CN107273809A - A kind of method of the real-time autonomous classification of fishing net under water for power buoy - Google Patents
A kind of method of the real-time autonomous classification of fishing net under water for power buoy Download PDFInfo
- Publication number
- CN107273809A CN107273809A CN201710362891.9A CN201710362891A CN107273809A CN 107273809 A CN107273809 A CN 107273809A CN 201710362891 A CN201710362891 A CN 201710362891A CN 107273809 A CN107273809 A CN 107273809A
- Authority
- CN
- China
- Prior art keywords
- image
- fishing net
- real
- video image
- under water
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Studio Devices (AREA)
Abstract
The present invention is to provide a kind of method of the real-time autonomous classification of fishing net under water for power buoy.Power buoy with the LED and Underwater Camera for automatically adjusting light intensity, the video image collected is real-time transmitted to by embedded image processing platform by Underwater Camera, correlated image processing method is run by embedded image processing platform, video image is handled, is instructed the propeller of power buoy to be acted accordingly according to the information after processing.The present invention can not only realize the real-time autonomous classification of fishing net under complex environment under water, and effectively improve the accuracy rate and efficiency of fishing net identification.
Description
Technical field
Independently know in real time the present invention relates to Underwater Targets Recognition field, especially a kind of fishing net under water for power buoy
Method for distinguishing.
Background technology
China was increasing to the protection of maritime rights and interests in recent years, and people are also more next for the demand of marine resources
Bigger, the exploitation of marine resources welcomes climax.With the increase of marine protection dynamics, ocean patrol and seabed operation are also more next
It is more extensive, but the accident that fishing net snarls propeller is happened occasionally, and very big directly or indirectly loss is brought to country.
Such as 2005, a wheel propeller for ship was snarled by fishing net and run out of steam, and is worked by turns under water in several divers several small
Shi Houcai recovers normal, but causes the damage of tailing axle oil sealing, causes oil leak, is brought to shipowner and plant huge
Economic loss.
Power buoy is that one kind need not anchor fixation, and the underwater propeller of itself can be used to realize autonomous and fixed
The buoy that point is stopped.Therefore as ship, when the underwater propeller of power buoy is rotated under water, also it is highly susceptible to fishing net
Interference.
Still lack effective detection means to fishing net at present, fishing net often in several meters under water even tens of rice, only with
It is difficult what is observed that naked eyes, which are, so the fishing net detection recognition method for developing a kind of view-based access control model is necessary.Present detection
Technology mainly includes acoustics imaging and optical imagery, and acoustics imaging is to judge the position of immersed body using sound wave, shape and away from
From, scope big advantage wide with the visual field.But compared with optical imagery, acoustics imaging resolution ratio is low, and noise is serious, image
It is of low quality, it is impossible to detect Small object.Optical imagery is then direct acquisition video image, and high resolution is conducive to detecting small mesh
Mark.According to the characteristics of fishing net itself, netting twine is only several millimeters, and mesh is centimetres, therefore optical imagery is to be adapted to detection fishing
The technological means of net.But it is due to the growth requirement of fishery, the fishing net in coastal waters is more, and water quality is poor, and definition under water is very poor,
And optical imagery is easily influenceed by illumination, make the interference of noise in video image than more serious, simultaneously because lacking effective
Fishing net means of identification, fishing net identification technology also have very big space go improvement；Therefore design one kind is needed to carry regulation light intensity
LED, and the fishing net autonomous classification method based on optical video image, and with this real-time instruction power buoy underwater propeller
Carry out corresponding actions, it is to avoid the generation for the phenomenon that propeller is snarled by fishing net.
The content of the invention
The real-time autonomous knowledge of fishing net can not only be realized under complex environment under water it is an object of the invention to provide a kind of
Not, and can effectively improve fishing net identification accuracy rate and efficiency the real-time autonomous classification of fishing net under water for power buoy
Method.
