CN108985170A - Transmission line of electricity hanger recognition methods based on Three image difference and deep learning - Google Patents
Transmission line of electricity hanger recognition methods based on Three image difference and deep learning Download PDFInfo
- Publication number
- CN108985170A CN108985170A CN201810621571.5A CN201810621571A CN108985170A CN 108985170 A CN108985170 A CN 108985170A CN 201810621571 A CN201810621571 A CN 201810621571A CN 108985170 A CN108985170 A CN 108985170A
- Authority
- CN
- China
- Prior art keywords
- image
- hanger
- deep learning
- difference
- transmission line
- 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.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The transmission line of electricity hanger recognition methods based on Three image difference and deep learning that the invention discloses a kind of, belongs to technical field of image processing, wherein the described method comprises the following steps: (1) input picture；(2) image is pre-processed；(3) three frame images are matched using HOG characteristics algorithm；(4) the three frame images after step (3) match are subjected to difference；(5) the doubtful hanger ROI region of original image is extracted and is detected；(6) yolov2 deep learning algorithm train classification models are used, image classification are carried out to variation ROI region by training pattern, it is determined whether be hanger；(7) hanger is screened according to the threshold value of classification；(8) image is exported, and will confirm that it is suspension object area rectangle mark.Overcome the deficiencies in the prior art of the present invention greatly can effectively avoid illumination, camera lens shake, the brought influence of scene changes, reduce rate of false alarm and can satisfy the requirement of detection processing speed.
Description
Technical field
The transmission line of electricity hanger recognition methods based on Three image difference and deep learning that the present invention relates to a kind of, belongs to figure
As processing technology field.
Background technique
To in electric transmission line channel hanger identification use method generally have two frame difference methods, template matching method,
Hough straight line converts algorithm etc..According to the positional relationship between high-voltage line and hanger, the pixel characteristic of foreign matter type etc. is carried out
Identification.Wherein two frame difference methods and the algorithm that combines of hough straight line transformation are one of most common methods, difficult point be how
Establish the adaptable Model of Target Recognition of scene；The target identification of doubtful hanger is determined based on the difference of front and back two field pictures
Then method converts algorithm by hough straight line and searches high-voltage line, carries out the location confirmation of line and doubtful hanger；Based on mould
The matched method of plate obtains doubtful suspension object area, using pre- first by image preprocessing and front and back two field pictures difference
Different high-voltage line hanger target templates are first set, are then matched using template and area image to be measured.
But the above method because caused by external environment background shaking, Changes in weather, influence of noise and target itself it is special
The reasons such as the diversity of sign cause following disadvantage；
1, main picture to be processed is the high-voltage line short distance scene for unmanned plane shooting, and high-voltage line all near linears are answered
High with hough transformation electric wire discrimination, for the remote scene of electric transmission line channel, high-voltage line is nearly all curve, application
Hough straight line transformation high-voltage line verification and measurement ratio is low, can not be by examining the spatial position of high-voltage line and hanger remove identification hanger；
2, the target identification method characteristic effect based on template matching is larger by weather and illumination effect, and rate of false alarm is very
It is high；
3, the inter-frame difference result of two frame difference methods is it is possible that target lap is not easy to be detected in two field pictures
Out the case where, moving target center are easy to appear cavitation.
Summary of the invention
It is identified the purpose of the present invention is to provide a kind of based on the transmission line of electricity hanger of Three image difference and deep learning
Method, overcome the deficiencies in the prior art greatly can effectively avoid illumination, camera lens shake, the brought influence of scene changes,
It reduces rate of false alarm and can satisfy the requirement of detection processing speed.
To solve the above problems, the present invention provides a kind of transmission line of electricity hanger based on Three image difference and deep learning
Recognition methods, wherein, it the described method comprises the following steps:
(1) input picture；
(2) image is pre-processed；
(3) three frame images are matched using improved HOG characteristics algorithm；
(4) the three frame images after step (3) match are subjected to difference；
(5) the doubtful hanger ROI region of original image is extracted and is detected
(6) yolov2 deep learning algorithm train classification models are used, figure is carried out to variation ROI region by training pattern
As classification, it is determined whether be hanger；
(7) hanger is screened according to the threshold value of classification；
(8) image is exported, and will confirm that it is suspension object area rectangle mark.
