CN101464148A - Three-dimensional image detecting, compiling and reconstructing system - Google Patents
Three-dimensional image detecting, compiling and reconstructing system Download PDFInfo
- Publication number
- CN101464148A CN101464148A CNA2007101621697A CN200710162169A CN101464148A CN 101464148 A CN101464148 A CN 101464148A CN A2007101621697 A CNA2007101621697 A CN A2007101621697A CN 200710162169 A CN200710162169 A CN 200710162169A CN 101464148 A CN101464148 A CN 101464148A
- Authority
- CN
- China
- Prior art keywords
- dimensional image
- target
- parameter
- function
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
A system for sensing, compiling and rebuilding three-dimensional images comprises an inputting and capturing unit, a training and interpreting unit, a displaying and compiling unit, and a rebuilding unit, wherein, the inputting and capturing unit receives a digital image and an airborne lidar datum and captures a first parameter to produce a three-dimensional image; the training and interpreting unit selects a target from the three-dimensional image, captures a second parameter of the target and calculates the second parameter to obtain a generated threshold value of the target, and detects and selects an area with similar properties with the target in the three-dimensional image according to the threshold value; the displaying and compiling unit establishes an enclosing tool for expanding and searching according to the threshold value; the enclosing tool can be used for adding, deleting and editing a detection result of the training and interpreting unit; and the rebuilding unit establishes a buffer area at the periphery of the target and captures a third parameter of the buffer area, and carries out curved surface pressurizing calculation to rebuild a previous condition of the target according to the third parameter.
Description
Technical field
The invention relates to a kind of image detecting, compile and reconstructing system, refer to especially a kind ofly be used for three-dimensional image detecting, compile and reconstructing system.
Background technology
TaiWan, China has the area more than 60 of percentage to belong to the mountain area and hillside landform, especially alpine terrain are precipitous, the streams is rapid, tectonic structure is complicated, lithology is fragile, soil is soft, and average yearly rainfall reaches 2,500 millimeters, is average 3 times in the whole world.Especially after the violent earthquake of 1999 921 collection collection, cause thin solum loosening,, very easily trigger native stone avalanche therefore whenever typhoon or torrential rain.Native stone avalanche is that soil is washed into the area, downstream along gully and river to direct influence the in gathering ground, and is piled up in the reservoir bottom, causes pondage to descend, and the while also causes the water quality dirt and influences water supply, causes people's livelihood predicament and economic loss.And avalanche ground is if fail to restore as early as possible, and rainwater will wash away veneer of soil stone gradually, further reduce the reservoir filling function, and reduce the reservoir life-span.Therefore in order to assist reservoir, forest and hillside effectively to manage, educate work again with comprising avalanche, and anti-disaster relief decision support is provided, therefore be necessary to carry out effectively and accurate the investigation work.
The mode of investigating traditionally be mainly dispatch officers terrain investigation with measure, or utilize telemetry to detect or know face of land zone of migration manually to declare in the robotization mode.The mode of terrain investigation wherein is subjected to the influence of factors such as landform, weather and traffic easily and can't effectively carries out, though the most not reliably, efficient is low and appropriate litigation fees is high.And utilize aeroplane photography and landsat image, with the apparent over the ground survey of telemetry, its shooting to contain area big, have wide area pull will, low observation dead angle and the characteristics such as ability of property repeated measures for a long time, the restriction that allows the user can surmount sense organ and space-time detects the information on the face of land apace.The most widely used at present boat telemetry source comprises aerophotograph (three-dimensional to or orthography), optical satellite image (orthography), radar image (Synthetic Aperture Radar, SAR), ground type and unloaded formula light reach (Light Detection And Ranging, LiDAR) etc.Below inquire into the correlation technique of utilizing telemetry to investigate with regard to data surface and technological side.
Aerophotograph
The technology that adopts is carried out interpretation for utilizing the aerophotograph solid right in the mode of manually carrying out stereo measurement.Though the method can obtain small size and high-precision face of land zone of migration, need expend considerable manpower and time, can't effectively provide related data to assist disaster investigation and the condition of a disaster assessment rapidly.Whole according to pertinent literature and experience remittance, the identification criterion of artificial interpretation comprises six of tone, position, shape, direction, the gradient and shades etc., and is as shown in table 1.Inquire into the applicability of these identification criterions from the angle of robotization, wherein " tone " partly can extract the possible scope that obtains in the robotization mode by the color information of aerophotograph." position " partly then must utilize other auxiliary data, and for example road, ridge and rivers and creeks vector plot so that make buffer zone, and extract possible face of land transition place in superimposed mode." shape " partly need utilize the kenel of landform, but this partly is to be difficult to the work that the robotization mode is detected most.(Digital Elevation Model DEM) calculates aspect, cooperates rivers and creeks information to detect the place that face of land transition may take place simultaneously then can to utilize the numerical value elevation model for the judgement of " direction ".What " gradient " was then same can calculate by the numerical value elevation model." shade " is visual perception's factor; fundamental purpose is for judging topographic relief; be difficult for reaching this purpose with robotization; therefore in artificial interpretation process; usually can adopt stereopsis; or with orthography cooperation numerical value elevation model (Digital Elevation Model, DEM) emulation three-dimensional sight.Therefore above-mentioned various interpretation criterions, and be not suitable for being used as fully the necessary condition of robotization interpretation.
