CN109636794B - Machine learning-based subway height adjusting valve fastening nut positioning method - Google Patents
Machine learning-based subway height adjusting valve fastening nut positioning method Download PDFInfo
- Publication number
- CN109636794B CN109636794B CN201811537193.9A CN201811537193A CN109636794B CN 109636794 B CN109636794 B CN 109636794B CN 201811537193 A CN201811537193 A CN 201811537193A CN 109636794 B CN109636794 B CN 109636794B
- Authority
- CN
- China
- Prior art keywords
- image
- height adjusting
- fastening nut
- adjusting valve
- subway
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration by the use of histogram techniques
-
- 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
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- 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
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
Abstract
The invention discloses a machine learning-based method for positioning a fastening nut of a subway height adjusting valve, which comprises the following steps of: s1, acquiring image information of the bottom of the whole subway train, and cutting the image according to different carriages to obtain a high-definition color image of the bottom of the whole subway train, which contains a height adjusting valve component; s2, establishing a deep neural network classifier to position the height adjusting valve component; establishing a support vector machine classifier to position a position area of a fastening nut in a height adjusting valve component; s3, performing primary segmentation on the positioned fastening nut image by using a K-Means clustering algorithm; and S4, accurately dividing the fastening nut of the subway height adjusting valve according to a boundary characteristic method.
Description
Technical Field
The invention relates to the field of subway mechanical part fault detection, in particular to a method for positioning a fastening nut of a subway height adjusting valve based on machine learning.
Background
At present, in the daily detection of a subway running gear, a fastening nut of a height adjusting valve is a part needing important monitoring. For guaranteeing the driving safety of the metro vehicle and realizing the accurate monitoring of key components, the accurate positioning of the fastening nut component of the height adjusting valve becomes the key premise for the abnormal detection of the whole component. The vibration or the overhauling error generated by the long-term operation of the subway train can cause the poor states of falling, loosening and the like of the fastening nut of the height adjusting valve, so that the bearing capacity of the adjusting valve is reduced, the mechanical strength is reduced, and the possibility of accidents is increased. Therefore, it is critical to achieve accurate positioning of the height-adjusting valve fastening nut.
At present still use artifical the detection as the owner to subway altitude mixture control valve fastening nut's unusual detection, owing to wait to detect regional area big, and wait to detect the nut area minimum, accomplish nut location and unusual detection and all need a large amount of manpowers, the work load is big, and efficiency is very low. The subway height adjusting valve fastening nut based on the digital image processing technology can complete on-line real-time positioning under the non-contact condition, and has strong advantages. The state monitoring of the special fastening nut for the height adjusting valve in the subway running part at home and abroad is still blank, and the nut positioning research for the part is more rarely reported. Because the size of the fastening nut is extremely small relative to the whole vehicle image, and the part is easily influenced by reflection, detection by adopting an image processing technology has certain difficulty, and a research report in the aspect is not searched at present.
Disclosure of Invention
According to the problems in the prior art, the invention discloses a method for positioning a fastening nut of a subway height adjusting valve based on a deep learning classifier, which specifically comprises the following steps:
s1, acquiring image information of the bottom of the whole subway train, and cutting the image according to different carriages to obtain a high-definition color image of the bottom of the whole subway train, which contains a height adjusting valve component;
s2, establishing a deep neural network classifier to position the height adjusting valve component; establishing a support vector machine classifier to position a position area of a fastening nut in a height adjusting valve component;
s3, performing primary segmentation on the positioned fastening nut image by using a K-Means clustering algorithm;
s4, accurately dividing the fastening nut of the subway height adjusting valve according to a boundary characteristic method: and (4) carrying out limited region searching on the preliminarily segmented binary image by adopting a region growing method to remove background miscellaneous points, and accurately extracting the fastening nut.
