CN107622231B - A kind of intelligent floating material collection system of water day one and its collection method - Google Patents
A kind of intelligent floating material collection system of water day one and its collection method Download PDFInfo
- Publication number
- CN107622231B CN107622231B CN201710806662.1A CN201710806662A CN107622231B CN 107622231 B CN107622231 B CN 107622231B CN 201710806662 A CN201710806662 A CN 201710806662A CN 107622231 B CN107622231 B CN 107622231B
- Authority
- CN
- China
- Prior art keywords
- rubbish
- image
- unmanned plane
- described image
- 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.)
- Active
Links
Abstract
The invention discloses a kind of intelligent floating material collection system of water day one and its collection methods.The system comprises image capture module, image processing module, device for cleaning water surface rubbish.Image capture module obtains the image of lake surface for clearance by unmanned plane, and converts by A/D, then image gray processing, carry out image preprocessing later.Image processing module divides lake surface background in image and foreground object；Identify the target rubbish in foreground object: the object by the way that area in foreground object to be greater than to preset value excludes；Coordinate is established using the camera of unmanned plane itself as coordinate origin to the target rubbish identified, calculates position coordinates under this coordinate.Device for cleaning water surface rubbish clears up target rubbish according to the position coordinates of target rubbish.Cleaning device and unmanned plane are combined the full-automation for realizing colleting garbage floated on water and the efficiency for greatly enhancing water surface cleaning work by the present invention.
Description
Technical field
The present invention relates to a kind of collection system and its collection methods more particularly to a kind of intelligent floating material of water day one to collect
System and its collection method.
Background technique
With the development of economy, people's lives level all increases, and house refuse also gradually increases, but due to one
The quality and behavioural habits of a little people does not have clear improvement, has ignored the protection of the protection to environment, especially water resource, water pollution
Have become serious problems.According to statistics, the ocean rubbish only on the Pacific Ocean just has reached 3,500,000 square kilometres, and only 2010
The whole world just has 8,000,000 tons of plastics to flow into ocean lake, and this number rises year by year.According to the French daily Le Figaro,
2013,1,500,000 animals became the victim of plastic garbage in ocean.
It is not only ocean, the pollution in lake river is also got worse, and causes the concern of society.Various floatings on lake surface
Rubbish has seriously affected the landscape of water quality and lake surface, or even also will cause water pollution, water quality deterioration, and will eventually endanger
The life and health of evil mankind itself.For landscape lake, since the tourist industry and natural conditions swift and violent by development are influenced, lake
Normal adrift various rubbish, make the travel value at scenic spot have a greatly reduced quality on face.The contamination phenomenon of landscape lake has at many scenic spots
In the presence of if clearing up not in time, not only resulting in visual pollution, influence natural landscape, can also be caused to the aquatile in lake
Life threat.The present invention is using efficiently, intelligently cleaning landscape lake lake surface floating refuse, to lake surface cleaning is maintained, is built as target
One good scenic environment is of great significance.
Foreign countries are more early to the research of water-surface cleaning device, the cleaning device of design also comparative maturity.Currently, Japan, America and Europe
It Deng country and puts into a huge sum of money in succession to be used to develop automatic device for cleaning water surface rubbish, for example, the enterpriser Li Cha in South Africa
De Hadiman (Richard Hardiman) develops the sea cleaner of a entitled " rubbish shark " (WasteShark)
Device people can independently cruise near harbour, clear up ocean rubbish.
In contrast, domestic to start late for the research of device for cleaning water surface rubbish, it utilizes manually drive mostly in the market
The mode for sailing garbage boat salvaging compares the rescue vessel for floating articles on water surface that typically Shanghai Waste Treatment Co. develops, adopts
It is salvaged with the mode of fuel oil driving and pilot steering.Since two thousand, under the effort of numerous scholars and engineer, I
State also achieves plentiful and substantial research achievement in terms of cleaning device automation.Such as " water surface robot " product of cloud continent science and technology with
Based on world's minimum high-performance unmanned boat platform of independent research, ship, communication, automation, robot control, remote has been merged
Various technologies such as range monitoring, networked system break Chinese water surface intelligent robot blank, which are mainly applied to it is military,
The fields such as security protection, exploration, environmental protection, rescue.At the beginning of 2016, the Chinese Academy of Sciences successfully has developed the cleaning of " wind light mutual complementing " autonomous type water surface
Robot is similar to household cleaning machine people, autonomous classification rubbish and can be purged, be mainly used in various oceans, lake
The remote job of the cleaning such as pool and the lake in scenic spot, the solid refuse in pond, duckweed and danger zone, however lake surface rubbish
The locating accuracy and recognition accuracy of rubbish are not very high.
Summary of the invention
In view of this, the present invention provides a kind of intelligent floating material collection system of water day one and its collection method, it will be clear
Clean device and unmanned plane combine the full-automation for realizing colleting garbage floated on water and to greatly enhance the water surface clear
The efficiency that science and engineering is made.
Solution of the invention is: a kind of intelligent floating material collection system of water day one comprising unmanned plane and the water surface
Garbage cleaning device；Unmanned plane is equipped with image capture module and image processing module；
Described image acquisition module obtains the image of lake surface for clearance by unmanned plane, and described image is turned by A/D
The discrete processes in the space and amplitude that convert analog signals into digital signal realization described image are changed, then described image gray scale
Change, carries out image preprocessing later；
Lake surface background in described image and foreground object are divided and obtain the foreground object by described image processing module；
Identify the target rubbish in the foreground object: the object by the way that area in the foreground object to be greater than to preset value excludes；It is right
The target rubbish identified establishes coordinate using the camera of unmanned plane itself as coordinate origin, calculates out position under this coordinate
Coordinate；
The device for cleaning water surface rubbish clears up target rubbish according to the position coordinates of target rubbish.
As a further improvement of the foregoing solution, it is identified the system also includes database, the database purchase
Target rubbish, and sorted generalization, figure of the described image acquisition module to acquisition are carried out according to the feature after target rubbish binaryzation
As calling the target rubbish of the database to compare, the target rubbish in quick obtaining described image.