As a preferred solution, image is carried out described in step (2) pre-processing the image progress referred to acquisition
Space filtering, smoothing denoising, enhancing, filtering, gray processing processing including image.
As a preferred solution, improved HOG characteristics algorithm described in step (3) refers to, image is transformed to pole
Under coordinate, piecemeal is carried out to image by rotation fixed angle, obtains each piece of gradient distribution histogram；Pass through gradient histogram
Figure carries out characteristic matching to three frame images.
Original HOG characteristics algorithm divides the image into multiple small connected regions, then counts the gradient of each connected region
Distribution histogram finally combines these histograms with shelf space information according to spatial order, in order to guarantee illumination not
Denaturation, algorithm normalize these local histograms in the comparing in larger scope of image, and the present invention is special to original HOG
Sign algorithm makes improvement, and improved algorithm has modified the partitioned mode of former algorithm, image is transformed under polar coordinates, passes through
It rotates fixed angle and piecemeal is carried out to image, obtain each piece of gradient distribution histogram；By histogram of gradients to three frame figures
As carrying out characteristic matching；
Using improved hog characteristics algorithm, can either geometry to image local and illumination variation keep constant well
Property increases the invariance to image rotation again, allows for images match.
As a preferred solution, the progress of three frame images described in step (4) difference refers to is obtained using Three image difference
Obtain region of variation.
As a preferred solution, the Three image difference is that present frame and upper frame are done calculus of differences, and next frame is again
Calculus of differences is done with present frame, then the result of this calculus of differences twice carries out logic and operation operation again, finally by setting
Good threshold value is split, and extracts variation targets.
The region of variation obtained at this time is more accurate than the region of variation that general two frame differences method obtains, can be in certain journey
" slur " phenomenon that two frame differential methods are eliminated on degree, also plays inhibiting effect to a certain extent for noise.
As a preferred solution, step (6) classification method is to restore all ROI region coordinates detected
Cutting feeding yolo classifier is carried out to original image to classify, and is finally screened hanger according to the threshold value of classification.
As a preferred solution, the yolo classifier is to be instructed using yolo algorithm to training sample set
Practice, obtains deep learning disaggregated model.
Using yolov2 deep learning algorithm train classification models, image is carried out to variation ROI region by training pattern
Classification, it is determined whether be hanger, yolo algorithm is end-to-end (End-to-End), and the deep learning without region nomination is calculated
Method, it has very fast recognition speed and a higher discrimination, and fast 1000 times of speed ratio R-CNN, 100 times faster than Fast RCNN,
It is currently the deep learning algorithm for being widely used in real-time objects detection in mobile end equipment, YOLOv2 has used one
New sorter network has used more 3*3 convolution kernel as characteristic extraction part, port number after the operation of pondization each time
It is double.The thought of network in network is used for reference, global average pond (the global average of Web vector graphic
Pooling), the convolution kernel of 1*1 is placed between the convolution kernel of 3*3, is used to compressive features, while also using batch
Normalization carrys out stable model training.This method uses yolov2 as Image Classifier, greatly improves the speed of classification
Degree and accuracy.
The invention has the advantages that:
The present invention well blends traditional machine learning algorithm and deep learning algorithm, excellent using respective algorithm
Point overcomes the defect of themselves, to reach more accurately recognition effect, illumination, camera lens shake, scene can effectively be avoided to become
It is influenced brought by changing, there is stronger rotational invariance, anti-distortion ability and good robustness and quick computing capability,
Target rate of false alarm can be effectively reduced as classifier using yolov2, target can be effectively reduced using Three image difference fail to report；Together
Shi Caiyong yolo deep learning algorithm improves deep learning module arithmetic speed, avoids traditional algorithm in conjunction with depth analysis
The slow problem of the speed of service afterwards, accelerates assorting process, can satisfy the requirement of detection processing speed.
