CN108764154A - A kind of garbage on water recognition methods based on multiple features machine learning - Google Patents

A kind of garbage on water recognition methods based on multiple features machine learning Download PDF

Info

Publication number
CN108764154A
CN108764154A CN201810537983.0A CN201810537983A CN108764154A CN 108764154 A CN108764154 A CN 108764154A CN 201810537983 A CN201810537983 A CN 201810537983A CN 108764154 A CN108764154 A CN 108764154A
Authority
CN
China
Prior art keywords
data
feature vector
machine learning
edge
counter
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
Application number
CN201810537983.0A
Other languages
Chinese (zh)
Other versions
CN108764154B (en
Inventor
吕霞付
李森浩
罗萍
程啟忠
林政�
陈勇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chongqing University of Post and Telecommunications
Original Assignee
Chongqing University of Post and Telecommunications
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Chongqing University of Post and Telecommunications filed Critical Chongqing University of Post and Telecommunications
Priority to CN201810537983.0A priority Critical patent/CN108764154B/en
Publication of CN108764154A publication Critical patent/CN108764154A/en
Application granted granted Critical
Publication of CN108764154B publication Critical patent/CN108764154B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • 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
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/2163—Partitioning the feature space
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/285—Selection of pattern recognition techniques, e.g. of classifiers in a multi-classifier system
    • 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
    • G06V10/34—Smoothing or thinning of the pattern; Morphological operations; Skeletonisation
    • 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
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Evolutionary Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Image Analysis (AREA)

Abstract

The present invention relates to a kind of garbage on water recognition methods based on multiple features machine learning, belong to waterborne target identification judgement field, S1：Identification model initializes, and generates the Database Systems that content is empty；S2：The positive example training data of label and counter-example training data are subjected to initialization pretreatment, normalized, the corresponding feature vector of extraction, bring the feature vector of the feature vector of positive example image and counter-example image into system model training, corresponding knowledge base is generated, is then tested to knowledge base using test data；S3：Knowledge base is brought into identifying system, water surface status data is read, acquires and upload image data；S4：Host computer pre-processes image data, extracts feature vector, and bring feature vector into machine learning model, reads corresponding knowledge base, the result that judges, and will determine that new feature vector is exported in the form of flag data.Energy intelligent recognition of the invention and pumped surface rubbish, meet running environment real-time.