The object of the present invention is achieved like this：
(1) video image is gathered using Underwater Camera；
(2) video image that Underwater Camera is obtained is entered into line definition evaluation；
(3) if the definition of the image collected is less than the threshold value of setting, open LED and be illuminated, by LED
Intensity of illumination be divided into several ascending gears, each gear correspond to certain definition scope, according to clear
The size of degree selects the gear of corresponding LED, by the brightness regulation of underwater LED lamp to corresponding gear；Return to step (1) weight
New collection video image, if the definition of the image collected proceeds step (4) not less than the threshold value of setting；
(4) image preprocessing is carried out to the video image collected, including：Noise Elimination from Wavelet Transform method removes Gauss and made an uproar
Sound, LPF removes random noise, and multiple dimensioned Retinex methods are used for image enhaucament, and bilateral filtering method is used to solve to shine
Degree estimation distortion；
(5) video image after processing is changed into bianry image；
(6) image processing platform detects fishing net netting twine using Hough transformation, including netting twine sidewards also has endways net
Line；
(7) according to the netting twine detected, any even number bar horizontal line, slope k ∈ [0, j) ∪ (- i, 0) and even number bar are calculated
Number of grid between vertical line, slope k ∈ [j ,+∞) ∪ (- ∞ ,-i], is stored in set D；Calculate any odd number bar horizontal line
Number of grid between odd number bar vertical line, is stored in set E；Calculate between any odd number bar horizontal line and even number bar vertical line
Number of grid, is stored in set F；
(8) if it is even number that each value in set D, which is each value in odd number, and E and F, it is judged as fishing
Net, otherwise, is unsatisfactory for any one condition therein and is all judged as non-fishing net, relevant information is submitted to the motion of power buoy
Operational module, instructs the working condition of its underwater propeller, is then back to step (1).
In order to solve the problem of underwater propeller of power buoy is snarled by fishing net, the present invention proposes a kind of for power
The method of the real-time autonomous classification of fishing net under water of buoy.Including：The LED of light intensity is automatically adjusted, for underwater lighting, is made under water
Video camera obtains clearly video image.Good B/W camera resistant to pressure, for gathering video image, and by video image
Transfer back to embedded image processing platform.Embedded image processing platform, for running a kind of real-time autonomous classification of fishing net under water
Method, and the underwater propeller working condition of the information guiding power buoy provided according to this method.
The present invention has following technological achievement：
1st, illumination visibility can be effectively improved using the method for automatically adjusting illumination intensity of the present invention and video is clear
Degree, obtains the less visual pattern of noise jamming, improves recognition efficiency and accuracy rate.
2nd, the autonomous classification of fishing net can be realized using the present invention, makes the underwater propeller of power buoy as the case may be
Independently judged, it is effective to reduce the loss that fishing net winding propeller is brought.
Brief description of the drawings
Fig. 1 is used for the flow chart of the method for the real-time autonomous classification of fishing net under water of power buoy.
Fig. 2 Noise Elimination from Wavelet Transform method and step figures.
Embodiment
The invention will be further described for citing below in conjunction with the accompanying drawings.
With reference to Fig. 1, the method for the real-time autonomous classification of fishing net under water for power buoy of the invention comprises the following steps：
Step 1：The video image of target area is obtained using underwater LED lamp and Underwater Camera.Underwater LED lamp is arranged on
Video camera is illuminated under the front of power buoy, water supply so that the image of shooting becomes apparent from reliably.Underwater Camera
Installed in the top of LED, the image for gathering target area carries out the identification of fishing net.
Step 2：The transmission of video images that step 1 is collected carries out underwater video figure to embedded image processing platform
The definition evaluation of picture.
Described definition evaluation method is the definition evaluation method based on Tenengrad evaluation functions.At image
In reason, it is considered that focus, namely the preferable image of definition has more sharp edge, therefore with bigger gradient letter
Numerical value.
Tenengrad functions use the Grad of Sobel operator extractions horizontally and vertically, and its detailed process is such as
Under：
Make Gx(x, y), Gy(x, y) is image slices vegetarian refreshments M (x, y) and Sobel boundary operators convolution respectively
The Tenengrad values for defining the image are
Ten values are bigger to represent that image is more clear.
Step 3：When Ten ＜ d (d is threshold value), illustrates that present image is unintelligible and be illuminated, it is necessary to open LED.Will
0 ＜ Ten ＜ d are equally divided into m gear, and the intensity of illumination of LED is divided into r ascending light intensity gear.IfWherein n ∈ r, then, can so as to improve illumination by LED regulation to n-th of gear
Degree of opinion and video definition, then effectively improve the recognition efficiency of fishing net, and return to step 1 resurveys video image.When Ten >=
During d, illustrate that present image is clear, meet the requirement that next step carries out image procossing, proceed step 4.