Detailed description of the invention
Fig. 1 is workflow schematic diagram of the present invention；
Fig. 2 is that Three image difference of the present invention realizes schematic diagram；
Fig. 3 is yolo sorter network structure used in the present invention.
Specific embodiment
Embodiment 1:
Referring to Figure of description, the present invention provides a kind of transmission line of electricity hanger based on Three image difference and deep learning
Recognition methods, wherein, it the described method comprises the following steps:
(1) input picture；
(2) image is pre-processed；
(3) three frame images are matched using improved HOG characteristics algorithm；
(4) the three frame images after step (3) match are subjected to difference；
(5) the doubtful hanger ROI region of original image is extracted and is detected
(6) yolov2 deep learning algorithm train classification models are used, figure is carried out to variation ROI region by training pattern
As classification, judge whether to be hanger；
(7) hanger is screened according to the threshold value of classification；
(8) image is exported, and will confirm that it is suspension object area rectangle mark；
(9) judge that non-hanger is then rejected.
As a preferred solution, image is carried out described in step (2) pre-processing the image progress referred to acquisition
Space filtering, smoothing denoising, enhancing, filtering, gray processing processing including image.
As a preferred solution, improved HOG characteristics algorithm described in step (3) refers to, image is transformed to pole
Under coordinate, piecemeal is carried out to image by rotation fixed angle, obtains each piece of gradient distribution histogram；Pass through gradient histogram
Figure carries out characteristic matching to three frame images.
Original HOG characteristics algorithm divides the image into multiple small connected regions, then counts the gradient of each connected region
Distribution histogram finally combines these histograms with shelf space information according to spatial order, in order to guarantee illumination not
Denaturation, algorithm normalize these local histograms in the comparing in larger scope of image, and the present invention is special to original HOG
Sign algorithm makes improvement, and improved algorithm has modified the partitioned mode of former algorithm, image is transformed under polar coordinates, passes through
It rotates fixed angle and piecemeal is carried out to image, obtain each piece of gradient distribution histogram；By histogram of gradients to three frame figures
As carrying out characteristic matching；
Using improved hog characteristics algorithm, can either geometry to image local and illumination variation keep constant well
Property increases the invariance to image rotation again, allows for images match.
As a preferred solution, the progress of three frame images described in step (4) difference refers to is obtained using Three image difference
Obtain region of variation.
As a preferred solution, the Three image difference is that present frame and upper frame are done calculus of differences, and next frame is again
Calculus of differences is done with present frame, then the result of this calculus of differences twice carries out logic and operation operation again, finally by setting
Good threshold value is split, and extracts variation targets.
The region of variation obtained at this time is more accurate than the region of variation that general two frame differences method obtains, can be in certain journey
" slur " phenomenon that two frame differential methods are eliminated on degree, also plays inhibiting effect to a certain extent for noise.
As a preferred solution, step (6) classification method is to restore all ROI region coordinates detected
Cutting feeding yolo classifier is carried out to original image to classify, and is finally screened hanger according to the threshold value of classification.
As a preferred solution, the yolo classifier is to be instructed using yolo algorithm to training sample set
Practice, obtains deep learning disaggregated model.
Using yolov2 deep learning algorithm train classification models, image is carried out to variation ROI region by training pattern
Classification, it is determined whether be hanger, yolo algorithm is end-to-end (End-to-End), and the deep learning without region nomination is calculated
Method, it has very fast recognition speed and a higher discrimination, and fast 1000 times of speed ratio R-CNN, 100 times faster than Fast RCNN,
It is currently the deep learning algorithm for being widely used in real-time objects detection in mobile end equipment, YOLOv2 has used one
New sorter network has used more 3*3 convolution kernel as characteristic extraction part, port number after the operation of pondization each time
It is double.The thought of network in network is used for reference, global average pond (the global average of Web vector graphic
Pooling), the convolution kernel of 1*1 is placed between the convolution kernel of 3*3, is used to compressive features, while also using batch
Normalization carrys out stable model training.This method uses yolov2 as Image Classifier, greatly improves the speed of classification
Degree and accuracy.