Description

A kind of garbage on water recognition methods based on multiple features machine learning
Technical field
The invention belongs to waterborne target identification judgement fields, and in particular to a kind of water surface rubbish based on multiple features machine learning Rubbish recognition methods.
Background technology
In recent years, China's economy develops rapidly, water pollution problems getting worse, and river water body is dirty by industry, sanitary wastewater Dye, river surface adrift a large amount of disposed wastes, such as plastic bottle, polybag etc. some be difficult to the rubbish degraded not only to the water surface Ill effect caused by landscape more causes serious influence to the production of the people and life, serious to have jeopardized the people Health, how effectively to lock, to clear up pollution sources extremely urgent.
Currently, the water surface is kept a public place clean and water quality monitoring relies primarily on artificial progress, water surface area is big, and aquatic environment is complicated, manually Junk-free and water quality sampling are heavy, inefficiency a job.And if it is toxic contaminants, add water surface operation With certain danger, artificial cleaning can threaten to the life security of personnel.
For some negative impacts of garbage on water, it is proposed many processing floater rubbish both at home and abroad at present Some measures, but due to the influence of some artificial origins, garbage on water is still difficult to handle completely.Due to some objective originals Cause, garbage on water neither one fix apparent increment trend, tend to random delta state completely.If using the people of timing always To salvage mode, not only waste of manpower, material resources, financial resources, while it may also result in secondary pollution.
Invention content
In view of this, energy automatic identification floater rubbish and the robot of cleaning are designed for this phenomenon, it will be fine Solve the problem of this this invention research floater rubbish identification (water bottle, carton) emphatically.
The purpose of the present invention is to provide a kind of reliability is higher, the water surface rubbish of the stronger multiple features machine learning of real-time Rubbish knows method for distinguishing.
In order to achieve the above objectives, the present invention provides the following technical solutions：
A kind of garbage on water recognition methods based on multiple features machine learning, includes the following steps：
S1：Identification model initializes, and generates the Database Systems that a content is empty, Database Systems include label Positive example training data, counter-example training data, positive example test data, the ginseng in counter-example test data and machine learning model Number；
S2：By the positive example training data of label and counter-example training data carry out initialization pretreatment, normalized, Corresponding feature vector is extracted, the feature vector of the feature vector of positive example image and counter-example image is brought into system model and is instructed Practice, generates corresponding knowledge base, then tested to knowledge base using test data, until knowledge base is met the requirements；
S3：Corresponding knowledge base is brought into identifying system, water surface situation number is dynamically read by boat-carrying camera According to, then the transmission of periodic acquisition camera image data, and be transmitted to host computer；
S4：For host computer when carrying out image data with above-mentioned modeling after identical pretreatment, extraction is identical with when modeling Feature vector, and bring feature vector into machine learning model, the corresponding knowledge base of machine learning model reading, to new feature Vector judges, and the result that will determine that is exported in the form of flag data.
Further, in step S1, the acquisition for positive example data and counter-example data, the difference shot using boat-carrying camera Scene (is specifically divided into the difference of intensity of illumination, whether the variation of solar azimuth, the difference of water environment, foreground target are rubbish Or be chaff interferent) video, corresponding linear list is established to map to different environment, maps the precision of linear list to knowledge The accuracy of library model has large effect, then intercepts positive example and counter-example data of the corresponding frame as acquisition in video.
Further, in step S2, the pretreatment to data image.Under water surface complex detection environment, there are water ripples, water The interference of other objects of face.It is confused to easily cause detection target and background, uses improved median filtering algorithm first Then projecting edge can be very good prominent foreground target using saliency extraction algorithm, while need to simplify the algorithm, Optimize run time.After prominent foreground target, need that separating treatment, common algorithm have threshold in the background by foreground target Value separation, multi-threshold separation, but above method adaptability robustness is poor, otsu algorithms can dynamically propose threshold value, point Binary image is obtained from foreground target and target context, then morphology operations is used to repair foreground subject edges.For one The interference of a little small foreground targets, this patent propose a kind of separate mode based on connected domain size.
Further, in step S2, the extraction to target signature.After the completion of image preprocessing, by extracting feature, come real Now to two classification of floating material and interference.Different targets is described by different features, be pattern-recognition it is most basic according to According to.This patent is extracted what description of the RGB color model based on water surface foreground target was combined with edge square theoretical description Feature generates feature vector.RGB color model：
Y=a | R-RA|+b|G-GA|+c|B-BA| (1)
R in formula (1)A、GA、BAIndicate that water surface background be averaged RGB color respectively, the face of R, G, B expression current pixel point Color, a, b, c indicate adjustable linear magnification.
Edge square theoretical description submodel：
ai=Li+Li+2-2Li+1 (2)
Formula (2) indicates neighboring edge point to the second-order difference equation of edge contour centre distance, wherein LiIndicate edge wheel Wide center to edge distance, i take 0 arrive Cnt-2, Cnt be peripheral edge pixel quantity, work as ai<Threshold, Threshold is predetermined threshold, needs to test adjusting according to actual conditions, then it is assumed that aiEqual to 0, then the change of target edge is reflected Change smaller.
Further, in step S3, to meet system real time requirement, according to the speed V that unmanned hull travels, sampling period T should meet relational expression：
TV≤0.2 (3)
Further,, can be in medium filtering in order to preferably protrude foreground target in step S2, color is close to the water surface The pixel of average color assigns lighter weight, and the farther pixel of color distance water surface mean pixel Euclidean distance is assigned Heavier weight；Edge repair morphology operations repair edge by the way of first expanding post-etching, for filtering core therein, The filtering core of dynamic size is used according to the distance of image.