Table 1
| The interpretation factor | Content |
| Tone | Drabon look, dark brown, filbert, green and brown look. |
| The position | Impact slope, road near ridge, river valley. |
| Shape | Long strip type, soupspoon type, dendroid river junction, the river valley is other may triangularity or rectangle. |
| Direction | Side slope gravity direction and the river direction relation of being orthogonal. |
| The gradient | The inclination hillside fields. |
| Shade | Hatching effect can obtain the sensation of 3D in order to find out river valley and ridge. |
Satellite image
Utilize satellite orthography interpretation avalanche ground, roughly similar to the technology of utilizing aerophotograph, but because the restriction of satellite image on spatial resolution, more rely on the transition analysis of using different times, compare two kinds of technology, utilize avalanche ground quantity that the satellite image Auto-Sensing gets and area all more than utilizing aerophotograph few, still have the place of deficiency with regard to the application on the engineering with the result of artificial interpretation.But, therefore after calamity, can carry out face of land transition detecting investigation fast, with the condition of a disaster investigation work after the assistance calamity because satellite has higher temporal resolution and sizable coverage.In addition, said method is declared when knowing utilizing single width of cloth satellite image to carry out the face of land, because the spatial information of the shortcoming third dimension, therefore to cooperate the numerical value elevation model to declare and know and compile with three-dimensional stereoscopic visual emulation mode indirect labor, be avalanche ground to avoid the judging smooth open ground of landform by accident.
Unloaded formula light reaches
Have the unloaded light of research and utilization to reach the avalanche ground that scanning 921 earthquakes cause, the vertical accuracy that scans achievement through checking on the spot can reach 12 centimeters, and so this report is not inquired into the detecting of avalanche ground and measured relevant problem.Utilize unloaded light to reach and produce high resolving power numerical value elevation model (DEM), utilize three dimensional analysis to obtain roughness of ground surface (Roughness), and then analyzed avalanche ground surface kenel (Morphology), by the analysis of this kind topography, can further understand avalanche ground feature, avalanche ground behavioral mechanism, and the ground of the avalanche in the assessment activity.Utilize this high resolving power numerical value elevation model, analyze roughness of ground surface, the gradient, aspect, semivariation number (Semi-Variance) and broken type yardstick (Fractal Dimension) etc., be used for analyzing avalanche the earth's surface kenel, composition and activity.
The unloaded light that above-mentioned research and utilization point cloud density is every square centimeter of point reaches and produces 1.8 meters grid type numerical value elevation models, and shade fluctuating stereographic map (Shaded Relief Map), slope map, topographic contour figure, topographic profile etc., by the ground surface form, analyze the change mechanism of face of land transition on space and time scale, and then analyze the ground surface kenel and interpretation obtains avalanche ground scope, estimate to have approximately 1/3rd light to reach a cloud in the data and can detect landform face under the tree, therefore manually compile the DEM that is produced compared to utilizing aerophotograph to reach with the image coupling, light reaches the interpretation that DEM helps historical avalanche ground.But for the exposed face of land transition of soil, utilize aerophotograph then can survey and draw more accurate transition border, the face of land.In addition, still do not have in the pertinent literature and utilize unloaded light to reach the research of detecting face of land zone of migration in the robotization mode.
Telemetry is in the applicability of face of land transition investigation
Take a broad view of and above-mentionedly utilize various telemetries to carry out face of land transition investigation methods, its applicability can be converged and be made into table 2.Wherein " degree of accuracy " part is meant the bearing accuracy of zone of migration boundary plane coordinate, its size depends primarily on the spatial resolution of data, therefore pinpoint accuracy is represented to reach a centimeter grade, and middle degree of accuracy represents to reach 1 meter grade, and low accuracy then is about 10 meters grades.Light reaches image can reach centimetre even millimeter grade in the detecting precision of elevation change, yet relatively low in the bearing accuracy of surface level, therefore estimates at the setting accuracy of ground table boundary low.
So-called " field of investigation " mainly is to obtain data with ground or high-altitude to distinguish in the table 2.And about the not strict at present standard of the definition of " face of land transition of small size ", be that major axis with transition place is a benchmark less than 50 meters at present, existing technology then is that the area with three pixels of the multispectral orthography of employed SPOT satellite defines just about 468.75m
2(=12.5 * 12.5 * 3).
" real-time " in the table 2 is meant after disaster takes place, and the time number of comprehensive disaster scope or face of land transition present situation is provided, and institute takes time, and real-time is low more more at most.Yet the maximum variable that influences this real-time is weather conditions, therefore do not consider earlier weather conditions, and satellite image scheduling and data transfer time, with regard to comprehensive face of land transition investigation, the ground type telemetry obtain the demand that is difficult for reaching real-time, and satellite image has higher temporal resolution than aeroplane photography, so the real-time of satellite image is higher.