The process of depth training: the transmission line of electricity image pattern collection of different scenes each set time is collected, feature is passed through
Matching and Three image difference obtain a large amount of difference sample set, carry out positive negative sample to difference sample set by the way of artificial
Mark (wherein, open, and negative sample 10000 is opened, and is trained using yolo algorithm to training sample set, obtains depth by positive sample 5000
Spend learning classification model.
The present invention well blends traditional machine learning algorithm and deep learning algorithm, excellent using respective algorithm
Point overcomes the defect of themselves, to reach more accurately recognition effect, illumination, camera lens shake, scene can effectively be avoided to become
It is influenced brought by changing, there is stronger rotational invariance, anti-distortion ability and good robustness and quick computing capability,
Target rate of false alarm can be effectively reduced as classifier using yolov2, target can be effectively reduced using Three image difference fail to report；Together
Shi Caiyong yolo deep learning algorithm improves deep learning module arithmetic speed, avoids traditional algorithm in conjunction with depth analysis
The slow problem of the speed of service afterwards, accelerates assorting process, can satisfy the requirement of detection processing speed.
The above description is only an embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (7)
1. a kind of transmission line of electricity hanger recognition methods based on Three image difference and deep learning, it is characterised in that, the side
Method the following steps are included:
(1) input picture；
(2) image is pre-processed；
(3) three frame images are matched using improved HOG characteristics algorithm；
(4) the three frame images after step (3) match are subjected to difference；
(5) the doubtful hanger ROI region of original image is extracted and is detected
(6) yolov2 deep learning algorithm train classification models are used, image point is carried out to variation ROI region by training pattern
Class, it is determined whether be hanger；
(7) hanger is screened according to the threshold value of classification；
(8) image is exported, and will confirm that it is suspension object area rectangle mark.
2. a kind of transmission line of electricity hanger identification side based on Three image difference and deep learning according to claim 1
Method, which is characterized in that image is carried out described in step (2) to pre-process the image progress space filtering referred to acquisition, including
The smoothing denoising of image, enhancing, filtering, gray processing processing.
3. a kind of transmission line of electricity hanger identification side based on Three image difference and deep learning according to claim 1
Method, which is characterized in that improved HOG characteristics algorithm described in step (3) refers to, image is transformed under polar coordinates, passes through rotation
Turn fixed angle and piecemeal is carried out to image, obtains each piece of gradient distribution histogram；By histogram of gradients to three frame images
Carry out characteristic matching.
4. a kind of transmission line of electricity hanger identification side based on Three image difference and deep learning according to claim 1
Method, it is characterised in that, the progress of three frame images described in step (4) difference, which refers to, utilizes Three image difference to obtain region of variation.
5. a kind of transmission line of electricity hanger identification side based on Three image difference and deep learning according to claim 4
Method, which is characterized in that the Three image difference is that present frame and upper frame are done calculus of differences, and next frame does difference with present frame again
Operation, then the result of this calculus of differences twice carries out logic and operation operation again, is divided finally by the threshold value set
It cuts, extracts variation targets.
6. a kind of transmission line of electricity hanger identification side based on Three image difference and deep learning according to claim 1
Method, which is characterized in that step (6) classification method be by all ROI region coordinates detected be restored to original image into
Row cuts feeding yolo classifier and classifies, and is finally screened hanger according to the threshold value of classification.