By adopting the above-described technical solution, the present invention has the advantage that：
The present invention models the target's feature-extraction method being combined using object edge detection modeling with color of object, relatively In traditional single type edge feature or pass through precision higher for the feature that region area changes, better adaptability more can The scene of outdoor garbage on water identification is adapted to, practicability is more preferable.Meanwhile in terms of pattern-recognition by the way of machine learning, and The mode of non-expert's rule so that accuracy of identification further increases.Overall plan is more reasonable effective.
Description of the drawings
In order to keep the purpose of the present invention, technical solution and advantageous effect clearer, the present invention provides following attached drawing and carries out Explanation：
Fig. 1 is the flow chart of the rubbish identification described in the embodiment of the present invention；
Fig. 2 is that the image described in the embodiment of the present invention obtains and pretreatment filters flow chart；
Fig. 3 is the feature extraction figure described in the embodiment of the present invention.
Specific implementation mode
Below in conjunction with attached drawing, the preferred embodiment of the present invention is described in detail.
A kind of garbage on water recognition methods based on multiple features machine learning, includes the following steps shown in referring to Fig.1：
101, identification model initializes, and generates the Database Systems that a content is empty, Database Systems are by label Positive example training data, counter-example training data, positive example test data, counter-example test data, the parameter in machine learning model.
102, the positive example training data of label and counter-example training data are subjected to initialization pretreatment, at normalization It manages, extract corresponding feature vector, the feature vector of the feature vector of positive example image and counter-example image is brought into system model Training, generate corresponding knowledge base, then tested to knowledge base using test data, until knowledge base meet the requirements for Only.
103, corresponding knowledge base is brought into identifying system, water surface shape is dynamically then read by boat-carrying camera Condition data, the image data that then periodic acquisition camera is transmitted, and it is transmitted to host computer.
104, identical when extraction is with modeling when host computer carries out with above-mentioned modeling image data after identical pretreatment Feature vector, and bring feature vector into machine learning model, the corresponding knowledge base of machine learning model reading, to new feature Vector judges, and the result that will determine that is exported in the form of flag data.
As shown in Fig. 2, in a step 101, the acquisition for positive example data and counter-example data is shot using boat-carrying camera Different scenes (be specifically divided into the difference of intensity of illumination, the variation of solar azimuth, the difference of water environment, foreground target whether Be rubbish or be chaff interferent) video, corresponding linear list is established to map to different environment, maps the precision of linear list There is large effect to the accuracy of knowledge base model, then intercepts positive example and counter-example of the corresponding frame as acquisition in video Data.
In a step 102, to the pretreatment of data image.Under water surface complex detection environment, there are water ripples, the water surface its The interference of his object.It is confused to easily cause detection target and background, is protruded first using improved median filtering algorithm Then edge can be very good prominent foreground target using saliency extraction algorithm, while need to simplify the algorithm, optimize Run time.After prominent foreground target, need that separating treatment, common algorithm have threshold value point in the background by foreground target From, multi-threshold separation, but above method adaptability robustness is poor, otsu algorithms can dynamically propose threshold value, before separation Scape target and background target obtains binary image, then morphology operations is used to repair foreground subject edges.It is small for some The interference of foreground target, this patent propose a kind of separate mode based on connected domain size.
As shown in figure 3, in a step 102, the extraction to target signature.It is special by extracting after the completion of image preprocessing Sign, to realize two classification to floating material and interference.Different targets is described by different features, is pattern-recognition most base This foundation.This patent is extracted the description of the RGB color model based on water surface foreground target and the sub- phase of edge square theoretical description In conjunction with feature, generate feature vector.RGB color model：
Y=a | R-RA|+b|G-GA|+c|B-BA| (1)
R in formula (1)A、GA、BAIndicate that water surface background be averaged RGB color respectively, the face of R, G, B expression current pixel point Color, a, b, c indicate adjustable linear magnification.
Edge square theoretical description submodel：
ai=Li+Li+2-2Li+1 (2)
Formula (2) indicates neighboring edge point to the second-order difference equation of edge contour centre distance, wherein LiIndicate edge wheel Wide center to edge distance, i take 0 arrive Cnt-2, Cnt be peripheral edge pixel quantity, work as ai<Threshold, Threshold is predetermined threshold, needs to test adjusting according to actual conditions, then it is assumed that aiEqual to 0, then the change of target edge is reflected Change smaller.
It in a step 102, can be in medium filtering, color be average close to the water surface in order to preferably protrude foreground target The pixel of color assigns lighter weight, the farther pixel of color distance water surface mean pixel Euclidean distance is assigned heavier Weight；Edge repair morphology operations repair edge by the way of first expanding post-etching, for filtering core therein, according to The distance of image uses the filtering core of dynamic size.
The invention discloses a kind of, and the garbage on water based on multiple features machine learning knows method for distinguishing, can identify and work as in real time Whether preceding water surface foreground target is rubbish either other interfering objects, and more completes the judgement to interfering object, judges it Whether it is barrier.To send out corresponding command signal, controls unmanned anti-pollution vessel and make corresponding action, it is preceding to go to salvage Rubbish or avoiding obstacles.The multiple characteristics extracted to foreground target using branch machine learning are classified, can be more accurate Identification rubbish and interfering object.
The foregoing is merely the preferred embodiment of the present invention, are not intended to restrict the invention, it is clear that those skilled in the art Various changes and modifications can be made to the invention by member without departing from the spirit and scope of the present invention.If in this way, the present invention Within the scope of the claims of the present invention and its equivalent technology, then the present invention is also intended to include these these modifications and variations Including modification and variation.

