CN104297814B - A kind of machine perspective vision implementation method - Google PatentsA kind of machine perspective vision implementation method Download PDF
- Publication number
- CN104297814B CN104297814B CN201410570417.1A CN201410570417A CN104297814B CN 104297814 B CN104297814 B CN 104297814B CN 201410570417 A CN201410570417 A CN 201410570417A CN 104297814 B CN104297814 B CN 104297814B
- Prior art keywords
- 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.)
- 238000001514 detection method Methods 0.000 claims abstract description 39
- 239000000463 material Substances 0.000 claims abstract description 21
- 238000005259 measurement Methods 0.000 claims abstract description 8
- 238000001228 spectrum Methods 0.000 claims description 22
- 238000002044 microwave spectrum Methods 0.000 claims description 12
- 239000000203 mixture Substances 0.000 claims description 8
- 230000003595 spectral Effects 0.000 claims description 8
- 239000004615 ingredient Substances 0.000 claims description 4
- 238000004422 calculation algorithm Methods 0.000 claims description 3
- 210000003491 Skin Anatomy 0.000 claims 1
- 230000035515 penetration Effects 0.000 abstract description 5
- 238000004364 calculation method Methods 0.000 abstract description 4
- 239000010410 layer Substances 0.000 description 55
- 239000011159 matrix material Substances 0.000 description 11
- 239000002689 soil Substances 0.000 description 7
- 229910052500 inorganic mineral Inorganic materials 0.000 description 6
- 238000000034 method Methods 0.000 description 6
- 239000011707 mineral Substances 0.000 description 6
- 239000002344 surface layer Substances 0.000 description 6
- 238000005481 NMR spectroscopy Methods 0.000 description 4
- 239000000919 ceramic Substances 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 239000000126 substance Substances 0.000 description 4
- 230000000875 corresponding Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 229910052742 iron Inorganic materials 0.000 description 3
- 238000002083 X-ray spectrum Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 2
- 239000002131 composite material Substances 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 238000009114 investigational therapy Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000002285 radioactive Effects 0.000 description 2
- 239000000523 sample Substances 0.000 description 2
- 230000002792 vascular Effects 0.000 description 2
- 206010028980 Neoplasm Diseases 0.000 description 1
- 210000004940 Nucleus Anatomy 0.000 description 1
- 240000007594 Oryza sativa Species 0.000 description 1
- 235000007164 Oryza sativa Nutrition 0.000 description 1
- 240000004678 Panax pseudoginseng Species 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 229940105648 Soma Drugs 0.000 description 1
- 229910000831 Steel Inorganic materials 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000003466 anti-cipated Effects 0.000 description 1
- 125000004429 atoms Chemical group 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005315 distribution function Methods 0.000 description 1
- 239000000428 dust Substances 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 238000005755 formation reaction Methods 0.000 description 1
- 238000005194 fractionation Methods 0.000 description 1
- 235000005035 ginseng Nutrition 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000010438 heat treatment Methods 0.000 description 1
- 229910052739 hydrogen Inorganic materials 0.000 description 1
- 239000001257 hydrogen Substances 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000002595 magnetic resonance imaging Methods 0.000 description 1
- 238000001225 nuclear magnetic resonance method Methods 0.000 description 1
- 238000002203 pretreatment Methods 0.000 description 1
- 235000009566 rice Nutrition 0.000 description 1
- 238000004805 robotic Methods 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000001340 slower Effects 0.000 description 1
- 238000004611 spectroscopical analysis Methods 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
- 238000001931 thermography Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
The invention discloses a kind of machine perspective vision implementation methods, by one object of multiband Based Microwave Radiometer Detection for the bright temperature of different frequency wave band after carrying out essential information identification, hierarchical mode combination Given information and a priori assumption are relied on again, come the density or temperature or material of inverting interior of articles, since the microwave of different frequency has different penetration depths, and the bright temperature and Density Distribution of each frequency, temperature distribution and dielectric parameter existence function relationship, therefore it is used to through part Given information and measurement come inverting unknown message, and constraint is realized by a-priori.Channel quantity should be no less than the quantity of parameter to be measured, do not have object layered for some, it is geometrically first being split as one by one there is object layered to give inverting respectively, to reduce calculation amount, the database of relevant density and dielectric constant etc. is established to familiar object, by to object first preliminary identification, its known parameters of search inquiry, then inverting.
The invention belongs to machine vision and microwave technical field more particularly to a kind of remote non-contact acquisition interior of articles The technical method of information.
About machine and robotic vision, predominantly stay in the past it is bionical on the basis of, be dedicated to allowing robot can The image of object is obtained by camera, is moved and is identified.
But this kind of vision means and equipment are only limitted to visible light or infrared range for the spectrum sensing range of object, The wavelength of these light is very short, and penetration depths are very low, it is seen that substantially without penetration depths, infrared light is also only capable of reaching in object light About one millimeter of depth distance under surface, therefore these methods cannot obtain the information of interior of articles.
Past needs special X-ray method or nuclear magnetic resonance method for the acquisition of interior of articles information.It is specific to introduce It is as follows.