7. a kind of transmission line of electricity hanger identification side based on Three image difference and deep learning according to claim 6
Method, which is characterized in that the yolo classifier is to be trained using yolo algorithm to training sample set, obtains deep learning
Disaggregated model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810621571.5A CN108985170A (en) | 2018-06-15 | 2018-06-15 | Transmission line of electricity hanger recognition methods based on Three image difference and deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810621571.5A CN108985170A (en) | 2018-06-15 | 2018-06-15 | Transmission line of electricity hanger recognition methods based on Three image difference and deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108985170A true CN108985170A (en) | 2018-12-11 |
Family
ID=64541355
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810621571.5A Withdrawn CN108985170A (en) | 2018-06-15 | 2018-06-15 | Transmission line of electricity hanger recognition methods based on Three image difference and deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108985170A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109813722A (en) * | 2019-03-12 | 2019-05-28 | 武汉武大卓越科技有限责任公司 | Contact Net's Suspension Chord defect inspection method |
| CN109886935A (en) * | 2019-01-28 | 2019-06-14 | 南京威翔科技有限公司 | A kind of road face foreign matter detecting method based on deep learning |
| CN110110684A (en) * | 2019-05-14 | 2019-08-09 | 深圳供电局有限公司 | Foreign object identification method and device for power transmission line equipment and computer equipment |
| CN110119730A (en) * | 2019-06-03 | 2019-08-13 | 齐鲁工业大学 | A kind of monitor video processing method, system, terminal and storage medium |
| CN110796682A (en) * | 2019-09-25 | 2020-02-14 | 北京成峰科技有限公司 | Detection and identification method and detection and identification system for moving target |
| CN110956614A (en) * | 2019-11-11 | 2020-04-03 | 国网山东省电力公司电力科学研究院 | Method and device for detecting foreign objects in conductors based on iterative search and projection method |
| CN111223129A (en) * | 2020-01-10 | 2020-06-02 | 深圳中兴网信科技有限公司 | Detection method, detection device, monitoring equipment and computer readable storage medium |
| CN111738211A (en) * | 2020-07-17 | 2020-10-02 | 浙江大学 | PTZ camera moving target detection and recognition method based on dynamic background compensation and deep learning |
| CN112364865A (en) * | 2020-11-12 | 2021-02-12 | 郑州大学 | Method for detecting small moving target in complex scene |
| CN112465781A (en) * | 2020-11-26 | 2021-03-09 | 华能通辽风力发电有限公司 | Method for identifying defects of main parts of wind turbine generator based on video |
| CN112909824A (en) * | 2021-03-24 | 2021-06-04 | 南方电网电力科技股份有限公司 | Method and device for identifying suspended foreign matters of power transmission line |
| CN113076800A (en) * | 2021-03-03 | 2021-07-06 | 惠州市博实结科技有限公司 | Road sign board detection method and device |
| CN113486865A (en) * | 2021-09-03 | 2021-10-08 | 国网江西省电力有限公司电力科学研究院 | Power transmission line suspended foreign object target detection method based on deep learning |
| CN113627299A (en) * | 2021-07-30 | 2021-11-09 | 广东电网有限责任公司 | Intelligent wire floater identification method and device based on deep learning |
-
2018
- 2018-06-15 CN CN201810621571.5A patent/CN108985170A/en not_active Withdrawn
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109886935A (en) * | 2019-01-28 | 2019-06-14 | 南京威翔科技有限公司 | A kind of road face foreign matter detecting method based on deep learning |
| CN109813722B (en) * | 2019-03-12 | 2021-09-24 | 武汉光谷卓越科技股份有限公司 | Contact net dropper defect detection method |
| CN109813722A (en) * | 2019-03-12 | 2019-05-28 | 武汉武大卓越科技有限责任公司 | Contact Net's Suspension Chord defect inspection method |
| CN110110684A (en) * | 2019-05-14 | 2019-08-09 | 深圳供电局有限公司 | Foreign object identification method and device for power transmission line equipment and computer equipment |
| CN110119730A (en) * | 2019-06-03 | 2019-08-13 | 齐鲁工业大学 | A kind of monitor video processing method, system, terminal and storage medium |
| CN110796682A (en) * | 2019-09-25 | 2020-02-14 | 北京成峰科技有限公司 | Detection and identification method and detection and identification system for moving target |
| CN110956614A (en) * | 2019-11-11 | 2020-04-03 | 国网山东省电力公司电力科学研究院 | Method and device for detecting foreign objects in conductors based on iterative search and projection method |
| CN110956614B (en) * | 2019-11-11 | 2023-04-07 | 国网山东省电力公司电力科学研究院 | Ground wire foreign matter detection method and device based on iterative search and projection method |
