CN104809350A - Method for distinguishing marine litter from bion - Google Patents

Abstract

The invention provides a method for distinguishing marine litter from bion. The method comprises the following steps: establishing a learning sample; performing Markov model learning based on the learning sample; performing classified judgment of the marine litter and the bion on a detected object through the Markov model obtained by learning, wherein the step of performing classified judgment of the marine litter and bion on the detected object through the markov model obtained by learning specifically comprises the following steps: performing preliminary orientation on the detected object; tracing the position of the detected object within a period of time to obtain moving parameters of the detected object at each monitoring moment, and forming a moving parameter varying sequence; calculating coinciding degree between the obtained moving parameter varying sequence and the Markov model obtained by learning; when the coinciding degree obtained by calculating is greater than or equal to a set threshold value, judging that the detected object is marine litter.

Description

The differentiating method of ocean rubbish and bion

Technical field

The present invention relates to marine eco-environment detection technique field, and in particular to the differentiating method of a kind of ocean rubbish and bion.

Background technology

Three kinds of technology of the quantity of current assessment floating refuse, kind and distribution: the article monitoring method of floating refuse " in the marine environment " from periodical " marine environment science " the 16th volume the 2nd phase describes following three kinds of technology:

1. statistician by inquiry passing ships estimate kind and the quantity of floating refuse.The rubbish quantity that the method is daily jettisoninged according to statistician, after carrying out the mathematical statistics of being correlated with, draws the quantity of the average annual floating refuse in certain marine site.Because the method needs the statistician collecting enough information and specialty, because the individuality of statistician there are differences, the conclusion therefore come out has larger difference.

2. density and the type of floating refuse is determined by field investigation.(1) use aircraft to carry out investigation observation to a certain marine site floating refuse, collect the data of the distribution of relevant floating refuse and quantity; (2) jointly carry out by other marine monitoring ship, pleasure-boat or freighter, finite observation is carried out in the course line according to ship.Two kinds of modes all require suitable sea situation (<3 level) and good visibility, and the former confidence level is high and the latter's expense is low.

3. the situation of artificial beach observation monitoring floating refuse.Because the fraction floats rubbish in ocean is subject to the impact of wind direction and ocean current, not necessarily float to bank completely, the method makes overall estimate amount on the low side.

To sum up, the detection of floating refuse also lacks economy, effective method; And the research and practice carrying out being correlated with is needed in the detection of sea low suspension rubbish and seabed rubbish badly.

Summary of the invention

There is accuracy of detection and economic cost to have much difficulty in healing the problem of weighing apparatus to overcome existing ocean garbage detection technique in the present invention, provides the differentiating method of a kind of economy, effective ocean rubbish and bion.

To achieve these goals, the differentiating method that the invention provides a kind of ocean rubbish and bion comprises:

Set up study sample;

Markov model study is carried out based on study sample;

With the Markov model learning to obtain, the classification that detecting object carries out ocean rubbish and bion is judged, specifically comprises:

Primary Location is carried out to detecting object;

Within a period of time, follow the tracks of the position of detecting object, obtain the moving parameter of each monitoring moment detecting object, and form moving parameter change sequence;

Calculate the moving parameter change sequence that obtains and study obtain the degree of agreement of Markov model;

When calculating gained degree of agreement and being more than or equal to setting threshold value, judge that detecting object is ocean rubbish.

In one embodiment of the invention, the step setting up study sample comprises:

Primary Location is carried out to the multiple sample object within the scope of monitoring marine site;

Within a period of time, follow the tracks of the position of each sample object of detection, obtain the moving parameter of each monitoring moment object, and form moving parameter change sequence sample;

Manually mark the multiple sample object within the scope of monitoring marine site, multiple sample object of identification are ocean rubbish or bion.

In one embodiment of the invention, moving parameter is translational speed and moving direction change.

In one embodiment of the invention, the acquisition step of moving parameter comprises:

Within a period of time, follow the tracks of the position by the detecting object after being tentatively decided to be or sample object, measure detecting object or the translational speed in the sample object unit interval;

Calculate the detecting object in each monitoring moment or the moving direction change of sample object.