The first is medically widely applied X-ray, and according to duality principle principle, X-ray energy is high, corpuscular property By force, there is very high energy and drill through ability.The method for generating X-ray is with the electronic impact metallic target after acceleration.Knockout process In, electronics slows down suddenly, and the kinetic energy of loss can be released with form of photons, forms the continuous part of X-ray spectrum, referred to as makes Dynamic radiation.By increasing acceleration voltage, the energy that electronics carries increases, it is likely that the inner electron of metallic atom is knocking-on. Then internal layer forms hole, and outer-shell electron transition returns internal layer and fills up hole, while releasing wavelength in the photon of 0.1 ran. Since the energy that outer-shell electron transition is released is quantized, so the wavelength for the photon released also concentrates on certain parts, shape At the characteristic curve in X-ray spectrum, referred to herein as characteristic radiation.However, X-ray has that three cannot be used for machine vision:
1. big to organism damage, general human body, which repeatedly receives X-ray i.e., larger danger.
2. reception pattern is detection through light, need to have receiving screen over there, it is therefore desirable to special X-ray room, it can not For robot to the observation vision of object.
3. must actively emit active probe, equipment be difficult to minimize, and normal object will not in nature daily life Actively emit X-ray.
Second is Magnetic resonance imaging.It utilizes nuclear magnetic resonance (nuclear magnetic resonance, abbreviation NMR) Principle, the decaying different in different structure environment inside the substance according to the energy that is discharged, is detected by additional gradient magnetic The electromagnetic wave launched can be depicted as in object accordingly it can be learnt that constituting the nuclear position of this object and type The structural images in portion.But nuclear magnetic resonance is needed to be placed into object in special system and be scanned.Its imaging system is one huge Big cylindric machine can manufacture the environment of a strong field regions around subject, hit by the pulse of radio wave The hydrogen nuclei in soma is hit, the arrangement of hydrogen atom in body is changed, when hydrogen atom is again introduced into position arrangement appropriate When, radio signals can be issued, this signal by computer reception and analyzed and conversion process, can be by body make-up and device Hydrogen atom activity in official, is converted into 2D image.Therefore, robot vision can not be also applied to, because of observed object root It is not controlled by robot originally, for observing building, can not need for the building to be placed into robot for observation. The equipment cost and use cost of nuclear magnetic resonance are also all very high.
In fact other than both the above means, low frequency microwave remote sensing also has penetrability.Microwave remote sensing mainly utilizes Infrared beam is projeced into body surface, and the microwave wavelength being reflected by it back changes and colour cast determines its size, form and movement The technology of speed.Various objects can generate reflection to external microwave signal on the earth, and object itself also has microwave spoke Penetrate ability.Microwave remote sensing be exactly the reflection of microwave or the microwave radiation of itself are come by detecting object perceptible volume morphing and Structure organization.
According to the difference of operation principle, microwave remote sensing equipment can be divided into two major classes, and one kind is active remote sensing device, that is, detect Equipment actively emits microwave signal, and the signal for then receiving detected target reflection or scattering again perceives detected target, Working principle is similar to known radar.Another kind of is passive remote sensing device, and working principle is similar to spectral detection device, i.e., Using highly sensitive microwave receiving technology, the faint microwave signal of target itself is received.Active at present in the world is micro- Wave remote sensor has microwave scatterometer, microwave altimeter, synthetic aperture radar (SAR) and real aperture radar etc., passive remote sensing device Mainly microwave radiometer.
In the practice of Satellite microwave remote sensing in recent years it has been realized that in the case where moisture content is low, low frequency microwave energy Through vegetation, ice and snow and dry sandy soil, to obtain soil information below near the ground.
However this characteristic is not used specially at present, also can not be applied to daily machine vision certainly. Because microwave remote sensing is all based on aerial remote sensing aeroplane or remote sensing satellite, mainly some wave band by obtaining at present The standard pictures of remote sensing images and some reference substances under the wave band compare, to identify the object type on remote sensing images Type has the following problems: 1. microwaves obtain bright temperature image be difficult to explain, microwave obtain bright temperature it is related to a large amount of factors, Material, density, specific heat, dielectric constant etc., therefore in general remote sensing, only it is used to bright temperature as one and standardized object pair The parameter of ratio is compared by the standard picture of the remote sensing images and some reference substances under the wave band of some wave band of acquisition, It to identify the object type on remote sensing images, for example is vegetation or soil, numerical value itself does not have specific physical significance, And in daily life, such comparison is just unnecessary, because visible light just has been able to tell object；2. microwave is distant Although it have been found that the available following information near the ground of microwave in sense, but the information is not anticipated substantially for remote sensing Justice, remote sensing object is substantially land and ocean, either soil or ocean lake with the presence of a large amount of water, and water (humidity, water Steam) microwave can be absorbed significantly, so that its investigation depth is no more than several centimeters in satellite remote sensing, and several lis of soil and ocean Substantially without difference, main purposes is at most exactly to penetrate the ice and snow covered, sand and dust, careless vegetation to obtain soil for rice depth and surface layer Earth type information using faciostenosis, therefore does not develop related art method accordingly.