Claims (6)

1. a kind of garbage on water recognition methods based on multiple features machine learning, it is characterised in that：Include the following steps：
S1：Identification model initializes, and generates the Database Systems that content is empty, Database Systems include label just Parameter in example training data, counter-example training data, positive example test data, counter-example test data and machine learning model；
S2：The positive example training data of label and counter-example training data are subjected to initialization pretreatment, normalized, extraction Corresponding feature vector brings the feature vector of the feature vector of positive example image and counter-example image into system model training, raw It at corresponding knowledge base, is then tested to knowledge base using test data, until knowledge base is met the requirements；
S3：Corresponding knowledge base is brought into identifying system, water surface status data is dynamically read by boat-carrying camera, then The image data of periodic acquisition camera transmission, and it is transmitted to host computer；
S4：Host computer after identical pretreatment, extracts feature identical with when modeling when carrying out image data with above-mentioned modeling Vector, and bring feature vector into machine learning model, machine learning model reads corresponding knowledge base, to new feature vector The result that judges, and will determine that is exported in the form of flag data.
2. the garbage on water recognition methods according to claim 1 based on multiple features machine learning, it is characterised in that：In step In rapid S1, the acquisition for the positive example training data, counter-example training data, positive example test data and counter-example test data is adopted With the video for the different scenes that boat-carrying camera is shot, corresponding linear list is established to different environment to map, mapping is linear The precision of table has large effect to the accuracy of knowledge base model, then intercept in video corresponding frame as acquisition just Example training data, counter-example training data, positive example test data and counter-example test data.
3. the garbage on water recognition methods according to claim 2 based on multiple features machine learning, it is characterised in that：In step In rapid S2, the pretreatment to data image includes the following steps：
S21：Using improved median filtering algorithm projecting edge；
S22：Foreground target is protruded using saliency extraction algorithm；
S23：Threshold value is dynamically proposed by otsu algorithms, detaches foreground target and target context obtains binary image；
S24：Volume is avoided by the separate mode based on connected domain size using morphology operations repairing foreground subject edges The interference of smaller foreground target.
4. the garbage on water recognition methods according to claim 3 based on multiple features machine learning, it is characterised in that：In step In rapid S2, the extraction to target feature vector includes：
The feature that the description of the RGB color model based on water surface foreground target and edge square theoretical description are combined is extracted, it is raw At feature vector, RGB color model：
Y=a | R-RA|+b|G-GA|+c|B-BA| (1)
R in formula (1)A、GA、BAIndicate that water surface background be averaged RGB color respectively, the color of R, G, B expression current pixel point, A, b, c indicate adjustable linear magnification；
Edge square theoretical description submodel：
ai=Li+Li+2-2Li+1 (2)
Formula (2) indicates neighboring edge point to the second-order difference equation of edge contour centre distance, wherein LiIndicate edge contour center To the distance at edge, i takes 0 to arrive Cnt-2, and Cnt is peripheral edge pixel quantity, works as ai<Threshold, Threshold are pre- Determine threshold value, needs to test adjusting according to actual conditions, then it is assumed that aiEqual to 0, then it is smaller to reflect the variation of target edge.
5. the garbage on water recognition methods according to claim 4 based on multiple features machine learning, it is characterised in that：In step In rapid S3, to meet system real time requirement, relational expression is met according to the speed V that unmanned hull travels, sampling period T：TV≤ 0.2。
6. the garbage on water recognition methods according to claim 5 based on multiple features machine learning, it is characterised in that：In step In rapid S2, in medium filtering, the pixel of color close to water surface average color assigns lighter weight, for color distance water Mean pixel Euclidean distance farther pixel in face assigns heavier weight；Edge repair morphology operations, which use, first expands post-etching Mode repair edge, for filtering core therein, the filtering core of dynamic size is used according to the distance of image.
CN201810537983.0A 2018-05-30 2018-05-30 Water surface garbage identification method based on multi-feature machine learning Active CN108764154B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810537983.0A CN108764154B (en) 2018-05-30 2018-05-30 Water surface garbage identification method based on multi-feature machine learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810537983.0A CN108764154B (en) 2018-05-30 2018-05-30 Water surface garbage identification method based on multi-feature machine learning