In one embodiment of the invention, carry out Markov model study based on study sample, comprise and calculate the transition probability of each monitoring moment detecting object translational speed and the transition probability of moving direction change.

In one embodiment of the invention, be according to detecting object or sample object signal wave to the Primary Location of detecting object or the Primary Location of sample object, positioned by solid region localization method.

In one embodiment of the invention, solid region localization method is the solid region localization method based on " ellipsoidal cavity " model.

In one embodiment of the invention, the solid region localization method based on " ellipsoidal cavity " model comprises:

Any node of ultrasonic sensor launches ultrasonic signal, and other node receives the ultrasound wave that the ultrasound wave and detecting object directly launched or sample object reflect;

Include detecting object or sample object distance launching site closest approach and solstics respectively and rotate two ellipsoids formed around launching site and any acceptance point line, occuring simultaneously and form " ellipsoidal cavity ";

Change acceptance point to obtain more " ellipsoidal cavity ", carry out common factor between " ellipsoidal cavity " and draw barycenter;

The barycenter of gained forms barycenter group, and obtain the barycenter of barycenter group, described barycenter is the position of detecting object or sample object.

In sum, the differentiating method of ocean provided by the invention rubbish and bion compared with prior art, has the following advantages:

Because the biology in ocean has autonomous mobility, therefore their state change does not have rule governed.But rubbish does not have an autonomous mobility in ocean, therefore their state change is that to have certain rule governed.The present invention is by under the hydrological characteristics of specified sea areas, by constantly measuring sampling, record ocean in each sample object not in the same time under moving parameter, form moving parameter change sequence sample, the moving parameter change sequence sample of multiple sample defines study sample, and then carries out Markov model study based on study sample.In subsequent detection, by measuring the moving parameter of each moment detecting object, and form moving parameter change sequence, judge that detecting object is ocean rubbish or bion by calculating the moving parameter change sequence formed in the degree of agreement of Markov model.

The differentiating method of ocean provided by the invention rubbish and bion is a kind of differentiating method based on Markov model, has very high differentiation precision.Meanwhile, differentiating method can come by means of computing machine, not only has very high differentiation efficiency, has extremely low differentiation cost simultaneously, effectively solves existing ocean rubbish detection method accuracy of detection and funds and to have much difficulty in healing the problem of weighing apparatus.

For above and other objects of the present invention, feature and advantage can be become apparent, preferred embodiment cited below particularly, and coordinate accompanying drawing, be described in detail below.

Accompanying drawing explanation

Figure 1 shows that the process flow diagram of the ocean rubbish that one embodiment of the invention provides and bion differentiating method.

Figure 2 shows that the process flow diagram setting up study sample in Fig. 1.

Figure 3 shows that in Fig. 1 with the process flow diagram that the Markov model learning to obtain judges the classification that detecting object carries out ocean rubbish and bion.

Embodiment

As shown in Figure 1, the differentiating method of the ocean rubbish that provides of the present embodiment and bion comprises:

Step S10, foundation study sample, comprising:

Step S11, to monitoring marine site within the scope of multiple sample object carry out Primary Location.

In the present embodiment, the solid region localization method based on " ellipsoidal cavity " model is adopted to carry out Primary Location to sample object.But the present invention is not limited in any way this.In other embodiment, other locator meamss such as node localization in wireless sensor networks can be adopted to position.