Summary of the invention
In view of the above-mentioned problems, proposing a kind of new technical solution in the present invention, set for the penetration capacity of low frequency microwave A kind of new equipment is counted, true physical quantity is analyzed by the technical program by the data detected.
A kind of machine perspective vision implementation method carries out preliminary detection to detected object by visible or infrared light and knows Not, essential information is obtained, and measures the different frequency of object under test transmitting or reflection by the microwave spectrum detection instrument of multi-frequency Microwave, obtain an object for the bright temperature of different frequency wave band, in conjunction with other Given informations or model；
The Density Distribution or Temperature Distribution or material structure information of inverting interior of articles are by having a multiband battle array The microwave spectrum detection instrument of column, wave band should be staggered as far as possible, more than one camera, a calculating equipment, the attachment network equipment, The workflow of system are as follows: the appearance or surface information of object are obtained by camera, then scan for according to this partial information or By the network equipment to data base querying, to obtain the given data of object, and by multiwave microwave spectrum detection instrument with The mode for receiving the microwave of object under test transmitting or reflection obtains object for the bright temperature of different-waveband, is then pressed by calculating equipment The density or Temperature Distribution of the inside of corresponding layered structure and inverting object are divided into according to above-mentioned method.
It is further limited as the present invention, this method uses following algorithm in inverting: object being first assumed to be one The layer of a subdivision does not have object layered for some, it is geometrically first being split as having layering knot one by one The basic object of structure gives inverting respectively again, then the bright temperature of the object of the different frequency wave band received be these subdivision layer for The superposition of the contribution of the bright temperature of different frequency, each layer of contribution can be expressed as each layer temperature multiplied by contribution proportion function, contribution Proportion function be the density of this layer, dielectric constant, thickness function, according to equation group " receive for different frequency wave band Bright temperature TB=subdivision layer for different frequency wave band bright temperature superposition=each layer physical temperature Ti multiplied by each layer contribution ratio The information such as the physical temperature or density of example each layer of Fi " inverting.
It is further limited as the present invention, this method also uses a priori assumption, is provided by Given information to inverting One assumption value of parameter, then carries out inverting for a priori assumption together with actual measured results, provides to inverted parameters Most probable solution.
It is further limited as the present invention, this method also establishes relevant database to familiar object, includes common object The dielectric constant or density of body or than thermal information, directly to be inquired in inverting.
It is further limited as the present invention, this method also first passes through detection method before inverting, by visible light, infrared light Detected object is tentatively identified, then the appearance information according to preliminary identification acquisition is in the machine or internet database Inquire associated materials, dielectric constant, density structure.
It is further limited as the present invention, this method also first passes through the surface of spectrometer detection testee before inverting The spectrum of transmitting or reflection is formed with the anti-material for pushing away object of spectral model according to the spectral information obtained, and then obtains dielectric Constant distribution.
It is further limited as the present invention, this method also first passes through radiometer or infrared radiation thermometer or infrared before inverting Thermal imaging system non-contactly obtains the temperature of body surface.
It is further limited as the present invention, system includes multiwave microwave spectrum detection instrument, camera, and calculating is set It is standby, the network equipment and attachment, the workflow of system are as follows: the appearance or surface information of object are obtained by camera, then according to this Partial information scans for or by the network equipment to data base querying, obtains the given data of object, which includes Material, composition, and object is obtained for the bright temperature of different frequency, then by calculating equipment by multiwave microwave spectrum detection instrument Density or Temperature Distribution or material structure information of the inside of inverting object etc..
It is further limited as the present invention, system also has a spectrum spectrum detection instrument, for obtaining object emission Or the spectrum of reflection, the spectroscopic data are used for the ingredient of inverting body surface.
It is further limited as the present invention, system has radiometer also to obtain the temperature of body surface.
Beneficial effects of the present invention:
Compared with machine vision technique scheme before, the technical program is had the advantage that
1. perspective.The letters such as the internal density distribution of object or building, internal structure, internal heat resource distribution can be obtained Breath, no longer stays in surface.
2. lossless.It does not destroy, influence object under test.Contact measured object is not needed, perspective can be completed.
3. low cost.For some remote objects perhaps high-altitude object or extraterrestrial object, if to carry out on the spot Probing, cost may be calculated with 10,000,000,000 dollars, and use this method, and cost that can be extremely low obtains more fully result.
4. safety.Building or blindage internal heat resource such as personnel positions can be judged at a distance, for anti-terrorism, national defence, dimension Shield social security has very big help.
5. conveniently and to human health.Equipment is such as checked oneself based on present invention manufacture human health, it can harmlessly Implement to check, can go to a doctor in time when finding that tumor forms density anomaly or vascular problem, that is, local vascular temperature anomaly.
6. round-the-clock, round-the-clock.It is influenced by weather such as cloud layer, thick fogs, is not also limited by the variation of day and night illumination condition.