The S2 specifically adopts the following mode:
a. generating a sample library of height adjusting valve components, wherein positive samples are images only containing the height adjusting valve components, and negative samples are images of other components not containing the height adjusting valves;
b. training the sample by adopting a deep neural network algorithm, establishing a deep learning classifier, and integrally positioning the height adjusting valve component;
c. generating a sample library of the fastening nuts in the height adjusting valve parts, wherein positive samples are images containing the fastening nuts, and negative samples are images of other parts not containing the fastening nuts;
d. and finishing the training of positive and negative samples by adopting a support vector machine, establishing a fastening nut classifier, and realizing the positioning of the fastening nut.
S3 specifically adopts the following mode:
s31, performing contrast enhancement processing on the positioned fastening nut image, and increasing the contrast difference between the interested fastening nut area and the background area in the image by adopting a histogram equalization method;
s32, clustering and dividing the image by utilizing a K-Means clustering algorithm, and separating a fastening nut area from an image background area, wherein the following method is specifically adopted:
s321, setting the number of sample classification to be 2, namely, dividing the sample classification into C 1 、C 2 Selecting a clustering center for each category, and determining an initial clustering center point by adopting a centralization algorithm;
s322, calculating the Euclidean distance between each pixel point and the clustering centers by adopting a formula (2) for each pixel point in the image, and classifying each pixel point to the class corresponding to the clustering center closest to the pixel point according to the closest criterion by adopting a formula (3);
where E is the square error, x is the sample point, μ i Is of the class C i The mean vector of (2);
s323, updating the clustering center: taking the mean values corresponding to all the objects in each category as the clustering center of the category to obtain 2 updated center points;
s324, setting the termination condition of the algorithm as the maximum cycle number M and the precision threshold N, comparing the difference between the updated 2 central points and the 2 initial clustering central points in the S321, and if the difference is smaller than the precision threshold N, giving gray values of pixel points 0 or 255 according to different categories and outputting images; if the difference is larger than the precision threshold, continuing S322-S324 until the cycle meets the condition;
s325, carrying out gray level judgment on the image, and outputting a final result after inverting the gray level of the image if the number of pixels of the output image gray level is more than 255 because the number of pixels of the ROI is less than that of pixels occupied by the background area; otherwise, directly outputting the final result;
and S33, processing the image by using the opening operation of the image morphology to eliminate the interference image points of the part.
S4 specifically comprises the following steps:
calculating horizontal direction projection and vertical direction projection of the ROI area, further determining the area where the nut is located according to boundary characteristics represented by a projection curve, dividing an original image by taking a horizontal straight line passing through a peak point vertical coordinate as a center in the horizontal direction projection, taking a half position of the respective height of an upper image and a lower image to define a horizontal line, and setting the horizontal line as an upper boundary and a lower boundary of the ROI respectively; in the vertical direction projection, the first trough appearing on the right side of the peak value is recorded as a ROI left boundary, the first non-zero value appearing on the right side of the image is recorded as a ROI right boundary, and accurate extraction of the region where the nut is located is completed.
Due to the adoption of the technical scheme, the method for positioning the fastening nut of the subway height adjusting valve based on the deep learning classifier directly utilizes the image processing technology to realize the positioning of the fastening nut of the subway height adjusting valve, can provide objective and accurate positioning results, and avoids the defects of the traditional manual positioning method. The method provides a meaningful reference for the subsequent detection of the bad state of the fastening nut of the subway height adjusting valve. According to the method, according to the data characteristics of a fastening nut of the height adjusting valve in the image of the whole vehicle, classifiers of deep learning and machine learning are combined and positioned step by step, and the problems of large data volume and time consumption in operation are effectively solved; the method fully considers the characteristics of the nut and the background environment, combines the K-Means algorithm, the region growing method, the target gray distribution characteristics and the like to design the nut positioning method, overcomes the influence of the nut which is easy to be misjudged under the complex background environment, and has accurate and effective positioning result; the invention relates to a method for positioning a fastening nut of a subway height adjusting valve, and related researches are rarely reported.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments described in the present application, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a block flow diagram of the method of the present invention.
FIG. 2 is an image containing a height-adjusting valve taken from an original image according to the present invention.
FIG. 3 is an illustration of a height adjustment valve assembly classified using a deep neural network according to the present invention.
