremum point as a vehicle body SIFT feature vector and a vehicle body HSI feature vector of corresponding sub-vehicle body image data.
The color number vector module 24 is configured to, for each color number in the color card table, select a pixel point in the color number area at will, and sequentially calculate a SIFT feature vector and an HSI feature vector as a color number SIFT feature vector and a color number HSI feature vector.
The identifying module 25 is configured to sequentially calculate a matching degree between each part of sub-vehicle body image data and each color number in the color card table according to the vehicle body SIFT feature vector, the vehicle body HSI feature vector, the color number S IFT feature vector and the color number HSI feature vector, take a color number corresponding to a maximum value of the matching degree greater than or equal to a first point number threshold value as a target color number of the sub-vehicle body image data, and identify a vehicle paint formula of the sub-vehicle body image data according to the target color number.
It will be clear to those skilled in the art that, for convenience and brevity of description, the specific working process of the identification system described above may refer to the corresponding process in the foregoing method embodiment, which is not described herein again.
Compared with the prior art, the embodiment of the application provides an accurate vehicle paint formula identification system based on image identification, which obtains the number of matching points between the sub-vehicle body S IFT feature map and each color number in the color card table according to SIFT feature vectors and HSI feature vectors, further obtains the matching degree between corresponding sub-vehicle body image data and each color number in the color card table, selects a target color number and a corresponding vehicle paint formula according to the matching degree, and ensures the accuracy of vehicle paint formula identification. The whole process does not need manual participation, avoids errors caused by manual participation and saves the time consumed in manually adjusting the paint formula.
In addition, if the target color number in the database does not meet the matching degree condition, the application can also increase the composition components of the vehicle paint formula corresponding to the color number of the target color number in the vehicle paint formula corresponding to the target color number according to the difference of the vehicle paint formulas between the target color number and the factory color number, and further adjust the target color number until the target color number meets the matching degree condition. The matching degree condition is met, namely, the color and the tone of each region can be kept consistent by the vehicle paint formula corresponding to the target color number after adjustment.
While the foregoing is directed to the preferred embodiments of the present application, it will be appreciated by those skilled in the art that changes and modifications may be made without departing from the principles of the application, such changes and modifications are also intended to be within the scope of the application.
Claims (10)
1. The vehicle paint formula accurate identification method based on image identification is characterized by comprising the following steps of:
acquiring a factory paint formula and a factory color number according to the vehicle type information; the factory color number refers to the color number of the factory vehicle paint formula corresponding to the color in a color chart, and each color number corresponds to one color and one vehicle paint formula in the color chart;
acquiring vehicle body image data and dividing the vehicle body image data into different sub-vehicle body image data;
carrying out Gaussian filtering of two different preset parameters on each part of sub-vehicle body image data, and obtaining a corresponding sub-vehicle body SIFT feature map after subtracting the two filtering results;
in each sub-vehicle body SIFT feature map, SIFT feature vectors and HSI feature vectors of each extreme point are calculated and used as vehicle body SIFT feature vectors and vehicle body HSI feature vectors of corresponding sub-vehicle body image data;
for each color number in the color card table, randomly selecting one pixel point in a color number area, and sequentially calculating SIFT feature vectors and HSI feature vectors as color number SIFT feature vectors and color number HSI feature vectors;
and according to the SIFT feature vector of the automobile body, the HSI feature vector of the automobile body, the SIFT feature vector of the color number and the HSI feature vector of the color number, calculating the matching degree of each part of sub-automobile body image data and each color number in a color card table in sequence, taking the color number corresponding to the maximum value of the matching degree which is larger than or equal to a first point number threshold value as a target color number of the sub-automobile body image data, and identifying the automobile paint formula of the sub-automobile body image data according to the target color number.
2. The precise identification method of the paint formulation based on the image identification according to claim 1, wherein the steps of acquiring the image data of the vehicle body and dividing the image data of the vehicle body into different sub-image data of the vehicle body are as follows:
obtaining a main view image, a first side view image, a second side view image, a rear view image and a top view image according to vehicle body image data, and taking vehicle edge pixel points in the main view image, the first side view image, the second side view image and the rear view image as vehicle contour lines;
and dividing the vehicle body image data according to the vehicle contour line to obtain front sub-vehicle body image data, rear sub-vehicle body image data, first side sub-vehicle body image data, second side sub-vehicle body image data and upper sub-vehicle body image data.