Table 2
| Degree of accuracy | The field of investigation | Detecting small size avalanche ground | Real-time | Automaticity | |
| The aerophotograph solid is right | High | On a large scale | Can | In | Low |
| The aerophotograph orthography | High | On a large scale | Can | In | High |
| SPOT satellite orthography | Low | On a large scale | Not | High | High |
| The SAR satellite image | Low | On a large scale | Not | Low | High |
| The high satellite orthography of resolving | In | On a large scale | Can | High | High |
| Unloaded light reaches (according to relief form with artificial interpretation) | High | On a large scale | Can | In | Low |
In " automaticity " partly, mainly be whether can reach to a certain degree robotization according to data characteristic, its degree is relevant with data processing technique and the artificial number of getting involved.Wherein the aerophotograph solid is right, and the artificial degree that gets involved is higher, and three-dimensional to image and volume coordinate direct corresponding relation a little less than, so automaticity is lower.Other orthography and unloaded light reach partly, then because data with the geographic coordinate fit, so can by suitable algorithm fast detecting obtain face of land zone of migration, reach robotization to a certain degree.Yet we know the degree height of robotization, do not represent achievement degree of accuracy and fiduciary level height, and therefore follow-up artificial interpretation work can't be avoided, and the right artificial degree that gets involved is to decide on the ability of algorithm, can't compare one by one at this.
Therefore, for the degree that reduces artificial intervention to improve the accuracy of automaticity, real-time and interpretation, the present invention is in view of the disappearance of known technology, be through testing and study and a creation spirit of working with perseverance concentratedly, create the present invention's " three-dimensional image detecting, compile and reconstructing system " eventually.
Summary of the invention
The present invention desires to provide a kind of three-dimensional image detecting, compiles and reconstructing system, use the characteristic parameter of target image among the 3-dimensional image figure do the detecting of extensive robotization, to the detecting result compile and to the image reconstruction of detecting target.According to fundamental purpose of the present invention, a kind of three-dimensional image detecting method is provided, its step comprises: receive a digitized video and a unloaded light reach data (Light Detection And Ranging, LiDAR) and capture one first parameter; Make a 3-dimensional image according to this first parameter; In this 3-dimensional image, choose a target and capture one second parameter of this target; Calculate this first and this second parameter to obtain producing a threshold value of this target; And in this 3-dimensional image, detect and choose zone with this target similar performance according to this threshold value.
Three-dimensional image detecting method according to above-mentioned conception, wherein this first parameter comprises a visible light wave range, a numerical value terrain model (DigitalSurface Model, DSM), a numerical value elevation model (DigitalElevation Model, DEM) and an object height model (ObjectHeight Model, OHM).
According to the three-dimensional image detecting method of above-mentioned conception, wherein this second parameter comprises a roughness of ground surface, a gradient, a ground object height and a visible light wave range.
According to the three-dimensional image detecting method of above-mentioned conception, wherein this digitized video be selected from a digital airborne photography image or a digital satellite photographic image one of them.
According to the three-dimensional image detecting method of above-mentioned conception, wherein this target is a landslide image.
According to another fundamental purpose of the present invention, provide a kind of 3-dimensional image to compile method, its step comprises: receive a digitized video and a unloaded light and reach data to make a 3-dimensional image; In this 3-dimensional image, choose a target and capture a parameter of this target; Set up a circle selection tool of searching in order to expansion according to this parameter; And use this circle selection tool to increase, delete and edit this target.
3-dimensional image according to above-mentioned conception is compiled method, and wherein this parameter comprises a roughness of ground surface, a gradient, a ground object height and a visible light wave range.
3-dimensional image according to above-mentioned conception is compiled method, and wherein this circle selection tool is fairy maiden's rod.
3-dimensional image according to above-mentioned conception is compiled method, and wherein this target is a landslide image.
According to another fundamental purpose of the present invention, a kind of 3-dimensional image automatic reconstruction method is provided, its step comprises: receive a digitized video and a unloaded light and reach data to make a 3-dimensional image; In this 3-dimensional image, choose a target; Set up a buffer zone around this target and capture a parameter of this buffer zone; And carry out a curved surface driving fit (SurfaceFitting) according to this parameter and calculate to rebuild an original state of this target.
According to the 3-dimensional image automatic reconstruction method of above-mentioned conception, wherein this parameter comprise a numerical value elevation model (Digital Elevation Model, DEM) and border, ground vector data.
According to the 3-dimensional image automatic reconstruction method of above-mentioned conception, wherein this curved surface driving fit is calculated and is utilized a toroidal function to carry out.
According to the 3-dimensional image automatic reconstruction method of above-mentioned conception, wherein this toroidal function be selected from a plane function (Planar surface), a bilinear surface function (Bi-linear surface), a quadric surface function (Quadratic Surface), a pair of quadric surface function (Bi-quadratic Surface), one cube of toroidal function (Cubicsurface) and a pair of cube toroidal function (Bi-cubic surface) one of them.
According to the 3-dimensional image automatic reconstruction method of above-mentioned conception, wherein this digitized video be selected from a digital airborne photography image or a digital satellite photographic image one of them.