The positioning principle of " ellipsoidal cavity " model is:

With 3 × 3 arrays $\left[\begin{array}{ccc}A& J& L\\ B& J& M\\ C& K& M\end{array}\right]$ For example: with seashore horizontal line for Z axis positive dirction is Y-axis positive dirction straight down normal vector direction perpendicular to emission array is X-axis positive dirction.Be emitted as example with I point, first get I (x ₁, y ₁, z ₁), J (x ₂, y ₂, z ₂), K (x ₃, y ₃, z ₃), B (x ₄, y ₄, z ₄) be reference point, O (x, y, z) is for object under test is apart from launching site I (x ₁, y ₁, z ₁) closest approach, and P (x ', y ', z ') be launching site I (x ₁, y ₁, z ₁) accessible point farthest, J point receives signal moment that I point launches and the signal moment that O point reflection is returned is respectively t ₁and t ₁', K point receives signal moment that I point launches and O point reflection signal moment of returning is respectively t ₂and t ₂', B point receives signal moment that I point launches and O point reflection signal moment of returning is respectively t ₃and t ₃'; Then:

d _ioj＝v·(t ₁′-t ₁)+d _ij，

d _iok＝v·(t ₂′-t ₂)+d _ik(1-1)

d _iob＝v·(t ₃′-t ₃)+d _ib

Then with I, J for focus, O is that on ellipsoid, any point can obtain ellipsoid TQ _iOJequation is:

$\sqrt{{(x-x_1)}^2+{(y-y_1)}^2+{(z-z_1)}^2}+\sqrt{{(x-x_2)}^2+{(y-y_2)}^2+{(z-z_2)}^2}=d_{\mathrm{ioj}}---(1-2)$

In like manner ellipsoid TQ _iOKwith ellipsoid TQ _iOBequation is respectively:

$\sqrt{{(x-x_1)}^2+{(y-y_1)}^2+{(z-z_1)}^2}+\sqrt{{(x-x_3)}^2+{(y-y_3)}^2+{(z-z_3)}^2}=d_{\mathrm{iok}}---(1-3)$

$\sqrt{{(x-x_1)}^2+{(y-y_1)}^2+{(z-z_1)}^2}+\sqrt{{(x-x_4)}^2+{(y-y_4)}^2+{(z-z_4)}^2}=d_{\mathrm{ioj}}---(1-4)$

Simultaneous equations (1-2), (1-3), (1-4) can obtain O (x, y, z) coordinate.

In like manner, for an I (x ₁, y ₁, z ₁), J (x ₂, y ₂, z ₂), K (x ₃, y ₃, z ₃), B (x ₄, y ₄, z ₄) be reference point, farthest accessible point P (x ', y ', z ') on ellipsoid a bit, J point receives the signal moment that signal moment of I point transmitting and P point reflection return and is respectively t ₄and t ₄', K point receives signal moment that I point launches and P point reflection signal moment of returning is respectively t ₅and t ₅', B point receives signal moment that I point launches and P point reflection signal moment of returning is respectively t ₆and t ₆', then can obtain reflection paths length and be respectively:

d _ipj＝v·(t ₄′-t ₄)+d _ij

d _ipk＝v·(t ₅′-t ₅)+d _ik(1-5)

d _ipb＝v·(t ₆′-t ₆)+d _ib

Based on the ellipsoid TQ of 3 _iPJ, TQ _iPKand TQ _iPBequation is respectively:

$\sqrt{{(x^{\&prime;}-x_1)}^2+{(y^{\&prime;}-y_1)}^2+{(z^{\&prime;}-z_1)}^2}+\sqrt{{(x^{\&prime;}-x_2)}^2+{(y^{\&prime;}-y_2)}^2+{(z^{\&prime;}-z_2)}^2}=d_{\mathrm{ipj}}---(1-6)$

$\sqrt{{(x^{\&prime;}-x_1)}^2+{(y^{\&prime;}-y_1)}^2+{(z^{\&prime;}-z_1)}^2}+\sqrt{{(x^{\&prime;}-x_3)}^2+{(y^{\&prime;}-y_3)}^2+{(z^{\&prime;}-z_3)}^2}=d_{\mathrm{ipk}}---(1-7)$

$\sqrt{{(x^{\&prime;}-x_1)}^2+{(y^{\&prime;}-y_1)}^2+{(z^{\&prime;}-z_1)}^2}+\sqrt{{(x^{\&prime;}-x_4)}^2+{(y^{\&prime;}-y_4)}^2+{(z^{\&prime;}-z_4)}^2}=d_{\mathrm{ipb}}---(1-8)$

Simultaneous equations (1-6), (1-7), (1-8) can obtain P (x ', y ', z ') coordinate.