As a further improvement of the foregoing solution, the system also includes paths to design module, and module is designed in the path
Plan the unmanned plane path design, the unmanned plane path design the following steps are included:
Thiessen polygon figure is drawn according to the condition distribution situation of known lake surface；
Maximum broken line is carried out to loke shore profile on this basis, loke shore profile is made to become the inscribed more of a fitting environment
Side shape, each edge of inscribed polygon are a route；
Since the longest edge of inscribed polygon, unmanned plane starts to scan according to Dijkstra's algorithm to lake surface, route
Between spacing be camera scanning range；
A way point is set for unmanned plane at a certain distance on route using differential GPS, way point is corrected, protects
The way point for meeting and sketching the contours of inscribed polygon feature completely is stayed, guarantees that unmanned plane will not conform to during navigation because of distance
It is suitable to lose next way point or deviate track；
Unmanned plane is just navigated by water and is scanned according to revised way point in real navigation.
As a further improvement of the foregoing solution, the system also includes paths to design module, and module is designed in the path
Plan that the path design of the device for cleaning water surface rubbish, the path design of the device for cleaning water surface rubbish are divided into two parts rule
Draw: consider free space should than hull size position beyond how many Rough Planning and stress space it is relatively narrow when ship positioning thin rule
It draws.
Further, in Rough Planning apply ant group algorithm, the Rough Planning the following steps are included:
Environmental modeling is carried out, grid division is carried out to entire lake surface: barrier being divided into black square, remaining is white square, no
Sufficient half-space is more than that half-space is calculated by a lattice, draws all possible route in white square between Origin And Destination by not calculating；
Optimal path is selected, first expanded scope: expanding as a unit for the multiple grids being connected in environmental modeling, statistics
The probability that barrier occurs in each unit, calculates the average value of these probability, and the range higher than average value is cast out, lower than flat
Continue iteration in aforementioned manners in the range of mean value, until finding out optimal path；
Optimal path is smoothed.
Further, C space law, the thin planning are applied in thin planning are as follows: first round barrier do one it is polygon
Shape, the distance of polygon to loke shore side are x, make a < x <b, and it is in place to cross the vertex of polygon, the starting point P of garbage boat and rubbish institute
Q line is set, line is not passed through barrier, keeps starting point P and position the Q row that is connected respectively with the vertex of polygon more at a closure
Side shape, the side of the closed polygon form two paths M, N, compare this two paths length, and short is optimal path.
As a further improvement of the foregoing solution, the gray processing of described image is by luminance quantization: being divided into 0 to 255 i.e.
256 ranks, 0 characterization is most dark or completely black, and 255 characterize most bright or Quan Bai；Described image processing module uses gray level threshold segmentation
Method is split described image.
As a further improvement of the foregoing solution, described image pretreatment comprising steps of
(1) noise reduction filtering processing is carried out first, and noise is eliminated by differential；
(2) processing is sharpened to described image again.
As a further improvement of the foregoing solution, described image processing module detects lake surface typical case interference before image segmentation
Factor is simultaneously dispelled.
The present invention also provides a kind of intelligent floating material collection methods of water day one, are applied to any one of the above water day one
Body intelligence floating material collection system；The intelligent floating material collection method of water day one the following steps are included:
The image in waters for clearance is obtained by unmanned plane shooting, described image is converted by A/D turns analog signal
It is changed to digital signal and realizes the space of described image and the discrete processes of amplitude, then described image gray processing, carry out figure later
As pretreatment；
Lake surface background in described image and foreground object are divided and obtain the foreground object；Identify the foreground object
In target rubbish: pass through by area in the foreground object be greater than preset value object exclude；To the target rubbish identified,
Using the camera of unmanned plane itself as coordinate origin, coordinate is established, calculates position coordinates under this coordinate；
Target rubbish is cleared up according to the position coordinates of target rubbish.
Beneficial effects of the present invention are as follows:
1. unmanned plane assists cleaning lake surface rubbish: previous device for cleaning water surface rubbish has due to the scanning range of itself
Limit, wastes time, the energy, and efficiency is very low, and " eyes " of the present invention application unmanned plane as garbage boat, " visual field " range are wide,
It greatly improves work efficiency；
2. identifying rubbish using exclusive method: on lake surface, biggish object (such as ship of area is also had in addition to floating refuse
Only, sheet of landscape plant lotus etc.), therefore area on lake surface is greater than one square metre of object when identifying rubbish by the present invention
It excludes, remaining foreground object can be determined as floating refuse；The method is succinctly effective, and it is biggish to also eliminated area
The influence that object generates the identification of target rubbish；
3. the path of water-surface cleaning device is designed: the present invention considers the case where there are islands in lake, if water-surface cleaning device
With target rubbish respectively in the two sides of island, then cleaning device cannot be simple line navigation；The present invention is known
In the case where island position, path design is carried out to cleaning device, has improved application range；
4. the path of unmanned plane is designed: unmanned plane can be autonomous complete in region specified in advance through the planning of way point
It is designed at path, to carry out course；
5. earth station: the present invention includes earth station, the earth station can real-time display unmanned plane pat the picture taken the photograph, and can
UAV Attitude control, setting of navigation area etc. are carried out by the earth station.
Detailed description of the invention
Fig. 1 is the histogram for the image gray levels that present system uses.
Fig. 2 is the image gray processing processing and binary conversion treatment flow chart that present system uses.
Fig. 3 is the path design drawing for the unmanned plane that present system uses.
Fig. 4 is the path design drawing for the water-surface cleaning device that present system uses.
Fig. 5 is the program flow chart that present system uses.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
The intelligent floating material collection system of water day one of the invention mainly includes two major parts on hardware: unmanned plane and
Device for cleaning water surface rubbish.It mainly include image capture module, image processing module, path design module on software.Nobody
Machine is equipped with image capture module, image processing module, path design module.Image processing module, path design module can also
It is mounted on device for cleaning water surface rubbish.
Image capture module obtains the image of lake surface for clearance, and described image is converted analog signals by A/D conversion
Digital signal realizes the space of described image and the discrete processes of amplitude, then described image gray processing, and it is pre- to carry out image later
Processing.
Lake surface background in described image and foreground object are divided and obtain the foreground object by image processing module；Identification
Target rubbish in the foreground object: the object by the way that area in the foreground object to be greater than to preset value excludes；To identification
Target rubbish out establishes coordinate using the camera of unmanned plane itself as coordinate origin, and out position is calculated under this coordinate and is sat
Mark.