According to the 3-dimensional image automatic reconstruction method of above-mentioned conception, wherein this target is a landslide image.
According to another fundamental purpose of the present invention, a kind of three-dimensional image detecting is provided, compiles and reconstructing system, comprise: an input and an acquisition unit in order to receiving a digitized video and a unloaded light reaches data, and captures one first parameter to make a 3-dimensional image; One training and interpretation unit, be connected with this input and acquisition unit, in order in this 3-dimensional image, to choose a target and to capture one second parameter of this target, and calculate this first and this second parameter obtaining producing a threshold value of this target, and in this 3-dimensional image, detect and choose zone with this target similar performance according to this threshold value; One shows and compiles the unit, links to each other with this training and interpretation unit, one encloses selection tool in order to set up according to this threshold value in order to what expansion was searched, and this circle selection tool can be used for increasing, delete and edit and the zone of this target similar performance; An and reconstruction unit, be connected with this training and interpretation unit, in order to having chosen one the 3rd parameter of setting up a buffer zone around the target and capturing this buffer zone at this, and carry out a curved surface driving fit (Surface Fitting) according to the 3rd parameter and calculate to rebuild an original state of this target.
According to the three-dimensional image detecting of above-mentioned conception, compile and reconstructing system, wherein this first parameter comprises a visible light wave range, a numerical value terrain model (Digital Surface Model, DSM), a numerical value elevation model (Digital Elevation Model, DEM) and an object height model (Object Height Model, OHM).
According to the three-dimensional image detecting of above-mentioned conception, compile and reconstructing system, wherein this second and the 3rd parameter comprises a roughness of ground surface, a gradient, a ground object height and a visible light wave range respectively.
According to the three-dimensional image detecting of above-mentioned conception, compile and reconstructing system, wherein this digitized video be selected from a digital airborne photography image or a digital satellite photographic image one of them.
According to the three-dimensional image detecting of above-mentioned conception, compile and reconstructing system, wherein this target is a landslide image.
According to the three-dimensional image detecting of above-mentioned conception, compile and reconstructing system, wherein this curved surface driving fit is calculated and is utilized a toroidal function to carry out.
According to the three-dimensional image detecting of above-mentioned conception, compile and reconstructing system, wherein this toroidal function be selected from a plane function (Planar surface), a bilinear surface function (Bi-linear surface), a quadric surface function (Quadratic Surface), a pair of quadric surface function (Bi-quadratic Surface), one cube of toroidal function (Cubicsurface) and a pair of cube toroidal function (Bi-cubic surface) one of them.
Description of drawings
For making the auditor further cognitive and understanding be arranged to feature of the present invention, purpose and function, hereinafter special theory reason with dependency structure accompanying drawing of the present invention and design describes, so that the auditor can understand characteristics of the present invention, detailed description is presented below, wherein:
Fig. 1 is three-dimensional image detecting of the present invention, compile and the process flow diagram of the first step of reconstructing system data input and three-dimensional image making.
Fig. 2 is three-dimensional image detecting of the present invention, compile and process flow diagram, the training of target interpretation and the automated detection target of second step of reconstructing system.
Fig. 3 is three-dimensional image detecting of the present invention, compile and the process flow diagram of the third step of reconstructing system, compiles the detecting result.
Fig. 4 is three-dimensional image detecting of the present invention, compile and the process flow diagram of the 4th step of reconstructing system, and former landform is rebuild.
The instrument that the expansion that Fig. 5 compiles for 3-dimensional image of the present invention is chosen, fairy maiden's rod, the synoptic diagram of choosing method.
Fig. 6 is the example of the employed curved surface polynomial function of 3D image reconstruction system of the present invention.
Embodiment
The present invention utilizes the unloaded light of having handled through classification to reach data, cooperates colored aviation orthophoto, in the mode of three-dimensional sight simulation, sets up the automated detection correlation technique, to promote the efficient and the accuracy of large-scale inquiry work.Yet, when carrying out automated detection, have unavoidably and leak the phenomenon of awarding (Omission) or erroneous judgement (Commission), thus be redevelopment one man-machine operation interface, the result who supplies artificial interpretation and compile automated detection.
The present invention can fully be understood by following embodiment explanation, make the personage who has the knack of present technique to finish it according to this, and right enforcement of the present invention is not can be limited it by following example to implement kenel.
See also Fig. 1, it is three-dimensional image detecting of the present invention, compile and the process flow diagram of the first step of reconstructing system data input and three-dimensional image making.At first, receive digital aerophoto earlier and carry out observation data (step 11) from the air to surface surface of height, the observing capacity that it has on a large scale with low dead angle after (routine) or change (disaster), can carry out comprehensive investigation work at ordinary times at large-area region of interest.Next capture information (step 12), use Spectrum Analysis that boat is done the increased surface covering analysis according to orthography, utilize its different visible light wave range clearly to define the border of different atural objects, carry out the three-dimensional spatial information extraction simultaneously, use is simultaneously climbed and is slided algorithm (being called for short CAS), the high density and the high accuracy three-dimensional point cloud information that provide in the data are provided obtained light carry out mechanized classification, light reached be divided into ground and landform two classes, be inserted into numerical value terrain model (the Digital SurfaceModel of one meter grid simultaneously, DSM) with numerical value elevation model (Digital ElevationModel, DEM) and calculate DSM and DEM difference, obtain object height degree model (Object Height Model on the ground, OHM), in order to the analysis of ground surface roughness to be provided, the landform face gradient is calculated the spatial information with the 3rd of atural object, just elevation change.