Any focus is identical, and two ellipsoidal surfaces a bit formed with O, P surface, as inside and outside ellipsoid, namely form one " ellipsoidal cavity ", are respectively: (TQ as above-mentioned equation can obtain three " ellipsoidal cavities " _iOJ, TQ _iPJ), (TQ _iOK, TQ _iPK) and (TQ _iOB, TQ _iPB), then can obtain the common factor of three " ellipsoidal cavities ", and obtain barycenter, be designated as Q (X ₁, Y ₁, Z ₁).

Q (X ₁, Y ₁, Z ₁) be the position of sample object.

Step S12, within a period of time, follow the tracks of the position of each sample object of detection, obtain the moving parameter of each monitoring moment object, and form moving parameter change sequence sample.In the present embodiment, moving parameter is translational speed and the moving direction change of sample object.The step obtaining this moving parameter is:

If a certain sample object is Q (x, y, z) at not three-dimensional coordinate in the same time.As T ₀the coordinate of this sample object of moment is Q (x ₀, y ₀, z ₀), T _ithe coordinate in moment is Q (x _i, y _i, z _i).Follow the tracks of by the position of the sample object after being tentatively decided to be within a period of time, measure the translational speed in the sample object unit interval.Specifically, at T _i-1to T _ithe displacement of time object is:

$D_{i-1,i}=\sqrt{{(x_i-x_{i-1})}^2+{(y_i-y_{i-1})}^2+{(z_i-z_{i-1})}^2}---\left(1\right)$

T _i-1to T _ithe average translational speed of time object O:

V _i-1,i＝D _i-1,i/(T _i-T _i-1) (2)

Known sample object is at the average velocity v of different time sections ₀, v ₁, v ₂...Known T ₀, T ₁and T ₂the coordinate of moment sample object, then can represent the moving direction of sample object with vector:

$\stackrel{\&RightArrow;}{O_0{O}_1}=(x_1-x_0,y_1{-y}_0,z_1-z_0)---\left(3\right)$

$\stackrel{\&RightArrow;}{O_1{O}_2}=(x_2-x_1,y_2-y_1,z_2-z_1)$

Two vectorial angles are:

${\mathrm{\&alpha;}}_0={\cos }^{-1}\frac{\stackrel{\&RightArrow;}{O_0{O}_1}\&CenterDot;\stackrel{\&RightArrow;}{O_1{O}_2}}{\left|\stackrel{\&RightArrow;}{O_0{O}_1}\right|*\left|\stackrel{\&RightArrow;}{O_1{O}_2}\right|}---\left(4\right)$

Then in the unit interval, the situation of change of the moving direction of sample object can use following equation expression:

wherein t ₀=(T ₂-T ₀)/2 (5)

Step S13, to monitoring marine site within the scope of multiple sample object manually mark, multiple sample object of identification are ocean rubbish or bion.Each sample object forms moving parameter change sequence sample < (V ₁, V ₂..., V _n), (M ₁, M ₂..., M _n) >, set up study sample.Wherein V _ibe the translational speed of the i-th moment sample object, M _iit is the moving direction change of the i-th moment sample object.

Step S20, based on study sample carry out Markov model study.Specifically:

First, if V _maxfor maximum speed, V _minfor minimum speed, V _avgfor average velocity.Meter state α ₁for high velocity α ₂for middling speed district α ₃for low regime according to the record of each ocean rubbish, calculate the probability matrix of each State Transferring $A=\left[\begin{array}{ccc}{a}_{11}& a_{12}& a_{13}\\ a_{21}& a_{22}& a_{23}\\ a_{31}& a_{32}& a_{33}\end{array}\right],$ Wherein a _ijfor state α _itransfer is α _jprobability, as a ₁₂for state α ₁transfer to state α ₂probability.The present embodiment divides three velocity bands.But the present invention is not construed as limiting this.