FIG. 4 is a sample support vector machine of the present invention for a height adjustment valve member fastening nut, FIG. 4 (a) positive sample, FIG. 4 (b) negative sample.
Fig. 5 is an image of a fastening nut in accordance with the present invention initially positioned.
Fig. 6 is a fastening nut image after contrast enhancement according to the present invention.
FIG. 7 is a schematic view of a fastening nut segmented by the K-Means clustering algorithm according to the present invention.
FIG. 8 is a schematic view of the fastening nut processed by the morphological algorithm of the present invention.
FIG. 9 is a schematic illustration of an image of a fastening nut of the present invention projected in both horizontal and vertical directions.
FIG. 10 is a schematic representation of the precise segmentation of the clinch nut based on boundary features of the present invention. Fig. 10 (a) shows a positive sample, and fig. 10 (b) shows a negative sample.
FIG. 11 is a schematic view of the positioning of the split nut after removing background noise by the region growing method.
Detailed Description
In order to make the technical solutions and advantages of the present invention clearer, the following describes the technical solutions in the embodiments of the present invention clearly and completely with reference to the drawings in the embodiments of the present invention:
as shown in fig. 1, the method for positioning the fastening nut of the subway height adjusting valve based on machine learning specifically comprises the following steps:
1. the method comprises the steps of collecting image information of the whole bottom of the subway train by using a high-definition color linear array camera and a speed measuring module, cutting according to different carriages to obtain a high-definition color image of the whole bottom of the subway train, wherein the high-definition color image comprises a height adjusting valve component, and intercepting and displaying a part of images for convenient observation due to overlarge images, as shown in figure 2.
2. Subway altitude mixture control valve fastening nut preliminary positioning based on degree of deep learning classifier:
2.1 establishing a sample library of the height adjusting valve fastening nut component, intercepting an image containing the height adjusting valve fastening nut as a sample, and setting a corresponding label, wherein the number of the samples is 3500;
2.2, completing sample training by adopting a deep neural network algorithm, establishing a component classifier, realizing the overall positioning of the height adjusting valve fastening nut component, and obtaining an image set of the height adjusting valve fastening nut component, wherein a sample image is shown in fig. 3;
2.3 establishing a sample library of the fastening nuts in the height adjusting valve parts, wherein positive samples are images of the fastening nuts, the number of which is 150, and negative samples are images of the rest parts, the number of which is 200, as shown in fig. 4;
2.4, completing positive and negative sample training by adopting a support vector machine, establishing a fastening nut classifier, extracting an interested fastening nut region, and obtaining an extraction result as shown in FIG. 5; and solving the optimal classification hyperplane solving problem, namely the convex quadratic programming problem, under the condition that the training data set is linearly inseparable by using a support vector machine.
3. Preliminary segmentation is carried out on a fastening nut based on a K-Means clustering algorithm;
3.1 because the nut to be detected has extremely small size relative to the whole train, the number of pixels occupied in the image is very small, in order to conveniently analyze the current state of the nut, the contrast enhancement processing needs to be carried out on the extracted fastening nut area, the image is processed by histogram equalization, the difference between the fastening nut and the background is increased, the segmentation is facilitated, and the processing result is as shown in fig. 6;
3.2 dividing the image into 2 clusters by using a K-Means clustering algorithm, presenting in a binary image form, adding a binary image criterion, and separating a fastening nut from an image background; the specific process is as follows:
(1) Setting the number of sample classes to 2, i.e. to C 1 、C 2 Selecting a clustering center for each category, and determining an initial clustering center point by adopting a centering algorithm;
(2) For each pixel point in the image, calculating Euclidean distance between the pixel point and the clustering centers, wherein E is a square error, x is a sample point, mu is i Is of class C i The mean vectors of (4) are divided into the classes corresponding to the clustering centers closest to the mean vectors according to the closest criterion, as shown in formula (3);
(3) Updating the clustering center: taking the mean values corresponding to all the objects in each category as the clustering center of the category to obtain new 2 central points;
(4) The algorithm termination condition was set to a maximum number of cycles of 10 and a precision threshold of 1.0. Comparing the difference values of the new 2 central points with the 2 central points in the step (1), if the difference values are smaller than the precision threshold, giving gray values of 0 or 255 to the pixel points according to different categories, and outputting an image; and if the precision threshold value is larger than the precision threshold value, continuing the steps (2) to (4) until the loop meets the termination condition.