Device for cleaning water surface rubbish clears up target rubbish according to the position coordinates of target rubbish.Path design module is not only advised
The path design of the device for cleaning water surface rubbish is also planned in the path design for drawing the unmanned plane.Module is designed in path can also
To be divided into two submodules, it is separately mounted on unmanned plane be respectively completed the path of unmanned plane on device for cleaning water surface rubbish
The path design planning of design planning and device for cleaning water surface rubbish.
The present invention is by Image Acquisition, image processing and analysis technological orientation and identifies lake surface rubbish, and position is believed
Breath is sent to device for cleaning water surface rubbish, carries out rubbish cleaning.
In the present embodiment, two parts are related to image processing techniques:
1. identification of the unmanned plane to rubbish；
2. in a small range, accurate positionin of the water-surface cleaning device to rubbish.
Wherein, unmanned plane uses identical technology to the accurate positionin of rubbish to the identification of rubbish and water-surface cleaning device.
Image capture module is mountable on unmanned plane, and Image Acquisition first obtains lake surface figure by the camera shooting of unmanned plane
Then picture carries out pretreatment and is converted to the image information for being conducive to subsequent analysis.
Since ccd image sensor is less and insensitive to ultraviolet light by the damage of vibration and impact, so of the invention
CCD camera is carried in design on unmanned plane, and unmanned plane during flying is carried out the shooting of image in lake surface overhead by camera.
Camera needs to convert analog signals into accessible digital signal by A/D conversion after obtaining image, i.e., empty
Between and amplitude discrete processes.Then image gray processing, that is, by luminance quantization, 0 to 255 i.e. 256 are commonly divided into
Rank, 0 most dark (completely black), 255 most bright (Quan Bai).Information after treatment is the gray value of image, but wherein comprising by each
The limitation and interference of kind of condition and the noise generated, must first carry out image preprocessing before subsequent processing, that is, carry out with
Lower two steps:
(1) noise reduction filtering processing is carried out first, and noise, the shadow for preventing noise from analyzing subsequent image are eliminated by differential
It rings；
(2) since filtering processing meeting is so that soft edge needs again for projecting edge and profile information to image
It is sharpened processing, i.e., useful information is reinforced by integral, so that edge is prominent, clear.
Image processing module is a key content of the invention.Since the image information of acquisition back includes two parts:
Lake surface background and foreground object.Firstly the need of by image segmentation, lake surface background is separated with foreground object and obtains prospect
Object.Due to the foreground object on general landscape lake surface and few, mainly there are two major classes: the biggish object of area (including it is in blocks
Aquatic ornamental plant, ship etc.) and floating refuse, since floating refuse is varied, in contrast, the biggish object of area
It more easily identifies, therefore the present invention utilizes exclusive method: excluding the object that area is greater than one square metre, remaining foreground object can be true
Being set to is floating refuse.
(1) detection of lake surface typical case disturbing factor with dispel
In a practical situation, the lake surface image of acquisition is not ideal, under the influence of various environmental conditions, there is also
Some cases can interfere the processing and analysis of image, and typical disturbing factor has the following:
1. because of situations such as by blow or aquatile movement influenced, the water surface is easy to appear ripple；
2. bank scenery can form inverted image in lake surface under the influence of natural cause；
3. exposure area may be generated in the image of acquisition due to the arbitrariness of camera and light and water surface angle
Or retroreflective regions.
The presence of these factors can all have an impact subsequent image processing, or even can generate the target knowledge of false error
Not.Therefore, these typical disturbing factors should be dispelled before being further processed, in order to avoid rubbish identification is impacted,
Step operation avoids uncertain factor that may be present and has an impact to analysis result, improves the robustness of target identification.
(2) image segmentation
The present invention needs to identify in lake surface rubbish, therefore should be first by foreground object and lake surface background separation, image segmentation
It is to divide the image into some regions not overlapped and there is respective feature, it can be according to the spy of foreground object and lake surface background
Sign represents the image as the set of physically significant connected region, i.e., carries out to foreground object, the lake surface background in image
Label, positioning, then separate foreground object from lake surface background, and image segmentation is the basis of image recognition, segmentation
The quality of quality directly affects the progress of subsequent image processing.Due to the gray feature and lake surface background of the common foreground object of lake surface
There is significant difference, therefore image is split using gray level threshold segmentation method.
If original image is f (x, y), a suitable gray value is looked in f (x, y) with certain criterion, as threshold value T,
Image g (x, y) after so dividing can be indicated with following form:
Or
As soon as it can be seen that the accurate Ground Split of image, threshold value can be chosen for image if a suitable threshold value can be found
The critical issue of segmentation.Threshold value once it is determined that, the gray scale of each pixel is compared with threshold value, rubbish can be provided after segmentation
Rubbish target area.
If the gray level histogram of image, in apparent bimodal, the gray level histogram of image as shown in Figure 1 then may be used
The gray value of the lowest point between two peaks is chosen as threshold value T.If Fig. 2 is original image, gray processing processing, after binary conversion treatment
The comparison of image three.
(3) target rubbish identifies
Since the foreground object extracted does not only have target rubbish, there are also the excessive object of some volumes (such as: ship), institutes
It is excluded so that area can be greater than to the object of certain value b, it is remaining to can be identified as target rubbish.
The actual size of one picture is codetermined by the pixel and resolution ratio of electronic pictures.It is assumed that electronic pictures
Pixel is p, and resolution ratio q then obtains the size of electronic pictures(note: obtained dimensional units are inch, need conversion unit
For rice).If the field range after the fixed height of unmanned plane is y, then ratioThe size of foreground object is l, therefore prospect in picture
The real area of object
To determine whether that, for target rubbish, the real area S and setting value b that need to only compare foreground object (are set herein
Be 1 square metre) size relation, it may be assumed that
After rubbish is identified, unmanned plane establishes coordinate using the camera of itself as coordinate origin.The rubbish that will identify that
Position calculated, obtain the corresponding position coordinate of rubbish, then send corresponding coordinate by wireless transport module to
Device for cleaning water surface rubbish.The program library of a garbage warehouse can be added in image processing algorithm, most of rubbish is scanned
Into library, feature after finding out every class rubbish binaryzation carries out sorted generalization, finally allow unmanned plane to after rubbish scanning directly with
It is compared in library, identification rubbish more rapidly and efficiently.