Step 4：Image preprocessing step mainly includes：Noise Elimination from Wavelet Transform method removes Gaussian noise, and LPF is gone
Except random noise, multiple dimensioned Retinex methods are used for image enhaucament (projecting edge feature), and bilateral filtering solves illumination estimation and lost
True problem.
Described Noise Elimination from Wavelet Transform method is mainly removal Gaussian noise, and the image for collecting Underwater Camera is clear
It is clear, the minutia of image is retained to greatest extent, Noise Elimination from Wavelet Transform Method And Principle is as follows：
Signals and associated noises are subjected to multi-scale wavelet transformation, wavelet field is transformed to from real domain, then under each yardstick as far as possible
Extraction signal wavelet coefficient, and remove the wavelet coefficient of noise.Finally use wavelet inverse transformation reconstruction signal.Its flow is as schemed
Shown in 2.
Described LPF method, which is mainly, removes the random noise point that Noise Elimination from Wavelet Transform method is not removed.
Described multiple dimensioned Retinex methods are mainly used in strengthening the image after noise reduction, and step is realized in this operation
It is rapid as follows：
Given piece image S (x, y), can be decomposed into reflected image R (x, y) and incident image L (x, y), given image
Relation such as following formula between reflected image and incident image：
S (x, y)=R (x, y) L (x, y) (3)
L is also known as illumination component, and R is also known as reflecting component.Multiple dimensioned Retinex basic thought is, in S, reduces L
Influence so that as far as possible retain object essence reflecting attribute, i.e. R.The expression formula of methods described is as follows：
R (x, y) is reflected image, and K is scale parameter, and it is 3 generally to take K, and weights are* volume is represented
Product computing, F is center ring around function, i.e. Gaussian function, convolution kernel：
λ therein is normaliztion constant, it is ensured that the integration inside convolution kernel is that 1, c is convolution kernel size.
Center ring is estimated that the low-frequency component of illumination component L corresponding images around function, and remove from S low frequency into
Point, left is exactly high fdrequency component, i.e. reflecting component R, remains the edge details of image.
The bilateral filtering method, which is mainly, to be solved region illumination estimation distortion that multiple dimensioned Retinex methods cause and goes out
Existing " halo artifact " phenomenon, improves the definition of image, is conducive to the edge extracting of later image.
Step 5：The image obtained is converted into bianry image.
Step 6：The Hough transformation is used to extract the fishing net netting twine in binary image, obtains the straight line where fishing net
Set.Its main thought is by the way that the parameter of (rectangular coordinate system) linear equation on image and variable are exchanged, so as to realize
Each non-zero pixels point on image, is transformed to the straight line of parameter space (polar coordinates), and belongs to same on image
The point of straight line just forms a plurality of straight line in parameter space and intersected at a point, and this is line correspondence in the coordinate of parameter space
Parameter.Therefore, in parameter space, by the maximum for calculating accumulated result, you can obtain the set of straight line on image.
Step 7：Grid set is obtained according to straight line set, any even number bar horizontal line is calculated
And vertical lineBetween number of grid, be stored in set D.Calculate any odd number bar vertical line and
Number of grid between horizontal line and arbitrarily between odd number bar horizontal line and even number bar vertical line, and be stored to respectively in set E and F.
Step 8：If any number in set D is that any number in odd number, set E and F is even number, then it is judged as
Fishing net, is unsatisfactory for wherein any condition person for non-fishing net.Relevant information is submitted to the motion operational module of power buoy, to refer to
The working condition of its underwater propeller is led, step 1 is then back to.
Above-mentioned Rule of judgment is according to being：Fishing net is made up of grid, and each grid is that have two horizontal lines and two vertical line groups
The closed figure of synthesis.There is geometric knowledge to understand, any odd number bar horizontal line and vertical line and odd number bar horizontal line and even number bar vertical line
Between be to contain the quadrangle that even number does not occur simultaneously certainly, and arbitrarily contained certainly between even number bar horizontal line and vertical line
The quadrangle that odd number does not occur simultaneously, so according to this result it may determine that going out whether target is network structure.