3. The method for accurately identifying the paint formulation of the vehicle based on the image identification of claim 1, wherein the gaussian filtering of two different preset parameters is performed on each part of the image data of the sub-vehicle body, and the corresponding sub-vehicle body SIFT feature map is obtained by subtracting the two filtering results, specifically comprising:
sequentially passing the sub-vehicle body image data through two Gaussian low-pass filters with different preset parameters and a representation function of a forward distribution function to obtain two Gaussian filtered images;
and subtracting the two Gaussian filtered images to obtain a sub-vehicle SIFT feature map.
4. The method for accurately identifying the paint formulation of the vehicle based on the image identification of claim 1, wherein before taking the extreme point in each sub-vehicle body SIFT feature map as the feature point of the sub-vehicle body image data and calculating the SIFT feature vector and the HSI feature vector of each extreme point as the vehicle body SIFT feature vector and the vehicle body HSI feature vector of the corresponding sub-vehicle body image data, the method further comprises:
calculating three-element second-order Taylor expansion at each extreme point in each sub-vehicle body SIFT feature map to obtain the contrast of each extreme point;
for each extreme point, if the contrast of the extreme point is smaller than a preset contrast threshold value, eliminating the extreme point; calculating a local curvature difference through the sea cucumber matrix, and eliminating the extreme points if the local curvature difference of the extreme points is larger than a preset curvature threshold.
5. The method for accurately identifying the paint formulation of the vehicle based on the image identification according to claim 1, wherein the calculating of the SIFT feature vector and the HSI feature vector of each extreme point as the vehicle body SIFT feature vector and the vehicle body HSI feature vector of the corresponding sub-vehicle body image data specifically comprises:
according to the gradient direction distribution characteristics of the extreme point neighborhood pixels, specifying a direction parameter for each extreme point;
according to the direction parameters, the positions and the scales of the extreme points, SIFT feature vectors of the extreme points are obtained and used as the SIFT feature vectors of the vehicle bodies corresponding to the sub-vehicle body image data;
and calculating an HSI characteristic vector of the extreme point to be used as an HSI characteristic vector of the vehicle body according to the R component value, the G component value and the B component value of the extreme point.
6. The method for accurately identifying the paint formulation of the vehicle based on the image identification according to claim 5, wherein the specifying the direction parameter for each extreme point according to the gradient direction distribution characteristic of the pixels in the neighborhood of the extreme point specifically comprises:
taking the extreme point as the center to sample a preset neighborhood window, and counting the gradient direction of the neighborhood pixels by using a histogram;
smoothing the histogram by using a preset Gaussian function;
and designating the direction parameter of the extreme point according to the smoothed histogram peak.
7. The vehicle paint formula accurate identification method based on image identification according to claim 1, wherein each color number in the color matching card table is selected at will, and a SIFT feature vector and an HSI feature vector are sequentially calculated in a color number region as a color number SIFT feature vector and a color number HSI feature vector, and specifically comprises the following steps:
according to the gradient direction distribution characteristics of the neighborhood pixels of the pixel points, direction parameters are designated for each extreme point;
according to the direction parameters, the positions and the scales of the pixel points, SIFT feature vectors of the pixel points are obtained and used as color number SIFT feature vectors of corresponding color numbers;
and calculating the HSI characteristic vector of the pixel point to be used as a color number HSI characteristic vector of the corresponding color number according to the R component value, the G component value and the B component value of the pixel point.
8. The method for accurately identifying the paint formulation of the vehicle based on the image identification according to claim 1, wherein the calculating the matching degree between each part of sub-vehicle body image data and each color number in the color card table according to the vehicle body SIFT feature vector, the vehicle body HSI feature vector, the color number SIFT feature vector and the color number HSI feature vector comprises the following steps:
normalizing the SIFT feature vector of the vehicle body, the HSI feature vector of the vehicle body, the SIFT feature vector of the color number and the HSI feature vector of the color number;
if the inner product of the SIFT feature vector of the vehicle body and the SIFT feature vector of the color number is larger than or equal to a matching threshold value and is smaller than or equal to 1, the SIFT feature vector of the vehicle body and the SIFT feature vector of the color number are matched;
if the inner product of the vehicle body HSI feature vector and the color number HSI feature vector is larger than or equal to a matching threshold value and smaller than or equal to 1, the vehicle body HSI feature vector and the color number HSI feature vector are matched;
when the SIFT feature vector of the automobile body is matched with the SIFT feature vector of the color number and the HSI feature vector of the automobile body is matched with the HSI feature vector of the color number, the extreme point in the SIFT feature map of the sub automobile body is matched with the corresponding color number;
and counting the number of extreme points matched with the corresponding color number to serve as the matching degree of the sub-body image data and the color number.