Detailed description of the invention
Fig. 1 method and system schematic diagram.Wherein 1 is multiband Microwave sounder, and the number of picture is only illustrated on figure, real Border can be any multiband, and 2 be camera, and the number of picture is only illustrated on figure, and 3 visit for optional infrared radiometer or spectrum Instrument is surveyed, the number of picture is only illustrated on figure, and 4 is calculate equipment, and 5 be the microwave of transmission, and 6 be object under test, and 7 be measuring targets The virtual hierarchy carried out when calculating, 8 be the heat source or density structure inside object under test.
The brightness temperature TB and every layer of actual physics temperature (T that Fig. 2 measurement obtains1-T6) relational graph.Wherein ε is each layer Dielectric constant, k indicates the wave number (wave number) of each layer, and d indicates that the thickness of each layer, e are the natural logrithm truth of a matter, subscript " Expression takes its imaginary part, in addition Rij=(kiz-kjz)/(kiz+kjz), Qij=1+Rij, and so on.
The expansion of Fig. 3 .Y, the concrete meaning of each parameter is in the explanation of Fig. 2.
The layer 5 actual physics temperature results (T5) that Fig. 4 case study on implementation inverting obtains.
The technical program passes through the microwave spectrum detection instrument with multichannel after obtaining necessary essential information first (microwave radiometer) carrys out the microwave band of the different frequency of detecting object transmitting or reflection, (if it is actively micro- If wave spectrum survey meter based on reflection), channel distribution can according to object type to be detected and stratified energy mechanism, than Such as, such as 200MHZ, 1GHZ, 2GHZ be can be set for building, the frequency bands such as 4GHZ, wherein the lower wavelength of frequency is more Long, investigation depth is deeper, and in the case where not too many moisture, the channel of 200MHZ can obtain the wall inside number meter Shen Signal.How many a frequency bands, which are arranged, should depend on commonly using the quantity of the unknown parameter of application scenarios and the level of object under test Number, if to judge what the heat sources such as its heating system were distributed from the internal physical Temperature Distribution of one building of external detection Words, it is assumed that several centimeters of the building surface are ceramic tiles, and several tens cm is concrete later, followed by for steel composites structure, Internal one layer is again concrete again, then building wall to be detected can be divided into 4 layers, (the air layer parameter inside and outside wall Be known), 4 layers of physical temperature be all it is unknown, be set as T1, T2, T3 and T4, it is noted that wherein each T is not a number Value, a but two-dimensional matrix, each element in two-dimensional matrix correspond to the practical temperature of one layer of an interior of building upper coordinate Degree.It can approximatively be interpreted as, each T corresponds to a flat image.In other parameters it is known that as every layer of density and dielectric are normal When number is all known, then 4 or more look-in frequencies should be at least designed, can be four above-mentioned frequency bands, but if in the presence of Unknowm coefficient, the density such as composite construction layer is also unknown, then more frequency ranges should be arranged, and channel quantity should be no less than ginseng to be measured Several quantity (being at this time well-posed situation), certainly, frequency band is more, and the information parameter that can be obtained is more and more Accurately (being at this time overconstrained situation).Therefore, when be applied to machine vision when, should on cost with the appearance of volume Bear range design it is as more as possible, distribution be staggered as far as possible frequency range (be distributed should be staggered, otherwise penetration depth approximation is unfavorable for subsequent Under operation).
An object will be obtained by Microwave sounder described above to be distributed the bright temperature of different frequency wave band, why is claimed Be for distribution because the bright temperature of each channel is not a numerical value and two-dimensional matrix yet, object under test each Coordinate corresponds to a numerical value, and drawing out is an image, sees that object different zones have not under visible light with the past Be with color it is similar, the bright temperature of microwave that the object of different zones emits or reflects also is different, if with mentioned above Wall for if, each wavelength period can show a bright temperature image.This bright temperature is not temperature, but and temperature With the same unit, the corresponding bright Wen Butong of each wavelength.
Certainly, the bright temperature image numerical value of these different-wavebands is different, but it seem that shape etc. is often much the same, because The information from each depth is contained for the bright temperature of each frequency, so the information such as hot spot, cold spot are almost the same. And past satellite microwave detects this step of often also arriving, and is compared according to its shape to study detection result such as SOIL DISTRIBUTION Deng.
However, in fact, bright temperature for different frequency, the ingredient of the information from different depth are different, if The bright temperature image that we can seem similar different-waveband to these shapes carries out information separation, can isolate each physics The density or Temperature Distribution of depth, so that the bright temperature for not having exact physical meaning is translated into true physical parameter.This hair It is bright to use following method to realize the purpose.