| CN111223129A (en) * | 2020-01-10 | 2020-06-02 | 深圳中兴网信科技有限公司 | Detection method, detection device, monitoring equipment and computer readable storage medium |
| CN111738211A (en) * | 2020-07-17 | 2020-10-02 | 浙江大学 | PTZ camera moving target detection and recognition method based on dynamic background compensation and deep learning |
| CN111738211B (en) * | 2020-07-17 | 2023-12-19 | 浙江大学 | PTZ camera moving object detection and recognition method based on dynamic background compensation and deep learning |
| CN112364865B (en) * | 2020-11-12 | 2022-09-23 | 郑州大学 | A detection method for moving small objects in complex scenes |
| CN112364865A (en) * | 2020-11-12 | 2021-02-12 | 郑州大学 | Method for detecting small moving target in complex scene |
| CN112465781A (en) * | 2020-11-26 | 2021-03-09 | 华能通辽风力发电有限公司 | Method for identifying defects of main parts of wind turbine generator based on video |
| CN113076800A (en) * | 2021-03-03 | 2021-07-06 | 惠州市博实结科技有限公司 | Road sign board detection method and device |
| CN112909824A (en) * | 2021-03-24 | 2021-06-04 | 南方电网电力科技股份有限公司 | Method and device for identifying suspended foreign matters of power transmission line |
| CN113627299A (en) * | 2021-07-30 | 2021-11-09 | 广东电网有限责任公司 | Intelligent wire floater identification method and device based on deep learning |
| CN113627299B (en) * | 2021-07-30 | 2024-04-09 | 广东电网有限责任公司 | Wire floater intelligent recognition method and device based on deep learning |
| CN113486865A (en) * | 2021-09-03 | 2021-10-08 | 国网江西省电力有限公司电力科学研究院 | Power transmission line suspended foreign object target detection method based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108985170A (en) | Transmission line of electricity hanger recognition methods based on Three image difference and deep learning | |
| CN106093066B (en) | A kind of magnetic tile surface defect detection method based on improved machine vision attention mechanism | |
| CN106709436B (en) | Track traffic panoramic monitoring-oriented cross-camera suspicious pedestrian target tracking system | |
| CN107133973B (en) | A ship detection method in a bridge anti-collision system | |
| CN102542289B (en) | Pedestrian volume statistical method based on plurality of Gaussian counting models | |
| CN107909081B (en) | A fast acquisition and fast calibration method for image datasets in deep learning | |
| CN103034852B (en) | The detection method of particular color pedestrian under Still Camera scene | |
| CN109101924A (en) | A kind of pavement marking recognition methods based on machine learning | |
| CN111310760B (en) | Oracle Bone Inscription Text Detection Method Combining Local Prior Features and Deep Convolution Features | |
| CN108921201B (en) | Dam defect identification and classification method based on feature combination and CNN | |
| CN107389701A (en) | A kind of PCB visual defects automatic checkout system and method based on image | |
| CN107316058A (en) | Improve the method for target detection performance by improving target classification and positional accuracy | |
| CN104166841A (en) | Rapid detection identification method for specified pedestrian or vehicle in video monitoring network | |
| CN107491720A (en) | A kind of model recognizing method based on modified convolutional neural networks | |
| CN104715238A (en) | Pedestrian detection method based on multi-feature fusion | |
| CN101853397A (en) | A bionic face detection method based on human visual characteristics | |
| CN106780504B (en) | It is a kind of based on distance keep level set beef image in carry on the back long flesh automatic division method | |
| CN109190456A (en) | Pedestrian detection method is overlooked based on the multiple features fusion of converging channels feature and gray level co-occurrence matrixes | |
| CN113221603A (en) | Method and device for detecting shielding of monitoring equipment by foreign matters | |
| Su et al. | A new local-main-gradient-orientation HOG and contour differences based algorithm for object classification | |
| CN103116890A (en) | Video image based intelligent searching and matching method | |
| CN114863464A (en) | Second-order identification method for PID drawing picture information | |
| CN106650824B (en) | Moving object classification method based on support vector machines | |
| CN107679565A (en) | Great-jump-forward sliding window accelerates detection method of license plate | |
| CN113139946A (en) | Shirt stain positioning device based on vision |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20181211 |