Path design of the invention seeks to the optimization road that the various situations of reply are found in certain spatial dimension
Diameter.In the present invention, since unmanned plane and water-surface cleaning device are directed to navigation, therefore path design is accordingly divided into two
Point: the path design of unmanned plane and the path of device for cleaning water surface rubbish are designed.
One, the path design of unmanned plane
Unmanned plane of the present invention navigates by water in lake surface overhead, needs in advance to be designed navigation route.It is mission planning and
A part of trajectory planning needs the constraint by environmental information and mission bit stream.Its main contents is first according to known to
Lake surface condition distribution situation draws Voronoi diagram；Maximum broken line is carried out to loke shore profile on this basis, becomes one
It is bonded the inscribed polygon of environment, then since the longest edge of inscribed polygon, according to Dijkstra (nothing as shown in Figure 3
Man-machine path design drawing) environment is started to scan, the spacing between route is camera scanning range；Using differential GPS on road
A way point is set for unmanned plane at a certain distance above line, way point will meet the feature for sketching the contours of polygon completely,
And guarantee that unmanned plane will not lose next way point or deviation track because of apart from improper during navigation；Last nothing
It is man-machine just to be navigated by water and scanned according to revised way point in real navigation, according to the actual situation further to route into
Row depth Conformity planning.
Dijkstra (Di Jiesitela) algorithm is typical signal source shortest path algorithm, arrives it for calculating a node
The shortest path of his all nodes.It is mainly characterized by centered on starting point extending layer by layer outward, until expanding to terminal.
Dijkstra's algorithm is very representational shortest path first, is all had as basic content in many specialized courses detailed
It introduces, such as data knot, graph theory, operational research etc..Notice that there is no negative power sides in the algorithm requirement figure.
Problem description: in non-directed graph G=(V, E), it is assumed that the length of each edge E [i] be w [i], find by vertex V0 to
The shortest path of remaining each point.(signal source shortest path)
(2) algorithm description
1. algorithm idea:
If G=(V, E) is a Weighted Directed Graph, vertex set V in figure is divided into two groups, it is first group most short to have found out
The vertex set in path (is indicated, only one source point in S when initial will be added as soon as often acquiring a shortest path later with S
Into set S, until whole vertex are all added in S, algorithm just finishes), second group does not determine shortest path for remaining
Vertex set (is indicated) with U, and successively second group of vertex is added in S by the increasing order of shortest path length.In addition
In the process, total shortest path length for keeping each vertex in from source point v to S no more than in from source point v to U any vertex it is most short
Path length.In addition, one distance of each vertex correspondence, the distance on the vertex in S is exactly the shortest path path length from v to this vertex
It spends, the distance on the vertex in U, is to only include the current shortest path length that the vertex in S is intermediate vertex from v to this vertex.
2. algorithm steps:
A. when initial, S only includes source point, i.e. the distance of S={ v }, v are 0.U includes other vertex in addition to v, it may be assumed that U=
{ remaining vertex }, if vertex u has side in v and U,<u,v>normally there is weight, if u is not the side abutment points out of v,<u,v>power
Value is ∞.
B. the smallest vertex k of a distance v is chosen from U, k, in addition S (the selected distance be exactly v to k most
Short path length).
C. it is the intermediate point newly considered with k, modifies the distance on each vertex in U；If the distance from source point v to vertex u (is passed through
Vertex k) than original distance (it is short without vertex k), then modify the distance value of vertex u, the vertex k of modified distance value away from
From plus the power on side.
D. step b and c is repeated until all vertex are included in S.
Two, the path design of water-surface cleaning device
Path planning is divided to two kinds, Rough Planning and thin planning.Rough Planning considers that free space should be super than hull size position
It is how many out, and carefully plan stress space it is relatively narrow when ship positionning it is accurate.It will be apparent that cruise ship should be more focused on Rough Planning.
2.2.1. ant group algorithm
Firstly, carry out environmental modeling, to entire lake surface carry out grid division, that is, environment be placed on a world as
In the chessboard of chess, unlike, barrier is divided into black square (for 0), remaining is white square (for 1), insufficient half-space by not calculating,
It calculates more than half-space by a lattice, so thinner, expansion barrier, that is, black square range, the boundary as far as possible of lattice point suitably reduce, handles
All possible route is just drawn in white square between more accurate Origin And Destination.Secondly, selection optimal path, but distance is most
It is short and cruising range is optimal to meet simultaneously.First expanded scope, if 25 small lattice in first step environmental modeling are one new
Unit, count the probability that barrier in each unit occurs, calculate the average value of these probability, the range higher than average value is given up
It goes, continues iteration in aforementioned manners in sub-average range, until finding out optimal route.Finally, to optimal route
It is smoothed.This algorithm foundation is ant group algorithm.
(1) range
The range that ant observes is a grid world, and it is speed radius (usually 3) that ant, which has a parameter, then
It it is observed that range be exactly the 3*3 grid world, and mobile distance is also within the scope of this.
(2) environment
Environment where ant is a virtual world, wherein there is barrier, there is other ant, and there are also pheromones, letters
There are two types of breath elements, the food information element under spilling one is the ant for finding food, the nest under spilling one is the ant for finding nest
Pheromones.Each ant is only capable of perceiving the environmental information within the scope of it.Environment allows pheromones to disappear with certain rate.
(3) it looks for food rule
Food has been looked for whether in the range of every ant can perceive, if there is just directly passing by.Otherwise it sees if there is
Pheromones, and compare which point pheromones in the range of can perceive are most, it is just walked towards the place more than pheromones, and
Every ant can all be made mistakes with small probability, rather than the point movement most toward pheromones.Ant looks for the rule of nest and above one
Sample, only it makes a response to the pheromones of nest, and does not react food information element.
(4) movement rule
Every ant moves both facing to the most direction of pheromones, also, when surrounding does not have pheromones guide, ant
The movement for the direction inertia that can be moved originally according to oneself is gone down, also, has a random small disturbance in the direction of movement.
Ant original place is turn-taked in order to prevent, it can remember which point of having passed by just now, if it find that the next point to be walked is before
It passes by, it will be avoided as far as possible.