As described in the building case of front, we are in inverting by object it is postulated that in order to segmenting one by one Layer, what detector received regards as the layer of these subdivisions for different frequency wave band the bright temperature B of different frequency wave band The contribution of bright temperature is superimposed, and the temperature T that each layer of contribution can be expressed as each layer is multiplied by each layer contribution proportion F, contribution proportion F The function of the density p of this layer and dielectric function ε (Dielectric Permittivity), according to equation group " receive for The layer of bright temperature=subdivision of different frequency wave band for the bright temperature of different frequency wave band superposition=each layer physical temperature multiplied by tribute Offer ratio ", i.e. B=FT equation group, wherein B and F and T is matrix, according to equation group can with the physical temperature of each layer of inverting or Density etc..Assuming that device measuring error matrix is S, the then general solution that the algorithm of inverting provides are as follows:
Wherein FTIndicate the transposed matrix of F, S-1Indicate the inverse matrix of S, (FTS-1F)-1Indicate (FTS-1F inverse matrix).
The inner layers temperature of detected object can be thus acquired according to B, S and T.Wherein S is in addition to relying on Emblem mark is fixed outer, can carry out immediate assessment according to the measurement to two known reference substances.
The composition error S of acquired resultsTFor ST=(FTS-1F)-1
If that internal density is not known yet if producing known variables in F, at this time, it is only necessary to increase and visit Measured channel obtains more equations.As long as there is enough detection channels, so that it may be offset not with enough equations Know parameter.
If that if there is disturbance in the layered structure for encountering object for example, in the above problem, close to indoor There may be the pipeline such as water pipe etc. of various layings in concrete layer, the position of these pipelines is unknown before detection, and is being located The density of its position has been presumed to be the density of concrete in reason, the temperature anomaly that will result in this way, at this point, needing It is iterated calculating (iteration), i.e., primary calculate not can solve, and need to calculate repeatedly and seek approximate solution.With above-mentioned by interior For concrete layer has pipeline, the region for showing abnormal temperature and being distributed is calculated when first time, then this partial region foundation is asked The Temperature Distribution solved be it is known that replacement density function be it is unknown, be iterated calculatings repeatedly, until occur matching stabilization Solution.
If that encounter some objects do not have layered structure again what if at this point, it geometrically first can torn open Being divided into one by one, there is respective object layered to be studied inverting respectively, and very complicated object is split for geometry, can To be split by computer program.In short, above-described layered structure is the model of the basic unit calculated, for complexity Object can be summed up in the point that finally and be handled in basic unit.Certainly, even if having enough detection channels, the Interior Solutions of complex object The calculated performance that analysis perspective needs is still biggish, and the system of inventor's design can be mentioned in following specific implementation case It is small still to have taken 20 under high performance Fortran language on the work station of an i7 processor for processing to an example When time, just finally obtain all thermal source distribution results of the inside of the object.
Theoretically, if there is sufficiently strong computing capability, by frequency band that is as more as possible and being in staggered distribution, it is enough repeatedly Generation calculating, careful fractionation, are able to solve the perspective demand of most objects.
The type of microwave spectrum detection instrument should be as needed resolution ratio determine use active type or passive type.It is main Dynamic Microwave sounder, i.e. detecting devices actively emit microwave signal, then receive the signal of detected target reflection or scattering again Detected target is perceived, similar radar.Passive microwave survey meter working principle is then set similar to spectrographic detection (radiometer) standby, i.e., using highly sensitive microwave receiving technology, receive the faint microwave signal of target itself.Quilt The spatial resolution of dynamic detection is lower, is not suitable for small-sized object detection.If object under test is small, interference is big, preferably use actively micro- Wave detection.
In practical operation, also preferably further apply constraint using following method, reduce error: using a priori assumption (a- Priori), the assumption value (an often matrix) to inverted parameters is provided by various Given informations, then by the elder generation It tests to assume to wait with the bright temperature being actually detected and carries out inverting together, provide most probable solution (the most probable to inverted parameters solution).Because if do not included a-priori, directly according to measured value inverting, even if often using the wave band of many frequencies The under operation is carried out, error can be still very big.
Or by taking building recited above perspective as an example, it is assumed that we guess according to other Given informations such as engineering data The distribution function T of one temperatureAError (error coviarance) * SA, then new solution calculation formula are as follows:
It after applying a-priori, is equivalent to and combines measurement and Given information, more optimized solution can be provided.When So, S and S are accurately providedAIt is particularly significant, if S i.e. measurement error are very big, and SAThe namely error of Given information solution It is considered very small words, that result will be very close to the T guessed according to Given informationA；It is on the contrary then close to T.
In order to reduce calculation amount, or reduction unknown parameter to the greatest extent, we can also establish relevant database to familiar object, The information such as dielectric constant or density or specific heat comprising familiar object, directly to be inquired in inverting.Many familiar things Density, dielectric constant etc. often can all be found.And the substantially classification of judgment object is wanted, or perhaps if what, The pre-treatment step an of machine vision can be notified to complete, for example, first passed through before inverting other detection method such as visible lights, Infrared light tentatively identifies detected object, determines that detected things is, then obtains according to preliminary identification outer The information such as sight inquire associated materials, dielectric constant, density structure etc. in the machine or internet database.It will inquire as far as possible The parameter information arrived is obtained by inquiry, to mitigate Inversion Calculation amount.