Publications (2)

Publication Number Publication Date
CN108764154A true CN108764154A (en) 2018-11-06
CN108764154B CN108764154B (en) 2020-09-08

Family

ID=64004277

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810537983.0A Active CN108764154B (en) 2018-05-30 2018-05-30 Water surface garbage identification method based on multi-feature machine learning

Country Status (1)

Country Link
CN (1) CN108764154B (en)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110009800A (en) * 2019-03-14 2019-07-12 北京京东尚科信息技术有限公司 A kind of recognition methods and equipment
CN110188680A (en) * 2019-05-29 2019-08-30 南京林业大学 Tea tree tender shoots intelligent identification Method based on factor iteration
CN110516625A (en) * 2019-08-29 2019-11-29 华育昌(肇庆)智能科技研究有限公司 A kind of method, system, terminal and the storage medium of rubbish identification classification
CN110555418A (en) * 2019-09-08 2019-12-10 无锡高德环境科技有限公司 AI target object identification method and system for water environment
CN111950357A (en) * 2020-06-30 2020-11-17 北京航天控制仪器研究所 Marine water surface garbage rapid identification method based on multi-feature YOLOV3
CN112733676A (en) * 2020-12-31 2021-04-30 青岛海纳云科技控股有限公司 Method for detecting and identifying garbage in elevator based on deep learning
CN113033313A (en) * 2021-02-26 2021-06-25 澜途集思生态科技集团有限公司 Deep learning-based water pollution judgment method
CN115588145A (en) * 2022-12-12 2023-01-10 深圳联和智慧科技有限公司 Unmanned aerial vehicle-based river channel garbage floating identification method and system
CN116563768A (en) * 2023-07-06 2023-08-08 生态环境部华南环境科学研究所(生态环境部生态环境应急研究所) Intelligent detection method and system for microplastic pollutants

Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1619593A (en) * 2004-12-09 2005-05-25 上海交通大学 Video frequency motion target adaptive tracking method based on multicharacteristic information fusion
CN101916369A (en) * 2010-08-20 2010-12-15 西安电子科技大学 Face recognition method based on kernel nearest subspace
CN103065331A (en) * 2013-01-15 2013-04-24 南京工程学院 Target tracking method based on correlation of space-time-domain edge and color feature
CN104209949A (en) * 2013-05-31 2014-12-17 董挺挺 Cleaning remote control robot device
CN105512666A (en) * 2015-12-16 2016-04-20 天津天地伟业数码科技有限公司 River garbage identification method based on videos
CN105868683A (en) * 2015-11-24 2016-08-17 乐视致新电子科技（天津）有限公司 Channel logo identification method and apparatus
CN106127240A (en) * 2016-06-17 2016-11-16 华侨大学 A kind of classifying identification method of plant image collection based on nonlinear reconstruction model
CN106203265A (en) * 2016-06-28 2016-12-07 江苏大学 A kind of Construction Fugitive Dust Pollution based on unmanned plane collection image is derived from dynamic monitoring and coverage prognoses system and method
CN106228547A (en) * 2016-07-15 2016-12-14 华中科技大学 A kind of view-based access control model color theory and homogeneity suppression profile and border detection algorithm
CN106250936A (en) * 2016-08-16 2016-12-21 广州麦仑信息科技有限公司 Multiple features multithreading safety check contraband automatic identifying method based on machine learning
CN106845408A (en) * 2017-01-21 2017-06-13 浙江联运知慧科技有限公司 A kind of street refuse recognition methods under complex environment
US20170221190A1 (en) * 2013-05-14 2017-08-03 Kla-Tencor Corporation Machine learning method and apparatus for inspecting reticles
CN107194395A (en) * 2017-05-02 2017-09-22 华中科技大学 A kind of object dynamic positioning method based on colour recognition and contours extract
CN108058790A (en) * 2017-12-28 2018-05-22 重庆邮电大学 A kind of rubbish cleaning unmanned boat waterborne

Patent Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1619593A (en) * 2004-12-09 2005-05-25 上海交通大学 Video frequency motion target adaptive tracking method based on multicharacteristic information fusion
CN101916369A (en) * 2010-08-20 2010-12-15 西安电子科技大学 Face recognition method based on kernel nearest subspace
CN103065331A (en) * 2013-01-15 2013-04-24 南京工程学院 Target tracking method based on correlation of space-time-domain edge and color feature
US20170221190A1 (en) * 2013-05-14 2017-08-03 Kla-Tencor Corporation Machine learning method and apparatus for inspecting reticles
CN104209949A (en) * 2013-05-31 2014-12-17 董挺挺 Cleaning remote control robot device
CN105868683A (en) * 2015-11-24 2016-08-17 乐视致新电子科技（天津）有限公司 Channel logo identification method and apparatus
CN105512666A (en) * 2015-12-16 2016-04-20 天津天地伟业数码科技有限公司 River garbage identification method based on videos
CN106127240A (en) * 2016-06-17 2016-11-16 华侨大学 A kind of classifying identification method of plant image collection based on nonlinear reconstruction model
CN106203265A (en) * 2016-06-28 2016-12-07 江苏大学 A kind of Construction Fugitive Dust Pollution based on unmanned plane collection image is derived from dynamic monitoring and coverage prognoses system and method
CN106228547A (en) * 2016-07-15 2016-12-14 华中科技大学 A kind of view-based access control model color theory and homogeneity suppression profile and border detection algorithm
CN106250936A (en) * 2016-08-16 2016-12-21 广州麦仑信息科技有限公司 Multiple features multithreading safety check contraband automatic identifying method based on machine learning
CN106845408A (en) * 2017-01-21 2017-06-13 浙江联运知慧科技有限公司 A kind of street refuse recognition methods under complex environment
CN107194395A (en) * 2017-05-02 2017-09-22 华中科技大学 A kind of object dynamic positioning method based on colour recognition and contours extract
CN108058790A (en) * 2017-12-28 2018-05-22 重庆邮电大学 A kind of rubbish cleaning unmanned boat waterborne