(5) avoidance rule
If the direction to be moved of ant has barrier to block, another direction of selection that it can be random, and have information
If element guides, it can be according to the regular behavior looked for food.
(6) pheromones rule
The pheromones that every ant spreads hair when just finding food or nest are most, and as it walks remote distance,
The pheromones sowed are fewer and fewer.According to this several rule, there is no direct relationship between ant, but every ant all and
Environment interacts, and by this tie of pheromones, actually associating between each ant.For example, when one
Ant has found food, it does not tell directly other ants to have food here, but sows pheromones to environment, when other
Ant by near it when, just may feel that the presence of pheromones, and then food is had found according to the guide of pheromones.
Negative-feedback can be suitably introduced into and carry out alignment path.
2.2.2.C space law
When water-surface cleaning device receives location information, need according to certain route navigation to target position.If clear
Clear between the position and target position of clean device, then can line navigation to target position；If there is barrier (such as island),
It then needs to design an optimal navigation path for water-surface cleaning device, target position is reached with bee line.
In lake barrier we using C space law (also known as Visual Graph space law) carry out path design:
Do a polygon round barrier (by taking lake Island as an example) first, the distance of polygon to bank is x, make a <
X <b (selection of wherein a, b make x be unlikely to too big or too small), x herein are that garbage boat and island bump against simultaneously in order to prevent
Be conducive to determine route.It crosses the vertex of polygon, the starting point P of garbage boat and rubbish position Q line and (is not passed through barrier
Line), so that P and Q is connected row with the vertex of polygon respectively into a closed polygon, the side of the polygon forms two
Path M, N compare this two paths length, and short is optimal path (water-surface cleaning device path design as shown in Figure 4
Figure).Water-surface cleaning device just presses the path and reaches target position.
In the selection of hardware system, by the comparison to common device, on-site programmable gate array FPGA is more suitable for this
Invention information treating capacity is big, the demanding feature of processing speed.It can be in existing unmanned plane model and water surface cleaning device
On the basis of be designed, build core of the hardware system based on programmable gate array FPGA as information processing.
The function and equivalent soft nuclear phase ratio realized due to FPGA stone have higher performance and lower power consumption, still
Soft core using more flexible, when the specific function run on hard nucleus management device encounters bottleneck, soft core is can be used in we
It is realized by programmable construction, so soft core of FPGA is used in combination with stone can be while providing high-performance and low-power consumption
Realize specific function.
In the present invention, core of the FPGA as information processing will not only simulate I2C bus protocol completes view
Frequency decoding, coding will also realize Image Acquisition control, image procossing rudimentary algorithm, and will figure to video stream synchronization signal extraction
In sram as signal storage.Meanwhile the route planning algorithm of water-surface cleaning device is also to realize in FPGA.As shown in Figure 5
Program flow chart, the present invention mainly to image procossing and path design this two large divisions be unfolded research.
The present invention identified using the unmanned plane in lake surface flying overhead and lake surface rubbish and positioned to rubbish, being detected
To location information send the water-surface cleaning device on lake surface to, then by cleaning device come clearing and retrieving rubbish, to maintain lake
Face cleaning.The present invention is for realizing that the automation of lake surface rubbish cleaning, intelligence have following significance.
1. unmanned plane assists cleaning lake surface rubbish.Previous device for cleaning water surface rubbish has due to the scanning range of itself
Limit, wastes time, the energy, and efficiency is very low, and " eyes " of the present invention application unmanned plane as garbage boat, " visual field " range are wide,
It greatly improves work efficiency.
2. identifying rubbish using exclusive method.On lake surface, biggish object (such as ship of area is also had in addition to floating refuse
Only, sheet of landscape plant lotus etc.), therefore area on lake surface is greater than one square metre of object when identifying rubbish by the present invention
It excludes, remaining foreground object can be determined as floating refuse.The method is succinctly effective, and it is biggish to also eliminated area
The influence that object generates the identification of target rubbish.
3. the path of water-surface cleaning device is designed.The present invention considers the case where there are islands in lake, if water-surface cleaning device
With target rubbish respectively in the two sides of island, then cleaning device cannot be simple line navigation.The present invention is known
In the case where island position, path design is carried out to cleaning device, has improved application range.
4. the path of unmanned plane is designed.Unmanned plane can be autonomous complete in region specified in advance through the planning of way point
It is designed at path, to carry out course.
5. earth station.The present invention includes earth station, the earth station can real-time display unmanned plane pat the picture taken the photograph, and can
UAV Attitude control, setting of navigation area etc. are carried out by the earth station.
It is summarized as follows in conclusion the present invention can extend.
1, in prototype system master-plan, settable structure: unmanned plane+water-surface cleaning device+earth station.
Unmanned plane is whole system " eyes ", is mainly responsible for the acquisition and processing of Surface Picture.Camera obtains image
After need by A/D convert convert analog signals into accessible digital signal, the i.e. discrete processes in space and amplitude.Then
Image gray processing, that is, by luminance quantization, it is divided into 0 to 255 i.e. 256 rank, 0 most dark (completely black), 255 is most bright (complete
It is white).Information after treatment is the gray value of image, but wherein comprising being limited and interference and making an uproar for generating by various conditions
Sound must first carry out image preprocessing before subsequent processing, that is, carry out following three steps:
1. carrying out noise reduction filtering processing first, noise, the influence for preventing noise from analyzing subsequent image are eliminated by differential；
2. since filtering processing meeting is so that soft edge needs again for projecting edge and profile information to image
It is sharpened processing, i.e., useful information is reinforced by integral, so that edge is prominent, clear；
3. detecting and dispelling water surface typical case's disturbing factor.
Since Image Information Processing amount is big, general processor is unable to satisfy the requirement of real-time.This system is in unmanned plane
It is upper to carry airborne computer, image can be quickly handled, and identify after rubbish that send the location information of rubbish to the water surface clear
Clean device.