In some cases, object is that unknown material is constituted, and spectrum detection instrument spy at this moment can be first passed through before inverting The surface emitting of testee or the spectrum of reflection are surveyed, according to the spectral information obtained with spectral model (often line spectrum Model) it is counter push away object material composition, and then obtain dielectric constant distribution.Such as above-mentioned building case, the one of most surface Layer is ceramic tile, and its different dielectric constant of the material of ceramic tile will be different, can first distinguish its material using spectrum detection instrument Distribution, to obtain the dielectric constant and density of ceramic tile layer.
In some cases, the surface temperature of object is unknown, can be measured at this time by radiometer such as infrared radiometer It obtains.
Many appliance arrangements can be designed according to above-mentioned method, to give smart machine or robot in seeing through objects The 3D visual capacity in portion.
Typically, as shown in Fig. 1, a system for being applied to robot perspective vision has a multiband array Microwave spectrum detection instrument 1, wave band should be staggered as far as possible, more than one camera 2, a calculating equipment 4, the attachment network equipment, The workflow of system are as follows: the appearance or surface information of object are obtained by camera, then scan for according to this partial information or By the network equipment to data base querying, to obtain the given data of object, and by multiwave microwave spectrum detection instrument with The mode for receiving the microwave 5 of object under test transmitting or reflection obtains object for the bright temperature of different-waveband, then by calculating equipment The density or Temperature Distribution 8 of the inside of corresponding layered structure 7 and inverting object are divided into according to above-mentioned method.
For the object of self-assembling formation, since its surface and internal material composition often have continuity, system at this time A spectrum spectrum detection instrument 3 can be also had, for obtaining the spectrum of object emission or reflection, which is used for inverting object The ingredient in body surface face, and then assume according to certain model the material composition of interior of articles.Material composition determines after determining The distribution of dielectric constant.If the temperature of body surface is unknown, it may also include a radiometer (Radiometer) or infrared Temperature measurer obtains the temperature of body surface.
Using many internal informations that the available past technology of the equipment is unable to get, for example, with above-mentioned building For object case, it is only necessary to which the robot by equipping the equipment carries out remote non-contacting observation, so that it may obtain building Internal heat source distribution (and then thermoelectric structure or mankind's activity inside deducibility), pipeline distribution and situation, for engineering, army Thing, medical observation, daily life are all significant.
In specific implementation case of the invention, inventor manufactured one realization the above method system, system band just like Lower part: Microwave sounder, spectrum detection instrument, radiometer, camera, the computer program of multiband.
The multi-frequency microwave survey meter has 4 look-in frequencies, respectively 120MHZ, 450MHZ, 1GHZ, with 5GHZ.There is master Dynamic model formula and Passive Mode use Passive Mode in case study on implementation.
Spectrum detection instrument is an imaging spectral survey meter.
Radiation is calculated as the infrared radiometer of measurement surface temperature.
Program is write using Fortran.It at present only include special to some observation objects due to being in the sample stage The function and library write can't directly observe arbitrary object.
To be viewed as an application case of the system to a dry mineral block that there is inside to bury radioactive heat source, The positioning of non-contacting inner radiation heat source is carried out using this system in this example.
The mineral block is observed by camera first in an implementation, and inquires multi view image, confirmed after comparing Mineral block type generally forms condition in turn to the block founding mathematical models according to the type, and model is divided into six layers, first layer For the weathering scrim layer on surface, the second layer to layer 5 is the argillic horizon formed various years, and layer 6 is lithosphere.Each layer All there is respective density p i Two dimensional Distribution and temperature Ti Two dimensional Distribution.It mainly needs due to finding radioactive heat source by heat point Cloth judgement, therefore approximating assumption density is every layer of numerical value, to reduce parameter.
Then mineral upper and lower surface mineralogical composition is observed by spectrum detection instrument, obtains the iron distribution of the sphere, mineral For mainly density and iron content being affected for dielectric constant, then by spectrum detection instrument observe side profile acquisition Each layer component material and every layer of thickness (depth) of mineral, has substantially determined the density of each layer.It is assumed that iron connects in object The distribution two-dimensional matrix of each layer dielectric constant of object has been inferred in continuous distribution according to Surface Fe distribution together in conjunction with each layer density.
Secondly infrared radiometer obtains the Two dimensional Distribution T1, T6 of the temperature of upper and lower surface.To be solved is that the second layer arrives The Temperature Distribution T2, T3, T4, T5 of layer 5.(note: temperature T herein is all referring to each mesh coordinate position on the internal layer The two-dimensional matrix of the distribution for the actual temp set, drawing out is a two dimensional image, definitely not single number)
Four different-wavebands according to Microwave sounder obtain four brightness temperatures, TB1, TB2, TB3 and TB4 respectively.
Program part is based on following physical principle and writes in Fortran: being first depending on fluctuation-dissipation theorem (fluctuation dissipation theorem, FDT) and dyadic Green's function obtain bright temperature and each layer in hierarchical system The relationship of actual temperature is shown in attached drawing 2.Specific parameter interpretation is all described in relevant Detailed description of the invention.