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
DEEPA.T.P 等: "Extraction of Invariant features of garbage area in multi-terrain image", 《INTERNATIONAL JOURNAL OF PURE AND APPLIED MATHEMATICS》 *
单建华: "改进权值函数的非局部均值去噪算法", 《中国图象图形学报》 *
微雪: "显著性论文学习阶段总结", 《HTTPS://WWW.CNBLOGS.COM/YINGYING0907/ARCHIVE/2013/03/22/2975172.HTML》 *
昝杰 等: "一种基于特征检测的机器人足球检测识别方法", 《广西大学学报：自然科学版》 *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110009800A (en) * 2019-03-14 2019-07-12 北京京东尚科信息技术有限公司 A kind of recognition methods and equipment
CN110009800B (en) * 2019-03-14 2023-04-07 北京京东乾石科技有限公司 Identification method and equipment
CN110188680A (en) * 2019-05-29 2019-08-30 南京林业大学 Tea tree tender shoots intelligent identification Method based on factor iteration
CN110188680B (en) * 2019-05-29 2021-08-24 南京林业大学 Tea tree tender shoot intelligent identification method based on factor iteration
CN110516625A (en) * 2019-08-29 2019-11-29 华育昌(肇庆)智能科技研究有限公司 A kind of method, system, terminal and the storage medium of rubbish identification classification
CN110555418A (en) * 2019-09-08 2019-12-10 无锡高德环境科技有限公司 AI target object identification method and system for water environment
CN111950357A (en) * 2020-06-30 2020-11-17 北京航天控制仪器研究所 Marine water surface garbage rapid identification method based on multi-feature YOLOV3
CN112733676A (en) * 2020-12-31 2021-04-30 青岛海纳云科技控股有限公司 Method for detecting and identifying garbage in elevator based on deep learning
CN113033313A (en) * 2021-02-26 2021-06-25 澜途集思生态科技集团有限公司 Deep learning-based water pollution judgment method
CN115588145A (en) * 2022-12-12 2023-01-10 深圳联和智慧科技有限公司 Unmanned aerial vehicle-based river channel garbage floating identification method and system
CN116563768A (en) * 2023-07-06 2023-08-08 生态环境部华南环境科学研究所(生态环境部生态环境应急研究所) Intelligent detection method and system for microplastic pollutants
CN116563768B (en) * 2023-07-06 2023-09-22 生态环境部华南环境科学研究所(生态环境部生态环境应急研究所) Intelligent detection method and system for microplastic pollutants

Also Published As

Publication number Publication date
CN108764154B (en) 2020-09-08

Similar Documents

Publication Publication Date Title
CN108764154A (en) A kind of garbage on water recognition methods based on multiple features machine learning
WO2021243743A1 (en) Deep convolutional neural network-based submerged oil sonar detection image recognition method
CN110414334B (en) Intelligent water quality identification method based on unmanned aerial vehicle inspection
CN106845408A (en) A kind of street refuse recognition methods under complex environment
CN110414368A (en) A kind of unsupervised pedestrian recognition methods again of knowledge based distillation
CN103383733B (en) A kind of track based on half machine learning video detecting method
CN109376792A (en) Photovoltaic cell open defect classification method based on multichannel residual error neural network
CN106934386B (en) A kind of natural scene character detecting method and system based on from heuristic strategies
CN103514448A (en) Method and system for navicular identification
CN109636758B (en) Water surface floater detection method based on space-time dynamic operator
CN105512666A (en) River garbage identification method based on videos
CN104732215A (en) Remote-sensing image coastline extracting method based on information vector machine
CN113065486A (en) Floater identification method, system, equipment and medium suitable for river channel
CN108073940B (en) Method for detecting 3D target example object in unstructured environment
CN105931241A (en) Automatic marking method for natural scene image
CN108710862A (en) A kind of high-resolution remote sensing image Clean water withdraw method
Zhao et al. Research on detection method for the leakage of underwater pipeline by YOLOv3
CN109284759A (en) One kind being based on the magic square color identification method of support vector machines (svm)
CN116246139A (en) Target identification method based on multi-sensor fusion for unmanned ship navigation environment
Zheng et al. Robust and automatic cell detection and segmentation from microscopic images of non‐setae phytoplankton species
CN110516565A (en) A kind of sea ship recognition methods based on Zernike and SVM
CN103150576A (en) Monitoring method of cyanobacterial bloom condition
CN109784389A (en) A kind of coal petrography recognition detection method based on Adaboost and Gabor algorithm
Pu et al. Comparison and Simulation of Deep Learning Detection Algorithms for Floating Objects on the Water Surface
CN113920300A (en) Visible light remote sensing image ship target identification method and system

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