The most important effect of water-surface cleaning device is the lake surface floating refuse collecting unmanned plane and being found.Cleaning device receives
After the rubbish location information sent to unmanned plane, optimal path will be voluntarily planned, to go to destination to clear up rubbish.Secondly the water surface
There are also some additional functions for cleaning device: the region that can not reached to danger zone or staff carries out lake water and adopts
Sample, for water quality monitoring；Feeding can be carried out to the biology in lake, reduce the workload of administrative staff；Last water surface cleaning
" gas station " of device or unmanned plane, when unmanned plane completes primary cruise or when not enough power supply is to work on, nobody
Machine will charge back to water-surface cleaning device.And all processes are all automatically, are not necessarily to manual operation.
Earth station is the ground surface platform that user is used to understand system health and control system.Earth station can be complete
At unmanned plane navigation area setting, unmanned plane during flying gesture stability, while can also check in real time water-surface cleaning device and
The working condition of unmanned plane.
2, in a practical situation, the lake surface image of acquisition is not ideal, under the influence of various environmental conditions, is also deposited
The processing and analysis of image can be interfered in some cases, therefore the present invention can detect typical disturbing factor simultaneously in image preprocessing
It dispels, water surface typical case's disturbing factor mainly has the following:
1. lake surface may have the floating material of some non-junk, such as some canoes for tourist's amusement, aquatile etc.,
Meeting impacts the rubbish identification of system；
2. there is ripple in the water surface because of situations such as being influenced by natural environment or aquatile movement；
3. if there are scenery or other objects in bank inverted image can be formed in lake surface；
4. there are exposure area or retroreflective regions in the image obtained.
3, the aquatile identified for influencing rubbish, the present invention are dispelled using the method for infrared detecting set detection.Due to
Aquatile and the maximum difference of floating refuse are exactly that biology has vital signs, and the thermoelectric element in infrared detecting set can detecte
To the presence of vital signs, effectively the bioturbation in water can be dispelled with this method.
4, image procossing target identification uses exclusive method
Since the foreground object gone out from image zooming-out not only has target rubbish, there are also the excessive objects of some volumes (such as: ship
Only), therefore the present invention is using exclusive method progress rubbish identification.Exclusive method be set a suitable threshold value, by foreground object into
Its mean radius and threshold value comparison is found out in row processing, if more than threshold value, then it is assumed that the foreground object is not target rubbish, and will
The object excludes, and remaining foreground object can then be determined as rubbish target.In the environment used in this system, row is used
Division identifies that rubbish is succinctly effective.
5, the main information processing of this system is completed on the airborne computer of UAV flight.Since image procossing is believed
Breath amount is bigger, and general common processor can not quickly complete image procossing, rubbish identification in real time.The present invention is in unmanned plane
It is upper to carry airborne computer, requirement of the system to real-time may be implemented, while without ground computer is depended on, information processing can
Directly to be completed on unmanned plane.
6, earth station is the ground surface platform that user is used to understand system health and control system.
1. earth station can complete the setting of unmanned plane navigation area: user can delimit lake surface region in earth station,
Unmanned plane can be with contexture by self cruise path after receiving corresponding location information, and then completes region cruise；
2. unmanned plane during flying gesture stability may be implemented in earth station, the unmanned plane during flying set is set in earth station
Posture, user can select different flight attitudes according to demand.Meanwhile under the premise of guaranteeing unmanned plane normal flight,
User can be with the flight attitude of customized unmanned plane；
3. can see the lake surface picture of unmanned plane shooting in real time in earth station, and it can also look at water surface cleaning
The working condition of device and unmanned plane；
7, the system may also include path design module, and the path of unmanned plane described in the path design module planning is set
Meter, the unmanned plane path design the following steps are included:
Thiessen polygon figure is drawn according to the condition distribution situation of known lake surface；
Maximum broken line is carried out to loke shore profile on this basis, loke shore profile is made to become the inscribed more of a fitting environment
Side shape, each edge of inscribed polygon are a route；
Since the longest edge of inscribed polygon, unmanned plane starts to scan according to Dijkstra's algorithm to lake surface, route
Between spacing be camera scanning range；
A way point is set for unmanned plane at a certain distance on route using differential GPS, way point is corrected, protects
The way point for meeting and sketching the contours of inscribed polygon feature completely is stayed, guarantees that unmanned plane will not conform to during navigation because of distance
It is suitable to lose next way point or deviate track；
Unmanned plane is just navigated by water and is scanned according to revised way point in real navigation.
8, the path planning of water-surface cleaning device uses C space law
When water-surface cleaning device receives location information, need according to certain route navigation to target position.If clear
Clear between the position and target position of clean device, then can line navigation to target position；If there is barrier (such as island),
It then needs to design an optimal navigation path for water-surface cleaning device, target position is reached with bee line.
In lake barrier we using C space law (also known as Visual Graph space law) carry out path design:
Do a polygon round barrier (by taking lake Island as an example) first, the distance of polygon to bank is x, make a <
X <b (selection of wherein a, b make x be unlikely to too big or too small), x herein are that garbage boat and island bump against simultaneously in order to prevent
Be conducive to determine route.It crosses the vertex of polygon, the starting point P of garbage boat and rubbish position Q line and (is not passed through barrier
Line), so that P and Q is connected row with the vertex of polygon respectively into a closed polygon, the side of the polygon forms two
Path M, N compare this two paths length, and short is optimal path (water-surface cleaning device path design as shown in Figure 4
Figure).Water-surface cleaning device just presses the path and reaches target position.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.
Claims (8)
1. a kind of intelligent floating material collection system of water day one, it is characterised in that: it includes image capture module, image procossing mould
Block, device for cleaning water surface rubbish；
Described image acquisition module obtains the image of lake surface for clearance by unmanned plane, and will by A/D conversion by described image
Analog signal is converted to digital signal and realizes the space of described image and the discrete processes of amplitude, then described image gray processing,
Image preprocessing is carried out later；
Lake surface background in described image and foreground object are divided and obtain the foreground object by described image processing module；Identification
Target rubbish in the foreground object: the object by the way that area in the foreground object to be greater than to preset value excludes；To identification
Target rubbish out establishes coordinate using the camera of unmanned plane itself as coordinate origin, and out position is calculated under this coordinate and is sat
Mark；
The device for cleaning water surface rubbish clears up target rubbish according to the position coordinates of target rubbish；
Wherein, the system also includes paths to design module, device for cleaning water surface rubbish described in the path design module planning
Path design, the path design of the device for cleaning water surface rubbish is divided into two parts planning: consideration free space should be than hull ruler
Very little position beyond how many Rough Plannings and stress space it is relatively narrow when ship positioning thin planning；
In Rough Planning apply ant group algorithm, the Rough Planning the following steps are included:
Environmental modeling is carried out, grid division is carried out to entire lake surface: barrier is divided into black square, remaining is white square, less than half
Lattice are more than that half-space is calculated by a lattice, draw all possible route in white square between Origin And Destination by not calculating；
Optimal path is selected, first expanded scope: the multiple grids being connected in environmental modeling being expanded as into a unit, statistics is each
The probability that barrier occurs in unit, calculates the average value of these probability, the range higher than average value is cast out, in subaverage
In the range of continue iteration in aforementioned manners, until finding out optimal path；
Optimal path is smoothed.