(5) Carrying out gray level judgment, and outputting a final result after inverting the gray level of the image if the number of pixels occupied by the target is less than that of pixels occupied by the background and the number of pixels with the gray level of 255 of the output image is more; otherwise, the final result is directly output, as shown in fig. 7.
3.3 processing the image using the opening operation of image morphology to eliminate the interference image point, as shown in FIG. 8.
4. Subway altitude control valve fastening nut is cut apart accurately based on boundary characteristics
And 4.1, performing limited region search on the segmented binary image by adopting a region growing method, removing background miscellaneous points and finishing accurate segmentation of the fastening nut. The specific process is as follows:
(1) Selecting seeds: firstly, setting a rectangular region where region growing seeds may appear, wherein the left boundary of the rectangle is the left boundary of an image, and the horizontal central line is the horizontal central line of the image; the horizontal width of the rectangle is 85% of the image, and the length of the screw in the image is approximate; the rectangular vertical height is 60% of the ROI image, approximately the nut height. 10000 points are randomly generated in the region, the pixel value of the 10000 points is judged, and 100 nonzero pixel points are selected as seeds for starting region growth.
(2) And (3) area growth: searching four-neighborhood pixel points of the seeds, if the gray level of the pixel point is 255 and the pixel point is not searched, combining the pixel points, and marking the pixel points as searched. And when the searching of the four neighborhood points is finished, setting newly combined pixels as new seeds according to the combination sequence, and continuing to fully perform the region growing process until all the pixel points are traversed.
(3) Generating a result image: and accumulating the generated 100 region growing result images, and setting a non-zero pixel value in the accumulated result image to be 255, so that a binary image containing all nut profiles and simultaneously removing complex background interference can be obtained, and the result is shown in fig. 9.
4.2 calculating the horizontal direction projection and the vertical direction projection of the image, and further determining the area where the nut is located according to the boundary characteristics represented by the projection curve. In horizontal direction projection, dividing an original image by taking a horizontal straight line passing through a longitudinal coordinate of a peak point as a center, and dividing a horizontal line by taking a half position of the height of each of an upper image and a lower image to set the horizontal line as an upper boundary and a lower boundary of the ROI respectively; in the vertical projection, the first valley appearing on the right side of the peak is taken as the ROI left boundary, the first non-zero value appearing on the right side of the image is taken as the ROI right boundary, and the horizontal and vertical projections are shown in fig. 10 (a), (b), respectively.
And accurately extracting the region where the nut is located according to the upper, lower, left and right boundaries of the ROI, returning ROI rectangular information by the algorithm as shown in the figure 11 according to the nut positioning result, and marking in the original image.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be considered to be within the technical scope of the present invention, and the technical solutions and the inventive concepts thereof according to the present invention should be equivalent or changed within the scope of the present invention.
Claims (3)
1. A subway height adjusting valve fastening nut positioning method based on machine learning is characterized by comprising the following steps:
s1, acquiring image information of the bottom of the whole subway train, and cutting the image according to different carriages to obtain a high-definition color image of the bottom of the whole subway train, which contains a height adjusting valve component;
s2, establishing a deep neural network classifier to position the height adjusting valve component; establishing a support vector machine classifier to position a position area of a fastening nut in a height adjusting valve component;
s3, primarily segmenting the positioned fastening nut image by using a K-Means clustering algorithm;
s4, accurately dividing the fastening nut of the subway height adjusting valve according to a boundary characteristic method: performing limited area searching on the preliminarily segmented binary image by adopting an area growing method to remove background miscellaneous points, and accurately extracting a fastening nut;
s4 specifically comprises the following steps: calculating horizontal direction projection and vertical direction projection of the ROI area, further determining the area where the nut is located according to boundary characteristics represented by a projection curve, dividing an original image by taking a horizontal straight line passing through a peak point vertical coordinate as a center in the horizontal direction projection, taking a half position of the respective height of an upper image and a lower image to define a horizontal line, and setting the horizontal line as an upper boundary and a lower boundary of the ROI respectively; in the vertical direction projection, the first trough appearing on the right side of the peak value is recorded as a ROI left boundary, the first non-zero value appearing on the right side of the image is recorded as a ROI right boundary, and accurate extraction of the region where the nut is located is completed.