It is T=[T2, T3, T4, T5] to inverted parameters T, but measured value is not at this time in this example according to inversion theory Bright temperature B itself, if note B be B=[TB1, TB2, TB3, TB4], then actual measured value at this time be B subtract surface layer contribution and The contribution of bottom, that is, measured value Y=(contribution of the surface layer B-TB contribution-TB bottom)=F*T.This is because wherein by infrared spoke The actual physics distribution that meter obtains surface layer and bottom is penetrated, the contribution on surface layer and the contribution of bottom are all that can directly calculate acquisition , therefore Y is only measured value, i.e. measurement vector at this time.
Wherein the expansion of Y is as shown in attached drawing 3.Specific parameter interpretation is all described in relevant Detailed description of the invention.
At this time according to T=(FTS-1F)-1FTS-1We obtain the inversion results of T by Y.T2-T3 is without abnormal, radioactivity Heat source is mainly in layer 5, and according to the 4th parameter T5 of T, we can find heat source position, and T5 is defeated according to transverse and longitudinal coordinate Out on image, as shown in Fig. 4.The unit of coloration item is degree Celsius.Wherein black portions correspond to higher temperature value, It is exactly the position place of internal heat resource.According to the image of T5, it is known that the position of inner radiation heat source is in upper right corner stain Place.
The another application case of this system is exactly the building illustrated in Summary.Its hierarchical mode is in invention Hold part also to have illustrated.After the processing of the implementation case system, (the process object class of specific calculating process and front Seemingly repeat no more), we obtain the heat source distributions in building.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, made any to repair Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
1. a kind of machine perspective vision implementation method, it is characterised in that: by including but is not limited to visible or infrared light to quilt Detecting object carries out preliminary detection identification, obtain include but is not limited to object outer dimension and external skin temperatures object it is basic Information, and the microwave of the different frequency by the measurement object under test transmitting of the microwave spectrum detection instrument of multi-frequency or reflection, obtain One object for different frequency wave band bright temperature, in conjunction with including but not limited to material, the Given information or model of dielectric constant, The information for carrying out physical temperature distribution, Density Distribution or structure that inverting includes but is not limited to each layer, carries out in the following manner:
This method uses following algorithm in inverting: object being first assumed to be to the layer segmented one by one, then the different frequencies received The bright temperature of the object of rate wave band is the superposition of the contribution of the layer temperature bright for different frequency of these subdivisions, and each layer of contribution is expressed as Each layer temperature multiplied by contribution proportion function, contribution proportion function be the density of this layer, dielectric constant, thickness function, according to side Journey group " the bright temperature T for different frequency wave band receivedB=subdivision layer for different frequency wave band bright temperature superposition= Each layer physical temperature TiMultiplied by each layer contribution proportion " physical temperature or density information of each layer of inverting.
2. the method according to claim 1, wherein do not have object layered for some, by its Geometrically first it is split as one by one there is basic object layered to give inverting respectively again.
3. passing through Given information the method according to claim 1, wherein this method also uses a priori assumption Provide an assumption value to inverted parameters, a priori assumption be then subjected to inverting together with actual measured results, provide to The most probable value of inverted parameters.
4. the method according to claim 1, wherein this method also establishes relevant database to familiar object, Dielectric constant or density comprising familiar object or than thermal information, directly to be inquired in inverting.
5. including but not limited to may be used the method according to claim 1, wherein this method is also first passed through before inverting Light-exposed, infrared light detection method tentatively identifies detected object, then includes but not according to what preliminary identification obtained It is limited to appearance information, it is the query-related information in the machine or internet database, including but not limited to material, dielectric constant, close Spend structure.
6. being tested the method according to claim 1, wherein this method also first passes through spectrometer detection before inverting The surface emitting of object or the spectrum of reflection are formed according to the spectral information obtained with the anti-material for pushing away object of spectral model, into And obtain dielectric constant distribution.
7. the method according to claim 1, wherein this method also first passes through radiometer or infrared survey before inverting Wen Yi or thermal infrared imager non-contactly obtain the temperature of body surface.
8. a kind of system for realizing claim 1 the method, which is characterized in that system includes at least multiwave microwave spectrum Survey meter, camera calculate equipment, the network equipment, the workflow of system are as follows: appearance or the surface of object are obtained by camera Information, then scan for or by the network equipment according to this partial information to data base querying, the given data of object is obtained, it should Given data includes but is not limited to material, composition, and obtains object for different frequency by multiwave microwave spectrum detection instrument Bright temperature, include but is not limited to then density or Temperature Distribution or material structure by the information of the inside of calculating equipment inverting object Information.
9. system according to claim 8, which is characterized in that system also has a spectrum spectrum detection instrument, for obtaining The spectrum of object emission or reflection is obtained, which is used for the ingredient of inverting body surface.
10. system according to claim 8, which is characterized in that system has radiometer also to obtain the temperature of body surface Degree.