Form DSM, DEM and the OHM rule triangulation network, and make multi-level fineness (LOD) geometric data structure, obtain Real-time and Dynamic 3-dimensional image (step 13) with making.And can be far and near according to object and observer's distance, select the triangulation network of suitable fineness to draw, reduce drawing triangle netting index, reach the purpose that Real-time and Dynamic is showed.
See also Fig. 2, it is three-dimensional image detecting of the present invention, compile and process flow diagram, the training of target interpretation and the automated detection target of second step of reconstructing system.In first step, select a target as training area (step 21) in the 3-dimensional image of made, system's automatic pick-up and the parameter of adding up the various attributes of this training area, comprise roughness, the gradient, object height and visible light wave range (step 22).Add up the mean value and the standard deviation of these property parameters, again with mean value+/-standard deviation of a certain multiplying power is used as threshold value (step 23).Then position and scope (step 24) that large-scale fast searching arrive the imagery zone of this parameter threshold value done with the probability type interpretation by boolean (Boolean) logical operation according to this parameter threshold value by system in 3-dimensional image.The result that will detect marks and stores (step 25) at last.
See also Fig. 3, it is three-dimensional image detecting of the present invention, compile and the process flow diagram of the third step of reconstructing system, compiles the detecting result.The detecting result (step 31) who shows second step in system, the function that has an increase in the system, deletes and compile the border is compiled leaking the result who awards or judge by accident among the detecting result so that the user to be provided.In this increase, deletion, and compile in the function on border, the user compiles the zone that instrument will judge by accident among the result detecting as the image of general pattern editting function and clicks and delete (step 32) and telescopic adjustment (step 34) is done on the border of chosen regional extent except using, also has the instrument that expansion is chosen, fairy maiden's rod, it can be to the border of the automatic extension of particular point of interest to the target area, can be used for doing large-area quick increase action (step 33) to leaking the zone of awarding among the detecting result, its method such as Fig. 5: with the point of interest is the center, near 5 * 5 window ranges of statistics central point are four kinds of attribute parameters of totally 25 pixels, calculate its mean value and standard deviation, and be used as the statistical value of property parameters in the specified window of the threshold value that block grows up with three times standard deviation, the neighbor point of 8 directions that the avalanche ground profile of detecting according to each step is outside gradually expands outwardly search, continue vicinity 8 direction searches at the same attribute pixel that increases newly, till no new point discovery, wherein the statistical value of this property parameters can also directly be applied mechanically the property parameters statistical value of training area in second step.To compile result's storage after compiling end.
See also Fig. 4, it is three-dimensional image detecting of the present invention, compile and the process flow diagram of the 4th step of reconstructing system, and former landform is rebuild.The user chooses earlier a target area (step 41) in 3-dimensional image, the surface configuration of this target area has drop with the surface configuration in zone on every side.Then with the block flop-in method produce automatically near the target area buffer zone (step 42) (Buffer-Zone), estimate original landform in curved surface driving fit (SurfaceFitting) mode, the surface configuration of supposing this target area can numerical value elevation model (Digital Elevation Model, DEM) and ground border vector data then by z=f (x, y) toroidal function is described, capturing dem data in the buffer zone then is used as reference point and finds the solution each coefficient (step 43) of toroidal function, carry out driving fit with least square method again and obtain each coefficient, follow " original " surface configuration (step 44), demonstrate the initial surface shape (step 45) of target area at last with the coefficient calculations target area of being tried to achieve.
How to determine that buffer size is the important problem of part, reference point quantity and target area both proportion of counting just is to the representativeness of the terrain surface after the driving fit.The present invention is the size that decides buffer zone according to certain multiplying power of target area area at present, for example 0.5 times, and 1 times or 2 times etc.This multiplying power can be determined that in case of the present invention, we utilize existing general landform, come the preceding landform of simulation objectives regional change by the user, and how this selects suitable toroidal function to study different variations, and sets suitable buffer size.
About toroidal function, the present invention tests with polynomial function, table 3 is six kinds of employed coefficients of polynomial function, be respectively plane function Planar, bilinear surface function Bi-linear, quadric surface function Q uadratic, biquadratic toroidal function Bi-quadratic, cube toroidal function Cubic to six kinds of two cube toroidal function Bi-cubic etc., its polynomial expression item time respectively from single order to three rank.And Fig. 6 is respectively the example of these six kinds of polynomial expression district faces.