Claims (2)
1. a kind of method of real-time autonomous classification of fishing net under water for power buoy, it is characterized in that：
(1) video image is gathered using Underwater Camera；
(2) video image that Underwater Camera is obtained is entered into line definition evaluation；
(3) if the definition of the image collected is less than the threshold value of setting, open LED and be illuminated, by the light of LED
It is divided into several ascending gears according to intensity, each gear correspond to certain definition scope, according to definition
Size selects the gear of corresponding LED, by the brightness regulation of underwater LED lamp to corresponding gear；Return to step (1) is adopted again
Collect video image, if the definition of the image collected proceeds step (4) not less than the threshold value of setting；
(4) image preprocessing is carried out to the video image collected；
(5) video image after processing is changed into bianry image；
(6) image processing platform detects fishing net netting twine using Hough transformation, including netting twine sidewards also has endways netting twine；
(7) according to the netting twine detected, calculating any even number bar horizontal line, slope k ∈, [0, j) ∪ (- i, 0) and even number bar are perpendicular
Number of grid between line, slope k ∈ [j ,+∞) ∪ (- ∞ ,-i], is stored in set D；Calculate any odd number bar horizontal line and
Number of grid between odd number bar vertical line, is stored in set E；Calculate the net between any odd number bar horizontal line and even number bar vertical line
Lattice quantity, is stored in set F；
(8) if it is even number that each value in set D, which is each value in odd number, and E and F, fishing net is judged as, it is no
Then, it is unsatisfactory for any one condition therein and is all judged as non-fishing net, relevant information is submitted to the motion work of power buoy
Module, instructs the working condition of its underwater propeller, is then back to step (1).
2. the method for the fishing net under water real-time autonomous classification according to claim 1 for power buoy, it is characterized in that：Institute
That states carries out image preprocessing to the video image that collects, including：Noise Elimination from Wavelet Transform method removes Gaussian noise, low pass filtered
Ripple removes random noise, and multiple dimensioned Retinex methods are used for image enhaucament, and bilateral filtering method is used to solve illumination estimation mistake
Very.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710362891.9A CN107273809A (en) | 2017-05-22 | 2017-05-22 | A kind of method of the real-time autonomous classification of fishing net under water for power buoy |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710362891.9A CN107273809A (en) | 2017-05-22 | 2017-05-22 | A kind of method of the real-time autonomous classification of fishing net under water for power buoy |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107273809A true CN107273809A (en) | 2017-10-20 |
Family
ID=60064452
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710362891.9A Pending CN107273809A (en) | 2017-05-22 | 2017-05-22 | A kind of method of the real-time autonomous classification of fishing net under water for power buoy |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107273809A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109003307A (en) * | 2018-06-11 | 2018-12-14 | 西北工业大学 | Fishing mesh sizing method based on underwater Binocular vision photogrammetry |
| CN109872499A (en) * | 2018-12-25 | 2019-06-11 | 大连理工大学 | A kind of block in-situ monitor alarm system based on image recognition |
| CN113344801A (en) * | 2021-03-04 | 2021-09-03 | 北京市燃气集团有限责任公司 | Image enhancement method, system, terminal and storage medium applied to gas metering facility environment |
| CN115409890A (en) * | 2022-11-02 | 2022-11-29 | 山东大学 | Self-defined mark detection method and system based on MSR and generalized Hough transform |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101750848A (en) * | 2008-12-11 | 2010-06-23 | 鸿富锦精密工业（深圳）有限公司 | Pick-up device and light filling method |
| CN101957539A (en) * | 2010-06-09 | 2011-01-26 | 杭州海康威视数字技术股份有限公司 | Method and device for monitoring and supplementing light |
| CN102118608A (en) * | 2009-12-30 | 2011-07-06 | 捷达世软件（深圳）有限公司 | System and method for adjusting video monitoring light |
| CN105096298A (en) * | 2014-05-08 | 2015-11-25 | 东北大学 | Grid feature point extraction method based on fast line extraction |
| CN105676230A (en) * | 2016-04-11 | 2016-06-15 | 中国科学院半导体研究所 | Real-time autonomous fishing net identification device and method for underwater obstacle avoidance navigation |