2. one intelligent floating material collection system in water day as described in claim 1, it is characterised in that: the system also includes numbers
According to library, the target rubbish that the database purchase identifies, and sorted generalization is carried out according to the feature after target rubbish binaryzation,
The target rubbish of described image acquisition module database described in the image call of acquisition compares, quick obtaining described image
In target rubbish.
3. one intelligent floating material collection system in water day as described in claim 1, it is characterised in that: the system also includes roads
Diameter designs module, and the path design of the path design of unmanned plane described in the path design module planning, the unmanned plane includes
Following steps:
Thiessen polygon figure is drawn according to the condition distribution situation of known lake surface；
Maximum broken line is carried out to loke shore profile on this basis, so that loke shore profile is become being inscribed for a fitting environment polygon
Shape, each edge of inscribed polygon are a route；
Since the longest edge of inscribed polygon, unmanned plane starts to scan according to Dijkstra's algorithm to lake surface, between route
Spacing be camera scanning range；
A way point is set for unmanned plane at a certain distance on route using differential GPS, way point is corrected, retains full
Foot sketches the contours of the way point of inscribed polygon feature completely, guarantees that unmanned plane will not be lost during navigation because of apart from improper
Fall next way point or deviates track；
Unmanned plane is just navigated by water and is scanned according to revised way point in real navigation.
4. one intelligent floating material collection system in water day as described in claim 1, it is characterised in that: apply C in thin planning
Space law, the thin planning are as follows: do a polygon round barrier first, the distance of polygon to loke shore side is x, make a < x <
B crosses the vertex of polygon, the starting point P of garbage boat and rubbish position Q line, line and is not passed through barrier, makes starting point
P and position Q is connected row into a closed polygon with the vertex of polygon respectively, and the side of the closed polygon forms two
Path M, N, compare this two paths length, and short is optimal path.
5. one intelligent floating material collection system in water day as described in claim 1, it is characterised in that: the gray processing of described image
It is by luminance quantization: is divided into 0 to 255 i.e. 256 rank, 0 characterization is most dark or completely black, and 255 characterize most bright or Quan Bai；The figure
As processing module is split described image using gray level threshold segmentation method.
6. one intelligent floating material collection system in water day as described in claim 1, it is characterised in that: described image pretreatment packet
Include step:
(1) noise reduction filtering processing is carried out first, and noise is eliminated by differential；
(2) processing is sharpened to described image again.
7. one intelligent floating material collection system in water day as described in claim 1, it is characterised in that: described image processing module
Lake surface typical case disturbing factor is detected before image segmentation and is dispelled.
8. a kind of intelligent floating material collection method of water day one, it is characterised in that: it is applied to as any in claim 1 to 7
The intelligent floating material collection system of the one of water day described in one；The intelligent floating material collection method of water day one includes following step
It is rapid:
The image in waters for clearance is obtained by unmanned plane shooting, described image is converted analog signals by A/D conversion
Digital signal realizes the space of described image and the discrete processes of amplitude, then described image gray processing, and it is pre- to carry out image later
Processing；
Lake surface background in described image and foreground object are divided and obtain the foreground object；It identifies in the foreground object
Target rubbish: the object by the way that area in the foreground object to be greater than to preset value excludes；To the target rubbish identified, with nothing
It is man-machine itself camera be coordinate origin, establish coordinate, calculate position coordinates under this coordinate；
Target rubbish is cleared up according to the position coordinates of target rubbish.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710806662.1A CN107622231B (en) | 2017-09-08 | 2017-09-08 | A kind of intelligent floating material collection system of water day one and its collection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710806662.1A CN107622231B (en) | 2017-09-08 | 2017-09-08 | A kind of intelligent floating material collection system of water day one and its collection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107622231A CN107622231A (en) | 2018-01-23 |
| CN107622231B true CN107622231B (en) | 2019-11-05 |
Family
ID=61088571
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710806662.1A Active CN107622231B (en) | 2017-09-08 | 2017-09-08 | A kind of intelligent floating material collection system of water day one and its collection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107622231B (en) |
Families Citing this family (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108510124B (en) * | 2018-04-09 | 2020-11-03 | 山东大学 | Water surface garbage cleaning path planning method and device based on machine learning |
| CN108646728B (en) * | 2018-04-09 | 2020-07-31 | 山东大学 | Cloud server-based automatic water surface garbage cleaning method and system |
| CN110443823A (en) * | 2018-05-03 | 2019-11-12 | 西南科技大学 | A kind of floater foreground segmentation method |
| CN108985274B (en) * | 2018-08-20 | 2023-07-14 | 上海磐导智能科技有限公司 | Water surface foreign matter identification method |
| WO2020058905A1 (en) | 2018-09-20 | 2020-03-26 | Cortexia Sa | Method and device for tracking and exploiting at least one environmental parameter |
| CN109392427B (en) * | 2018-10-26 | 2021-01-19 | 大连理工高邮研究院有限公司 | Novel mowing device, system and method for mower |
| CN110414334B (en) * | 2019-06-20 | 2021-05-11 | 浙江大学 | Intelligent water quality identification method based on unmanned aerial vehicle inspection |
| CN110244763B (en) * | 2019-06-21 | 2021-11-02 | 浙江海洋大学 | Remote monitoring system and monitoring method for seawater pollutants |