Table 3
| 1 | x | y | xy | x 2 | y 2 | x 2 y | xy 2 | x 2 y 2 | x 3 | y 3 | x 3 y | xy 3 | x 3 y 2 | x 2 y 3 | x 3 y 3 | |
| The plane function | V | V | V | |||||||||||||
| The bilinear surface function | V | V | V | V | ||||||||||||
| The quadric surface function | V | V | V | V | V | V | ||||||||||
| The biquadratic toroidal function | V | V | V | V | V | V | V | V | V | |||||||
| Cube toroidal function | V | V | V | V | V | V | V | V | V | V | ||||||
| Two cubes of toroidal functions | V | V | V | V | V | V | V | V | V | V | V | V | V | V | V | V |
Three-dimensional image detecting of the present invention, compile and reconstructing system is applicable to large-scale monitoring is done on the face of land, the scope of monitoring has comprised the monitoring of landslide and monitoring of forest group or the like.When being applied to the monitoring of landslide, native system reach by unloaded light and the data of digital aerophoto in obtain data such as DSM, DEM, OHM and different visible light wave band, utilize these data to depict the 3-dimensional image map.At TaiWan, China, the situation on common avalanche ground is that soil has exposed phenomenon, near forest is dense and landform is precipitous, and wherein the exposed of soil makes slightly degree of making increase of the face of land, so near so viridescent sharp contrast of the dense meeting of the big and forest of the precipitous gradient of landform.Therefore, the zone of being found out avalanche by the user on the 3-dimensional image map is as selected training area, system can analyze and add up roughness of ground surface in the training area, the gradient, the mean value of object height and visible light wave range property parameters such as (green degree) and standard deviation to set the threshold that detecting is differentiated, just on the 3-dimensional image map, do large-scale Auto-Sensing search by boolean calculation then and the zone that will detect marks situation for record avalanche generation with the interpretation of probability type according to the threshold value of each parameter.Utilize the instrument of compiling of native system then, the Auto-Sensing result is done to increase/delete compiling of zone and telescopic adjustment border, to increase detecting result's accuracy with manual type.At last can be former the landform function of rebuilding produce buffer zone around in the avalanche zone, and choose numerical value elevation model (Digital Elevation Model in this buffer zone, DEM) be used as reference point, estimate the preceding landform of avalanche in curved surface driving fit (Surface Fitting) mode again, obtain the earth flow vector or the accumulating amount on avalanche ground at last in the mode of subtracting each other.So the number of dropouts that just can assess native stone in the avalanche zone is further assessed the influence of avalanche to the area, downstream, as the influence to reservoir sedimentation.
Aspect the monitoring of forest group, because the performance of chlorophyll and kenel has nothing in common with each other between different types of forest, so focus on the resolution of visible light wave range, in order to make the fast investigation that different groups distribute, its investigation result can be done equally and manually compile and the original appearance reconstruction, and wherein original appearance is rebuild the assessment that can be used for forest volume of timber amount.This investigation result can be used as the assessment of the forest reserves or timber resources, and is high as the gradient, atural object if cooperate the investigation of other landform attribute again, just can be used for the usefulness of the planning of land development.
In sum, the three-dimensional image detecting of the present invention design, compile and reconstructing system can accurately and fast judge whether the generation of face of land transition, can also estimate simultaneously the seriousness of variation or disaster (as landslide), and then earthquake, influence that typhoon or torrential rain caused reached the function of detecting real-time and caution, reach the detecting effect better than known technology.The real innovative design that belongs to difficult energy, dark tool industrial value, file an application in the whence in accordance with the law.
The present invention must be appointed by the person skilled in the art and executes that the craftsman thinks and be to modify the content of right neither disengaging farsighted person's scope of the present invention institute desire protection as all.
Claims (22)
1, a kind of 3-dimensional image automatic detection method is characterized in that, its step comprises:
Receiving a digitized video and a unloaded light reaches data and captures one first parameter;
Make a 3-dimensional image according to this first parameter;
In this 3-dimensional image, choose a target and capture one second parameter of this target;
Calculate this second parameter to obtain producing a threshold value of this target; And
According to this threshold value Auto-Sensing and choose a zone that reaches this threshold value in this 3-dimensional image.
2,3-dimensional image automatic detection method as claimed in claim 1 is characterized in that, wherein this first parameter comprises a visible light wave range, a numerical value terrain model, a numerical value elevation model and an object height model.
3,3-dimensional image automatic detection method as claimed in claim 1 is characterized in that, wherein this second parameter comprises a roughness of ground surface, a gradient, a ground object height and a visible light wave range.
4,3-dimensional image automatic detection method as claimed in claim 1 is characterized in that, wherein this digitized video be selected from a digital airborne photography image or a digital satellite photographic image one of them.
5,3-dimensional image automatic detection method as claimed in claim 1 is characterized in that, wherein this target is a landslide image.
6, a kind of 3-dimensional image is compiled method, it is characterized in that, its step comprises:
Receive a digitized video and a unloaded light and reach data to make a 3-dimensional image;
In this 3-dimensional image, choose a target and capture a parameter of this target;
Set up a circle selection tool of searching in order to expansion according to this parameter; And
Use this circle selection tool to increase, delete and edit this target.