Then, if M _maxfor the change of maximum direction, M _minfor the change of minimum direction, M _avgfor mean direction change.Meter state β ₁for Rapid Variable Design district β ₂for middling speed variation zone β ₃for low speed variation zone according to the record of each ocean rubbish, calculate the probability matrix of each State Transferring $B=\left[\begin{array}{ccc}{b}_{11}& b_{12}& b_{13}\\ b_{21}& b_{22}& b_{23}\\ b_{31}& b_{32}& b_{33}\end{array}\right],$ Wherein b _ijfor state β _itransfer is β _jprobability.The present embodiment divides three direction region of variation.But the present invention is not construed as limiting this.

Finally, sample object velocity series degree of agreement threshold value P is calculated _s1with direction change sequence degree of agreement threshold value P _s2.

P _s1account form as follows:

If V _i∈ α _k, V _j∈ α _l, then PV (V _i, V _j)=a _kl.(6)

Velocity series degree of agreement threshold value $P_{S1}=\underset{i=1}{\overset{n-1}{\mathrm{\&Pi;}}}\mathrm{PV}(V_i,V_{i+1})---\left(7\right)$

Wherein, k, l=1,2,3; PV (V _i, V _j) be V _iaffiliated state α _ktransfer to V _jaffiliated state α _lprobability.

P _s2account form as follows:

If M _i∈ β _k, M _j∈ β _l, then PM (M _i, M _j)=b _kl.(8)

Direction change sequence degree of agreement threshold value $P_{S2}=\underset{i=1}{\overset{n-1}{\mathrm{\&Pi;}}}\mathrm{PM}(M_i,M_{i+1})---\left(9\right)$

Wherein, k, l=1,2,3; PM (M _i, M _j) be M _iaffiliated state β _ktransfer to M _jaffiliated state β _lprobability.

Step S30, with learn obtain Markov model to detecting object carry out ocean rubbish and bion classification judge, specifically comprise:

Step S31, Primary Location is carried out to detecting object.This step is identical with step S11, adopts the solid region localization method based on " ellipsoidal cavity " model to position detecting object, obtains Q _t(X ₁, Y ₁, Z ₁).

Step S32, within a period of time, follow the tracks of the position of detecting object, obtain the moving parameter of each monitoring moment detecting object, and form moving parameter change sequence.This step is identical with step S12, and the moving parameter change sequence formed after formula (1)-(2)-(3)-(4)-(5) calculate is < (V _t1, V _t2..., V _tn), (M _t1, M _t2..., M _tn) >.Wherein V _tifor detecting object T is in the translational speed in the i-th moment, M _tifor detecting object T is in the moving direction change of the i-th moment.

The moving parameter change sequence < (V that step S33, calculating obtain _t1, V _t2..., V _tn), (M _t1, M _t2..., M _tn) > and study obtain the degree of agreement of Markov model, comprise velocity series degree of agreement P _{speed is coincide}and P _{direction change coincide}.Wherein P _{speed is coincide}by sequence (V _t1, V _t2..., V _tn) substitute into formula (6) and (7) obtain afterwards; P _{direction change coincide}by sequence (M _t1, M _t2..., M _tn) substitute into formula (8) and (9) obtain afterwards.

Step S34, when calculating gained degree of agreement and being more than or equal to setting threshold value, judge that detecting object is ocean rubbish.Specifically, P is worked as _{speed is coincide}>=P _s1and P _{direction change coincide}>=P _s2, then this object is judged as being ocean rubbish.

In sum, the differentiating method of ocean provided by the invention rubbish and bion compared with prior art, has the following advantages:

Because the biology in ocean has autonomous mobility, therefore their state change does not have rule governed.But rubbish does not have an autonomous mobility in ocean, therefore their state change is that to have certain rule governed.The present invention is by under the hydrological characteristics of specified sea areas, by constantly measuring sampling, record ocean in each sample object not in the same time under moving parameter, form moving parameter change sequence sample, the moving parameter change sequence sample of multiple sample defines study sample, and then carries out Markov model study based on study sample.In subsequent detection, by measuring the moving parameter of each moment detecting object, and form moving parameter change sequence, judge that detecting object is ocean rubbish or bion by calculating the moving parameter change sequence formed in the degree of agreement of Markov model.