| CN110320163A (en) * | 2019-06-26 | 2019-10-11 | 深圳市宇驰检测技术股份有限公司 | Online water quality monitoring method, system and the storage medium of unmanned plane |
| CN110758658A (en) * | 2019-11-08 | 2020-02-07 | 湘潭大学 | Method for cleaning water surface floating garbage by using unmanned aerial vehicle and unmanned ship in cooperation |
| CN111079724B (en) * | 2020-03-25 | 2020-06-30 | 速度时空信息科技股份有限公司 | Unmanned aerial vehicle-based sea floating garbage identification method |
| CN111931597A (en) * | 2020-07-20 | 2020-11-13 | 苏州富鑫林光电科技有限公司 | Garbage sorting method and system based on artificial intelligence, storage medium and equipment |
| US11308603B2 (en) | 2020-08-12 | 2022-04-19 | International Business Machines Corporation | Detection and monitoring of waste disposal in an environmental location using artificial intelligence |
| CN112947202B (en) * | 2021-02-26 | 2022-06-14 | 厦门理工学院 | Intelligent cleaning system for water surface floating garbage and control method |
| CN114063640B (en) * | 2021-10-18 | 2024-04-12 | 广州大学 | Water area garbage cleaning method, device, equipment and storage medium |
| CN113985876B (en) * | 2021-10-27 | 2023-09-26 | 广州大学 | Ocean garbage recycling path planning method and system based on bionic fish foraging |
| CN116195912A (en) * | 2021-11-30 | 2023-06-02 | 美智纵横科技有限责任公司 | Control method and device of cleaning robot, electronic equipment and storage medium |
| CN114565635B (en) * | 2022-03-08 | 2022-11-11 | 安徽新宇环保科技股份有限公司 | Unmanned ship system capable of intelligently identifying river channel garbage and performing classified collection |
| CN114545952B (en) * | 2022-04-20 | 2022-07-29 | 季华实验室 | Water surface cleaning method, operation system, device, electronic equipment and storage medium |
| CN115318760B (en) * | 2022-07-29 | 2024-04-16 | 武汉理工大学 | Unmanned aerial vehicle laser cleaning method and system for power transmission tower |
| CN115049825B (en) * | 2022-08-16 | 2022-11-01 | 北京大学 | Water surface cleaning method, device, equipment and computer readable storage medium |
| CN116591131B (en) * | 2023-07-18 | 2023-09-22 | 浙江大学海南研究院 | Garbage collection module and garbage collection device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103542852A (en) * | 2013-10-23 | 2014-01-29 | 上海大学 | Unmanned plane path planning method based on cutting method |
| CN106444773A (en) * | 2016-10-25 | 2017-02-22 | 大连理工大学 | Environment modeling method based on recursive reduced visibility graph |
| CN107092914A (en) * | 2017-03-23 | 2017-08-25 | 广东数相智能科技有限公司 | Refuse classification method, device and system based on image recognition |
| CN107117268A (en) * | 2017-05-12 | 2017-09-01 | 西南科技大学 | The ocean rubbish recovering method and system of a kind of heterogeneous system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9975040B2 (en) * | 2015-06-07 | 2018-05-22 | Apple Inc. | Procedural navigation graphs for identifying a route for a virtual object |
-
2017
- 2017-09-08 CN CN201710806662.1A patent/CN107622231B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103542852A (en) * | 2013-10-23 | 2014-01-29 | 上海大学 | Unmanned plane path planning method based on cutting method |
| CN106444773A (en) * | 2016-10-25 | 2017-02-22 | 大连理工大学 | Environment modeling method based on recursive reduced visibility graph |
| CN107092914A (en) * | 2017-03-23 | 2017-08-25 | 广东数相智能科技有限公司 | Refuse classification method, device and system based on image recognition |
| CN107117268A (en) * | 2017-05-12 | 2017-09-01 | 西南科技大学 | The ocean rubbish recovering method and system of a kind of heterogeneous system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107622231A (en) | 2018-01-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107622231B (en) | A kind of intelligent floating material collection system of water day one and its collection method | |
| CN106816038A (en) | A kind of inland waters abnormal behaviour ship automatic identification system and method | |
| CN108121991A (en) | A kind of deep learning Ship Target Detection method based on the extraction of edge candidate region | |
| CN110188696A (en) | A kind of water surface is unmanned to equip multi-source cognitive method and system | |
| CN109711295A (en) | A kind of remote sensing image offshore Ship Detection | |
| CN109766830A (en) | A kind of ship seakeeping system and method based on artificial intelligence image procossing | |
| Salehi et al. | Automatic moving vehicles information extraction from single-pass WorldView-2 imagery | |
| CN103886285A (en) | Method for detecting optical remote-sensing image ship with assistance of prior geographic information | |
| CN101145200A (en) | Inner river ship automatic identification system of multiple vision sensor information fusion | |
| CN111738112A (en) | Remote sensing ship image target detection method based on deep neural network and self-attention mechanism | |
| CN108090911A (en) | A kind of offshore naval vessel dividing method of remote sensing image | |
| CN111222701A (en) | Marine environment map layer-based automatic planning and evaluation method for ship route | |
| CN111582218A (en) | Remote sensing image island extraction method based on deep learning | |
| CN104992172B (en) | Port remote sensing image shore ship detection method based on coastal outburst and different area scanning | |
| CN111753677A (en) | Multi-angle remote sensing ship image target detection method based on characteristic pyramid structure | |
| CN109919026A (en) | A kind of unmanned surface vehicle local paths planning method | |
| CN107731011A (en) | A kind of harbour is moored a boat monitoring method, system and electronic equipment | |
| Zhang et al. | Research on unmanned surface vehicles environment perception based on the fusion of vision and lidar | |
| Zhang et al. | Nearshore vessel detection based on Scene-mask R-CNN in remote sensing image | |
| CN113723371B (en) | Unmanned ship cleaning route planning method and device, computer equipment and storage medium | |
| CN108446721A (en) | A kind of aquatic environment analytic method based on multiple-limb network | |
| Wang et al. | Accurate real-time ship target detection using Yolov4 | |
| CN113159042A (en) | Laser vision fusion unmanned ship bridge opening passing method and system | |
| US20230306730A1 (en) | Information extraction method of offshore raft culture based on multi-temporal optical remote sensing images | |
| CN114743059B (en) | Automatic classification method for submarine geographic entities by integrating topographic features |
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 |