9. The method for accurately identifying a paint formulation based on image identification according to claim 1, wherein after using a color number corresponding to a maximum value of the matching degree greater than or equal to a first point threshold as a target color number of the sub-vehicle body image data and identifying the paint formulation of the sub-vehicle body image data according to the target color number, further comprises:
if the maximum value of the matching degree is smaller than the first point threshold value, taking the corresponding color number when the matching degree is the maximum value as the target color number of the sub-vehicle body image data and recording all color numbers with the matching degree larger than the second point threshold value;
identifying a target paint formula corresponding to the target color number according to the color chart;
identifying a different vehicle paint formula corresponding to each color number with the matching degree larger than a second point threshold according to the color chart;
comparing the color master composition of each differential paint formulation with the color master composition of the target paint formulation, and counting the color master existing in the differential paint formulation and not in the target paint formulation as an additive color master;
and adding the added color masterbatch into the target paint formula, adjusting the proportion of the added color masterbatch to form a new color until the matching degree of sub-vehicle body image data and the target color number is greater than or equal to a first point threshold value, and recording the target paint formula at the moment as a paint formula identification result.
10. Accurate identification system of car lacquer formula based on image recognition, characterized by comprising:
the obtaining module is used for obtaining a factory vehicle paint formula and a factory color number according to the vehicle type information; the factory color number refers to the color number of the factory vehicle paint formula corresponding to the color in a color chart, and each color number corresponds to one color and one vehicle paint formula in the color chart;
the segmentation module is used for acquiring the vehicle body image data and segmenting the vehicle body image data into different sub-vehicle body image data;
the filtering module is used for carrying out Gaussian filtering of two different preset parameters on each part of sub-vehicle body image data, and obtaining a corresponding sub-vehicle body SIFT feature map after subtracting the two filtering results;
the extreme value vector module is used for calculating SIFT feature vectors and HSI feature vectors of each extreme point in each sub-vehicle body SIFT feature map as vehicle body SIFT feature vectors and vehicle body HSI feature vectors of corresponding sub-vehicle body image data;
the color number vector module is used for selecting one pixel point in the color number area at will for each color number in the color card table, and sequentially calculating SIFT feature vectors and HSI feature vectors as color number SIFT feature vectors and color number HSI feature vectors;
the identification module is used for sequentially calculating the matching degree of each part of sub-vehicle body image data and each color number in the color card table according to the vehicle body SIFT feature vector, the vehicle body HSI feature vector, the color number SIFT feature vector and the color number HSI feature vector, taking the color number corresponding to the maximum value of the matching degree which is larger than or equal to a first point number threshold value as a target color number of the sub-vehicle body image data, and identifying the vehicle paint formula of the sub-vehicle body image data according to the target color number.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310731844.2A CN116740383B (en) | 2023-06-19 | Vehicle paint formula accurate identification method and system based on image identification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310731844.2A CN116740383B (en) | 2023-06-19 | Vehicle paint formula accurate identification method and system based on image identification |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116740383A true CN116740383A (en) | 2023-09-12 |
| CN116740383B CN116740383B (en) | 2024-06-21 |
Family
ID=
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1139234A1 (en) * | 2000-03-28 | 2001-10-04 | E.I. Dupont De Nemours And Company | Color matching method for automotive refinishing |
| CN102012939A (en) * | 2010-12-13 | 2011-04-13 | 中国人民解放军国防科学技术大学 | Method for automatically tagging animation scenes for matching through comprehensively utilizing overall color feature and local invariant features |
| CN102236901A (en) * | 2011-06-30 | 2011-11-09 | 南京大学 | Method for tracking target based on graph theory cluster and color invariant space |
| CN102549545A (en) * | 2008-10-31 | 2012-07-04 | Ppg工业俄亥俄公司 | Color formulation selection process with visual display |
| CN103150904A (en) * | 2013-02-05 | 2013-06-12 | 中山大学 | Bayonet vehicle image identification method based on image features |