7,3-dimensional image as claimed in claim 6 is compiled method, it is characterized in that, wherein this parameter comprises a roughness of ground surface, a gradient, a ground object height and a visible light wave range.
8,3-dimensional image as claimed in claim 6 is compiled method, it is characterized in that, wherein this circle selection tool is fairy maiden's rod.
9,3-dimensional image as claimed in claim 6 is compiled method, it is characterized in that, wherein this target is a landslide image.
10, a kind of 3-dimensional image automatic reconstruction method is characterized in that, its step comprises:
Receive a digitized video and a unloaded light and reach data to make a 3-dimensional image;
In this 3-dimensional image, choose a target;
Set up a buffer zone around this target and capture a parameter of this buffer zone; And
Carrying out a curved surface driving fit according to this parameter calculates to rebuild an original state of this target.
11,3-dimensional image automatic reconstruction method as claimed in claim 10 is characterized in that, wherein this parameter comprises a numerical value elevation model and border, ground vector data.
12,3-dimensional image automatic reconstruction method as claimed in claim 10 is characterized in that, wherein this curved surface driving fit is calculated and utilized a toroidal function to carry out.
13,3-dimensional image automatic reconstruction method as claimed in claim 12, wherein this toroidal function be selected from a plane function, a bilinear surface function, a quadric surface function, a pair of quadric surface function, one cube of toroidal function and a pair of cube toroidal function one of them.
14,3-dimensional image automatic reconstruction method as claimed in claim 10 is characterized in that, wherein this digitized video be selected from a digital airborne photography image or a digital satellite photographic image one of them.
15,3-dimensional image automatic reconstruction method as claimed in claim 10 is characterized in that, wherein this target is a landslide image.
16, a kind of three-dimensional image detecting, compile and reconstructing system, it is characterized in that, comprise:
One input and acquisition unit in order to receiving a digitized video and a unloaded light reaches data, and captures one first parameter to make a 3-dimensional image;
One training and interpretation unit, be connected with this input and acquisition unit, in order in this 3-dimensional image, to choose a target and to capture one second parameter of this target, and calculate this second parameter obtaining producing a threshold value of this target, and in this 3-dimensional image, detect and choose zone with this target similar performance according to this threshold value;
One shows and compiles the unit, links to each other with this training and interpretation unit, and in order to set up a circle selection tool of searching in order to expansion according to this threshold value, this circle selection tool can be used for increasing, deleting and edits one of this training and interpretation unit and detect the result; And
One reconstruction unit is connected with this training and interpretation unit, in order to having chosen one the 3rd parameter of setting up a buffer zone around the target and capturing this buffer zone at this, and carries out a curved surface driving fit according to the 3rd parameter and calculates to rebuild an original state of this target.
17 three-dimensional image detectings as claimed in claim 16, compile and reconstructing system, it is characterized in that wherein this first parameter comprises a visible light wave range, a numerical value terrain model, a numerical value elevation model and an object height model.
18, three-dimensional image detecting as claimed in claim 16, compile and reconstructing system, it is characterized in that wherein this second and the 3rd parameter comprises a roughness of ground surface, a gradient, a ground object height and a visible light wave range respectively.
19, three-dimensional image detecting as claimed in claim 16, compile and reconstructing system, it is characterized in that wherein this digitized video is to be selected from a digital airborne photography image or a digital satellite photographic image ㄧ wherein.
20, three-dimensional image detecting as claimed in claim 16, compile and reconstructing system, it is characterized in that wherein this target is a landslide image.
21, three-dimensional image detecting as claimed in claim 16, compile and reconstructing system, it is characterized in that wherein this curved surface driving fit is calculated and utilized a toroidal function to carry out.