The differentiating method of ocean provided by the invention rubbish and bion is a kind of differentiating method based on Markov model, has very high differentiation precision.Meanwhile, differentiating method can come by means of computing machine, not only has very high differentiation efficiency, has extremely low differentiation cost simultaneously, effectively solves existing ocean rubbish detection method accuracy of detection and funds and to have much difficulty in healing the problem of weighing apparatus.

Although the present invention discloses as above by preferred embodiment; but and be not used to limit the present invention, anyly know this those skilled in the art, without departing from the spirit and scope of the present invention; can do a little change and retouching, therefore protection scope of the present invention is when being as the criterion depending on claims scope required for protection.

Claims (8)

1. a differentiating method for ocean rubbish and bion, is characterized in that, comprising:

Set up study sample;

Markov model study is carried out based on study sample;

With the Markov model learning to obtain, the classification that detecting object carries out ocean rubbish and bion is judged, specifically comprises:

Primary Location is carried out to detecting object;

Within a period of time, follow the tracks of the position of detecting object, obtain the moving parameter of each monitoring moment detecting object, and form moving parameter change sequence;

Calculate the moving parameter change sequence that obtains and study obtain the degree of agreement of Markov model;

When calculating gained degree of agreement and being more than or equal to setting threshold value, judge that detecting object is ocean rubbish.

2. the differentiating method of ocean according to claim 1 rubbish and bion, is characterized in that, the described step setting up study sample comprises:

Primary Location is carried out to the multiple sample object within the scope of monitoring marine site;

Within a period of time, follow the tracks of the position of each sample object of detection, obtain the moving parameter of each monitoring moment object, and form moving parameter change sequence sample;

Manually mark the multiple sample object within the scope of monitoring marine site, the multiple sample object described in identification are ocean rubbish or bion.

3. the differentiating method of ocean according to claim 1 and 2 rubbish and bion, is characterized in that, described moving parameter is translational speed and moving direction change.

4. the differentiating method of ocean according to claim 3 rubbish and bion, is characterized in that, the acquisition step of described moving parameter comprises:

Within a period of time, follow the tracks of the position by the detecting object after being tentatively decided to be or sample object, measure detecting object or the translational speed in the sample object unit interval;

Calculate the detecting object in each monitoring moment or the moving direction change of sample object.

5. the differentiating method of ocean according to claim 3 rubbish and bion, it is characterized in that, describedly carry out Markov model study based on study sample, comprise the transition probability calculating each monitoring transition probability of moment detecting object translational speed and moving direction change.

6. the differentiating method of ocean according to claim 1 and 2 rubbish and bion, it is characterized in that, be according to detecting object or sample object signal wave to the Primary Location of detecting object or the Primary Location of sample object, positioned by solid region localization method.

7. the differentiating method of ocean according to claim 6 rubbish and bion, is characterized in that, described solid region localization method is the solid region localization method based on " ellipsoidal cavity " model.

8. the differentiating method of ocean according to claim 7 rubbish and bion, is characterized in that, the described solid region localization method based on " ellipsoidal cavity " model comprises:

Any node of ultrasonic sensor launches ultrasonic signal, and other node receives the ultrasound wave that the ultrasound wave and detecting object directly launched or sample object reflect;

Include detecting object or sample object distance launching site closest approach and solstics respectively and rotate two ellipsoids formed around launching site and any acceptance point line, occuring simultaneously and form " ellipsoidal cavity ";

Change acceptance point to obtain more " ellipsoidal cavity ", carry out common factor between " ellipsoidal cavity " and draw barycenter;

The barycenter of gained forms barycenter group, and obtain the barycenter of barycenter group, described barycenter is the position of detecting object or sample object.