2. The machine learning-based subway height adjusting valve fastening nut positioning method according to claim 1, further characterized by comprising: the S2 specifically adopts the following mode:
a. generating a sample library of height adjusting valve components, wherein positive samples are images only containing the height adjusting valve components, and negative samples are images of other components not containing the height adjusting valves;
b. training the sample by adopting a deep neural network algorithm, establishing a deep learning classifier, and integrally positioning the height adjusting valve component;
c. generating a sample library of the fastening nuts in the height adjusting valve component, wherein positive samples are images containing the fastening nuts, and negative samples are images of other components not containing the fastening nuts;
d. and finishing the training of positive and negative samples by adopting a support vector machine, establishing a fastening nut classifier, and realizing the positioning of the fastening nut.
3. The machine learning-based method for positioning the fastening nut of the subway height adjusting valve according to claim 1, further characterized in that: s3 specifically adopts the following mode:
s31, performing contrast enhancement processing on the positioned fastening nut image, and increasing the contrast difference between the interested fastening nut area and the background area in the image by adopting a histogram equalization method;
s32, clustering and dividing the image by utilizing a K-Means clustering algorithm, and separating a fastening nut area from an image background area, wherein the following method is specifically adopted:
s321, setting the number of sample classifications to 2, namely, dividing into C 1 、C 2 Selecting a clustering center for each category, and determining an initial clustering center point by adopting a centralization algorithm;
s322, calculating the Euclidean distance between each pixel point and the clustering centers by adopting a formula (2) for each pixel point in the image, and classifying each pixel point to the class corresponding to the clustering center closest to the pixel point according to the closest criterion by adopting a formula (3);
where E is the square error, x is the sample point, μ i Is of class C i The mean vector of (2);
s323, updating the clustering center: taking the mean values corresponding to all the objects in each category as the clustering center of the category to obtain 2 updated center points;
s324, setting the termination condition of the algorithm as the maximum cycle number M and the precision threshold N, comparing the difference between the updated 2 central points and the 2 initial clustering central points in the S321, and if the difference is smaller than the precision threshold N, giving gray values of pixel points 0 or 255 according to different categories and outputting images; if the difference is larger than the precision threshold, continuing S322-S324 until the cycle meets the condition;
s325, carrying out gray level judgment on the image, and outputting a final result after inverting the gray level of the image if the number of pixels of the output image with the gray level value of 255 is more because the number of pixels of the ROI is less than that of pixels occupied by the background area; otherwise, directly outputting the final result;
and S33, processing the image by using the opening operation of the image morphology to eliminate the interference image points of the part.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811537193.9A CN109636794B (en) | 2018-12-14 | 2018-12-14 | Machine learning-based subway height adjusting valve fastening nut positioning method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811537193.9A CN109636794B (en) | 2018-12-14 | 2018-12-14 | Machine learning-based subway height adjusting valve fastening nut positioning method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109636794A CN109636794A (en) | 2019-04-16 |
| CN109636794B true CN109636794B (en) | 2023-02-28 |
Family
ID=66074261
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811537193.9A Active CN109636794B (en) | 2018-12-14 | 2018-12-14 | Machine learning-based subway height adjusting valve fastening nut positioning method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109636794B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112365481B (en) * | 2020-11-13 | 2021-06-18 | 哈尔滨市科佳通用机电股份有限公司 | Method for detecting bolt loss of cross beam assembly based on image processing |