| CN103544484A (en) * | 2013-10-30 | 2014-01-29 | 广东工业大学 | Traffic sign identification method and system based on SURF |
| CN106340007A (en) * | 2016-06-13 | 2017-01-18 | 吉林大学 | Image processing-based automobile body paint film defect detection and identification method |
| WO2018040756A1 (en) * | 2016-09-05 | 2018-03-08 | 华为技术有限公司 | Vehicle body colour identification method and device |
| CN109934092A (en) * | 2019-01-18 | 2019-06-25 | 深圳壹账通智能科技有限公司 | Identify color method, apparatus, computer equipment and storage medium |
| CN114332254A (en) * | 2021-12-31 | 2022-04-12 | 北京奔驰汽车有限公司 | AI-based vehicle body color and cover type identification system and method and electronic equipment |
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1139234A1 (en) * | 2000-03-28 | 2001-10-04 | E.I. Dupont De Nemours And Company | Color matching method for automotive refinishing |
| CN102549545A (en) * | 2008-10-31 | 2012-07-04 | Ppg工业俄亥俄公司 | Color formulation selection process with visual display |
| CN102012939A (en) * | 2010-12-13 | 2011-04-13 | 中国人民解放军国防科学技术大学 | Method for automatically tagging animation scenes for matching through comprehensively utilizing overall color feature and local invariant features |
| CN102236901A (en) * | 2011-06-30 | 2011-11-09 | 南京大学 | Method for tracking target based on graph theory cluster and color invariant space |
| CN103150904A (en) * | 2013-02-05 | 2013-06-12 | 中山大学 | Bayonet vehicle image identification method based on image features |
| CN103544484A (en) * | 2013-10-30 | 2014-01-29 | 广东工业大学 | Traffic sign identification method and system based on SURF |
| CN106340007A (en) * | 2016-06-13 | 2017-01-18 | 吉林大学 | Image processing-based automobile body paint film defect detection and identification method |
| WO2018040756A1 (en) * | 2016-09-05 | 2018-03-08 | 华为技术有限公司 | Vehicle body colour identification method and device |
| CN109934092A (en) * | 2019-01-18 | 2019-06-25 | 深圳壹账通智能科技有限公司 | Identify color method, apparatus, computer equipment and storage medium |
| CN114332254A (en) * | 2021-12-31 | 2022-04-12 | 北京奔驰汽车有限公司 | AI-based vehicle body color and cover type identification system and method and electronic equipment |
Non-Patent Citations (3)
| Title |
|---|
| SWATHI M, K.V. SURESH: "Automatic Traffic Sign Detection and Recognition: A Review", 《2017 INTERNATIONAL CONFERENCE ON ALGORITHMS, METHODOLOGY, MODELS AND APLLICATIONS IN EMERGING TECHNOLOGIES》, 14 December 2017 (2017-12-14), pages 1 - 6 * |
| 郭家伟: "抛光石材表面颜色识别系统研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑 》, 15 April 2019 (2019-04-15), pages 038 - 965 * |
| 陈孝之，谢莉青: "织物颜色配准到标准色卡的计算机识别与仿真", 《纺织学报》, vol. 37, no. 5, 31 May 2016 (2016-05-31), pages 150 - 154 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110490914B (en) | Image fusion method based on brightness self-adaption and significance detection | |
| CN106651872B (en) | Pavement crack identification method and system based on Prewitt operator | |
| CN108765325B (en) | Small unmanned aerial vehicle blurred image restoration method | |
| CN106204509B (en) | Infrared and visible light image fusion method based on regional characteristics | |
| CN111145161A (en) | Method for processing and identifying pavement crack digital image | |
| CN112819772B (en) | High-precision rapid pattern detection and recognition method | |
| CN108389175B (en) | Image defogging method integrating variation function and color attenuation prior | |
| CN107967482A (en) | Icon-based programming method and device | |
| CN109101924A (en) | A kind of pavement marking recognition methods based on machine learning | |
| CN109242870A (en) | A kind of sea horizon detection method divided based on image with textural characteristics | |
| CN114118144A (en) | Anti-interference accurate aerial remote sensing image shadow detection method | |
| CN114998198A (en) | Injection molding surface defect identification method | |
| CN108491498A (en) | A kind of bayonet image object searching method based on multiple features detection | |
| CN107832674B (en) | Lane line detection method | |
| CN110705553B (en) | Scratch detection method suitable for vehicle distant view image | |
| Gao et al. | Colour vision model-based approach for segmentation of traffic signs | |
| CN116740383B (en) | Vehicle paint formula accurate identification method and system based on image identification | |
| CN114842201A (en) | Sandstone aggregate image segmentation method based on improved Mask _ Rcnn | |
| CN117152135B (en) | Road construction crack defect evaluation and detection method | |
| CN107886549B (en) | Fabric pattern color transfer method based on weaving texture detail enhancement | |
| CN111862178B (en) | Improved LBP feature extraction method | |
| CN116740383A (en) | Vehicle paint formula accurate identification method and system based on image identification | |
| CN115100510B (en) | Tire wear degree identification method | |
| CN116205903A (en) | Method for detecting defects of vehicle body finish paint | |
| CN112669360B (en) | Multi-source image registration method based on non-closed multi-dimensional contour feature sequence |
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 |