Priority Applications (1)
|Application Number||Priority Date||Filing Date||Title|
|CN201410570417.1A CN104297814B (en)||2014-10-24||2014-10-24||A kind of machine perspective vision implementation method|
Applications Claiming Priority (1)
|Application Number||Priority Date||Filing Date||Title|
|CN201410570417.1A CN104297814B (en)||2014-10-24||2014-10-24||A kind of machine perspective vision implementation method|
|Publication Number||Publication Date|
|CN104297814A CN104297814A (en)||2015-01-21|
|CN104297814B true CN104297814B (en)||2019-11-12|
Family Applications (1)
|Application Number||Title||Priority Date||Filing Date|
|CN201410570417.1A Active CN104297814B (en)||2014-10-24||2014-10-24||A kind of machine perspective vision implementation method|
Country Status (1)
|CN (1)||CN104297814B (en)|
Families Citing this family (1)
|Publication number||Priority date||Publication date||Assignee||Title|
|CN106124556B (en) *||2016-06-08||2019-04-16||哈尔滨工业大学||Thermal parameter inverting measuring device based on thermal infrared imager|
Family Cites Families (8)
|Publication number||Priority date||Publication date||Assignee||Title|
|US5777481A (en) *||1996-04-01||1998-07-07||University Corporation For Atmospheric Research||Ice detection using radiometers|
|JP2004012452A (en) *||2002-06-11||2004-01-15||Marcom:Kk||Microwave moisture meter|
|CN101295022A (en) *||2008-06-25||2008-10-29||中国农业科学院农业资源与农业区划研究所||Method for ground surface temperature and emissivity inversion by remote sensing data ASTER|
|CN101963664B (en) *||2010-09-28||2013-03-20||中国科学院东北地理与农业生态研究所||Microwave remote sensing pixel element decomposing method based on land and water living beings classifying information|
|CN102135573B (en) *||2010-11-23||2013-05-08||吉林大学||Device and method for measuring dielectric constant of low-loss medium|
|CN102155996A (en) *||2011-03-29||2011-08-17||吉林大学||Device and method for measuring microwave thermal radiation characteristic of dense medium|
|CN102393537B (en) *||2011-10-30||2014-04-23||北京无线电计量测试研究所||Scan method of human body security system utilizing frequency division technology|
|CN103512663B (en) *||2013-09-13||2016-03-02||华中科技大学||A kind of computing method of the bright temperature of lunar surface microwave that rises and falls|
- 2014-10-24 CN CN201410570417.1A patent/CN104297814B/en active Active
Also Published As
|Publication number||Publication date|
|CN109073348A (en)||For to bury object detected, position and image obtain airborne system and method, characterize subsoil ingredient method|
|Kurrant et al.||Regional estimation of the dielectric properties of inhomogeneous objects using near-field reflection data|
|Di Donato et al.||A new linear distorted-wave inversion method for microwave imaging via virtual experiments|
|CN105866771B (en)||A kind of cache detecting system and its detection method|
|Bevacqua et al.||Boundary indicator for aspect limited sensing of hidden dielectric objects|
|CN104297814B (en)||A kind of machine perspective vision implementation method|
|Ludeno et al.||A comparison of linear inverse scattering models for contactless GPR imaging|
|Danese et al.||Geophysical methods and spatial information for the analysis of decaying frescoes|
|Aboudourib et al.||A processing framework for tree-root reconstruction using ground-penetrating radar under heterogeneous soil conditions|
|Blanche et al.||Analysis of geomaterials using frequency modulated continuous wave radar in the X-band|
|Nguyen et al.||A coherent model of forest scattering and SAR imaging in the VHF and UHF-band|
|Johnson et al.||Through-wall sensing with multifrequency microwave radiometry: a proof-of-concept demonstration|
|Sood et al.||Lunar advanced radar orbiter for subsurface sounding (LAROSS): Lava tube exploration mission|
|Peichl et al.||Passive microwave remote sensing for security applications|
|Angell et al.||Computational modeling analysis of radar scattering by metallic body-worn explosive devices covered with wrinkled clothing|
|Govind et al.||Microwave nondestructive imaging of buried objects using improved scattered-field calibration technique|
|Bhattacharya et al.||Remote estimation of dielectric permittivity of lunar surface regolith using compact polarimetric synthetic aperture radar data|
|Sato et al.||Estimation of subsurface fracture roughness by polarimetric borehole radar|
|Zhang et al.||Resolution and robustness to noise of the sensitivity-based method for microwave imaging with data acquired on cylindrical surfaces|
|De Leo et al.||An EM modeling for rescue system design of buried people|
|CN109061769A (en)||Infrared system operating distance based on controllable target tests appraisal procedure|
|Wang et al.||High-resolution through-wall ghost imaging algorithm using chaotic modulated signal|
|Bellez et al.||Full polarimetric bistatic radar imaging experiments on sets of dielectric cylinders above a conductive circular plate|
|Rezgui et al.||Development of an ultra wide band microwave radar based footwear scanning system|
|Panov et al.||Method of active radio-wave imagery of the areas extensively contaminated by radionuclides from the aircraft and spacecraft|
|SE01||Entry into force of request for substantive examination|
|SE01||Entry into force of request for substantive examination|