| CN116012375B (en) * | 2023-03-22 | 2023-08-04 | 西南交通大学 | Method and system for detecting cotter pin defects of overhead contact system soft crossing suspension pulley |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008151471A1 (en) * | 2007-06-15 | 2008-12-18 | Tsinghua University | A robust precise eye positioning method in complicated background image |
| CN106485694A (en) * | 2016-09-11 | 2017-03-08 | 西南交通大学 | A kind of high ferro contact net double-jacket tube connector six-sided nut based on cascade classifier comes off defective mode detection method |
| CN108648233A (en) * | 2018-03-24 | 2018-10-12 | 北京工业大学 | A kind of target identification based on deep learning and crawl localization method |
-
2018
- 2018-12-14 CN CN201811537193.9A patent/CN109636794B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008151471A1 (en) * | 2007-06-15 | 2008-12-18 | Tsinghua University | A robust precise eye positioning method in complicated background image |
| CN106485694A (en) * | 2016-09-11 | 2017-03-08 | 西南交通大学 | A kind of high ferro contact net double-jacket tube connector six-sided nut based on cascade classifier comes off defective mode detection method |
| CN108648233A (en) * | 2018-03-24 | 2018-10-12 | 北京工业大学 | A kind of target identification based on deep learning and crawl localization method |
Non-Patent Citations (1)
| Title |
|---|
| 基于深度学习的脑部核磁图像分割算法;程俊龙等;《轻工科技》;20171231(第08期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109636794A (en) | 2019-04-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110334706B (en) | Image target identification method and device | |
| CN105261017B (en) | The method that image segmentation based on road surface constraint extracts pedestrian's area-of-interest | |
| CN109840521B (en) | Integrated license plate recognition method based on deep learning | |
| CN109101924B (en) | Machine learning-based road traffic sign identification method | |
| CN102902967B (en) | Method for positioning iris and pupil based on eye structure classification | |
| CN102509091B (en) | Airplane tail number recognition method | |
| CN105809184B (en) | Method for real-time vehicle identification and tracking and parking space occupation judgment suitable for gas station | |
| CN105844621A (en) | Method for detecting quality of printed matter | |
| CN105335743A (en) | Vehicle license plate recognition method | |
| CN113724231B (en) | Industrial defect detection method based on semantic segmentation and target detection fusion model | |
| Murugan et al. | A deep learning RCNN approach for vehicle recognition in traffic surveillance system | |
| CN108052904B (en) | Method and device for acquiring lane line | |
| CN111340855A (en) | Road moving target detection method based on track prediction | |
| CN108509950B (en) | Railway contact net support number plate detection and identification method based on probability feature weighted fusion | |
| CN109636794B (en) | Machine learning-based subway height adjusting valve fastening nut positioning method | |
| CN101526994A (en) | Fingerprint image segmentation method irrelevant to collecting device | |
| CN115100196B (en) | Method for evaluating derusting effect of stamping part based on image segmentation | |
| Yang et al. | An efficient method for vehicle model identification via logo recognition | |
| CN110706235A (en) | Far infrared pedestrian detection method based on two-stage cascade segmentation | |
| CN111476804A (en) | Method, device and equipment for efficiently segmenting carrier roller image and storage medium | |
| CN112668725A (en) | Metal hand basin defect target training method based on improved features | |
| CN109543498B (en) | Lane line detection method based on multitask network | |
| CN107247967B (en) | Vehicle window annual inspection mark detection method based on R-CNN | |
| CN105354547A (en) | Pedestrian detection method in combination of texture and color features | |
| Zhou et al. | Learning-based seat belt detection in image using salient gradient |
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 | ||
| CP03 | Change of name, title or address |
Address after: No.52a (1001), Beiyi West Road, Tiexi District, Shenyang, Liaoning Province, 110000 Patentee after: LIAONING DINGHAN QIHUI ELECTRONIC SYSTEM ENGINEERING Co.,Ltd. Address before: 110000 10th floor, building AC, Jingu project, no.52a, beiyixi Road, Tiexi District, Shenyang City, Liaoning Province Patentee before: LIAONING QUICKHIGH ELECTRONIC SYSTEM ENGINEERING CO.,LTD. |
|
| CP03 | Change of name, title or address |