-
2017
- 2017-05-22 CN CN201710362891.9A patent/CN107273809A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101750848A (en) * | 2008-12-11 | 2010-06-23 | 鸿富锦精密工业（深圳）有限公司 | Pick-up device and light filling method |
| CN102118608A (en) * | 2009-12-30 | 2011-07-06 | 捷达世软件（深圳）有限公司 | System and method for adjusting video monitoring light |
| CN101957539A (en) * | 2010-06-09 | 2011-01-26 | 杭州海康威视数字技术股份有限公司 | Method and device for monitoring and supplementing light |
| CN105096298A (en) * | 2014-05-08 | 2015-11-25 | 东北大学 | Grid feature point extraction method based on fast line extraction |
| CN105676230A (en) * | 2016-04-11 | 2016-06-15 | 中国科学院半导体研究所 | Real-time autonomous fishing net identification device and method for underwater obstacle avoidance navigation |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109003307A (en) * | 2018-06-11 | 2018-12-14 | 西北工业大学 | Fishing mesh sizing method based on underwater Binocular vision photogrammetry |
| CN109003307B (en) * | 2018-06-11 | 2021-10-22 | 西北工业大学 | Underwater binocular vision measurement-based fishing mesh size design method |
| CN109872499A (en) * | 2018-12-25 | 2019-06-11 | 大连理工大学 | A kind of block in-situ monitor alarm system based on image recognition |
| CN113344801A (en) * | 2021-03-04 | 2021-09-03 | 北京市燃气集团有限责任公司 | Image enhancement method, system, terminal and storage medium applied to gas metering facility environment |
| CN115409890A (en) * | 2022-11-02 | 2022-11-29 | 山东大学 | Self-defined mark detection method and system based on MSR and generalized Hough transform |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108444447B (en) | Real-time autonomous detection method for fishing net in underwater obstacle avoidance system | |
| CN107273809A (en) | A kind of method of the real-time autonomous classification of fishing net under water for power buoy | |
| CN106548463B (en) | Sea fog image automatic defogging method and system based on dark and Retinex | |
| CN104217215B (en) | A kind of classifying identification method of water surface foggy image and picture rich in detail | |
| CN109978869A (en) | A kind of sea horizon detection method and system based on gray level co-occurrence matrixes and Hough transform | |
| CN101915910B (en) | Method and system for identifying marine oil spill object by marine radar | |
| CN103530853B (en) | Infrared light intensity image and infrared polarization figure image intensifying fusion method | |
| CN107818326A (en) | A kind of ship detection method and system based on scene multidimensional characteristic | |
| CN109086824A (en) | A kind of sediment sonar image classification method based on convolutional neural networks | |
| CN104933680B (en) | A kind of intelligent quick sea fog minimizing technology of unmanned boat vision system video | |
| CN109376589B (en) | ROV deformation small target identification method based on convolution kernel screening SSD network | |
| CN110532897A (en) | The method and apparatus of components image recognition | |
| CN108764154B (en) | Water surface garbage identification method based on multi-feature machine learning | |
| CN111476120B (en) | Unmanned aerial vehicle intelligent ship water gauge identification method and device | |
| CN109117802A (en) | Ship Detection towards large scene high score remote sensing image | |
| CN110516606A (en) | High-resolution satellite image any direction Ship Target Detection method | |
| CN105676230B (en) | Real-time fishing net autonomous classification device and recognition methods for the navigation of underwater avoidance | |
| CN110288623B (en) | Data compression method for unmanned aerial vehicle maritime net cage culture inspection image | |
| CN106709426A (en) | Ship target detection method based on infrared remote sensing image | |
| CN109815807A (en) | A kind of ship detecting method of pulling in shore based on edge line analysis and converging channels feature | |
| CN102750683B (en) | Filtering method for sea surface stripe noise and sea surface stripe cloud in moderate resolution imaging spectroradiometer (MODIS) remote sensing image | |
| CN104966272A (en) | Underwater sea cucumber image processing method and system | |
| CN106156758B (en) | A kind of tidal saltmarsh method in SAR seashore image | |
| CN110335245A (en) | Cage netting damage monitoring method and system based on monocular space and time continuous image | |
| CN116152115B (en) | Garbage image denoising processing method based on computer 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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171020 |