22, three-dimensional image detecting as claimed in claim 21, compile and reconstructing system, wherein this toroidal function be selected from a plane function, a bilinear surface function, a quadric surface function, a pair of quadric surface function, one cube of toroidal function and a pair of cube toroidal function one of them.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2007101621697A CN101464148B (en) | 2007-12-21 | 2007-12-21 | Three-dimensional image detecting, compiling and reconstructing system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2007101621697A CN101464148B (en) | 2007-12-21 | 2007-12-21 | Three-dimensional image detecting, compiling and reconstructing system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101464148A true CN101464148A (en) | 2009-06-24 |
| CN101464148B CN101464148B (en) | 2012-05-23 |
Family
ID=40804837
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2007101621697A Expired - Fee Related CN101464148B (en) | 2007-12-21 | 2007-12-21 | Three-dimensional image detecting, compiling and reconstructing system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101464148B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102819023A (en) * | 2012-07-27 | 2012-12-12 | 中国地质大学（武汉） | Method and system of landslide recognition of complicated geological background area based on LiDAR |
| CN106575475A (en) * | 2014-08-21 | 2017-04-19 | 松下知识产权经营株式会社 | Information management device, vehicle, and information management method |
| CN108344397A (en) * | 2017-12-28 | 2018-07-31 | 中国公路工程咨询集团有限公司 | Automation modeling method, system and its auxiliary device based on oblique photograph technology |
| CN110213003A (en) * | 2019-05-21 | 2019-09-06 | 北京科技大学 | A kind of wireless channel large-scale fading modeling method and device |
| CN110569786A (en) * | 2019-09-06 | 2019-12-13 | 中国农业科学院农业资源与农业区划研究所 | fruit tree identification and quantity monitoring method and system based on unmanned aerial vehicle data acquisition |
| CN110645960A (en) * | 2018-06-26 | 2020-01-03 | 凌上科技(北京)有限公司 | Distance measurement method, terrain following distance measurement method, obstacle avoidance distance measurement method and device |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3622094B2 (en) * | 1995-10-05 | 2005-02-23 | 株式会社日立製作所 | Map update support apparatus and map information editing method |
-
2007
- 2007-12-21 CN CN2007101621697A patent/CN101464148B/en not_active Expired - Fee Related
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102819023A (en) * | 2012-07-27 | 2012-12-12 | 中国地质大学（武汉） | Method and system of landslide recognition of complicated geological background area based on LiDAR |
| CN102819023B (en) * | 2012-07-27 | 2014-09-17 | 中国地质大学（武汉） | Method and system of landslide recognition of complicated geological background area based on LiDAR |
| CN106575475A (en) * | 2014-08-21 | 2017-04-19 | 松下知识产权经营株式会社 | Information management device, vehicle, and information management method |
| CN106575475B (en) * | 2014-08-21 | 2019-08-09 | 松下知识产权经营株式会社 | Apparatus for management of information, vehicle and approaches to IM |
| CN108344397A (en) * | 2017-12-28 | 2018-07-31 | 中国公路工程咨询集团有限公司 | Automation modeling method, system and its auxiliary device based on oblique photograph technology |
| CN110645960A (en) * | 2018-06-26 | 2020-01-03 | 凌上科技(北京)有限公司 | Distance measurement method, terrain following distance measurement method, obstacle avoidance distance measurement method and device |
| CN110213003A (en) * | 2019-05-21 | 2019-09-06 | 北京科技大学 | A kind of wireless channel large-scale fading modeling method and device |
| CN110569786A (en) * | 2019-09-06 | 2019-12-13 | 中国农业科学院农业资源与农业区划研究所 | fruit tree identification and quantity monitoring method and system based on unmanned aerial vehicle data acquisition |
| CN110569786B (en) * | 2019-09-06 | 2022-03-29 | 中国农业科学院农业资源与农业区划研究所 | Fruit tree identification and quantity monitoring method and system based on unmanned aerial vehicle data acquisition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101464148B (en) | 2012-05-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| TWI353561B (en) | 3d image detecting, editing and rebuilding system | |
| Gigli et al. | Semi-automatic extraction of rock mass structural data from high resolution LIDAR point clouds | |
| Wilson | Digital terrain modeling | |
| Maguya et al. | Adaptive algorithm for large scale dtm interpolation from lidar data for forestry applications in steep forested terrain | |
| CN101464148B (en) | Three-dimensional image detecting, compiling and reconstructing system | |
| CN110988909B (en) | TLS-based vegetation coverage measuring method for sand vegetation in severe cold fragile area | |
| Hsiao et al. | Change detection of landslide terrains using ground-based LiDAR data | |
| Mason et al. | Data utilization in flood inundation modelling | |
| Vetter et al. | Vertical Vegetation structure analysis and hydraulic roughness determination using dense ALS point cloud data-A voxel based approach | |
| Rau et al. | Semi-automatic shallow landslide detection by the integration of airborne imagery and laser scanning data | |
| CN114674277A (en) | Method for monitoring surface subsidence of deep mining and evaluating land reclamation suitability by combining whole-field survey lines | |
| de Oliveira et al. | Obtaining morphometric variables from gullies using two methods of interpolation laser scanner data: the case study of Vassouras, Brazil | |
| Niederheiser et al. | Dense image matching of terrestrial imagery for deriving high-resolution topographic properties of vegetation locations in alpine terrain | |
| CN112166688B (en) | Method for monitoring desert and desertification land based on minisatellite | |
| Liu et al. | Architecture planning and geo-disasters assessment mapping of landslide by using airborne LiDAR data and UAV images | |
| Tian et al. | A process-oriented method for rapid acquisition of canopy height model from RGB point cloud in semiarid region | |
| Soderman et al. | Three-dimensional environment models from airborne laser radar data | |
| Liu et al. | Processing outcrop point clouds to 3D rock structure using open source software | |
| Liu et al. | The geomorphometry of rainfall-induced landslides in taiwan obtained by airborne lidar and digital photography | |
| Graham et al. | Remote sensing applications for hedgerows | |
| Wu et al. | Voxel-based marked neighborhood searching method for identifying street trees using vehicle-borne laser scanning data | |
| Rutzinger et al. | Object detection in airborne laser scanning data-an integrative approach on object-based image and point cloud analysis | |
| Gao | Algorithms and software tools for extracting coastal morphological information from airborne LiDAR data | |
| Che et al. | A Simple and Reliable Method for Estimating Building-scale Height Based on Multi-source Datasets | |
| Dowman | Automatic feature extraction for urban landscape models |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20120523 Termination date: 20191221 |