A kind of method from airborne laser point cloud data identification overhead power transmission line
Technical field
The present invention relates to a kind of method from airborne laser point cloud data identification overhead power transmission line, belong to the method field of LiDAR data Objects extraction treatment technology.
Background technology
LiDAR(Light Detection And Ranging) be a kind of emerging earth observation technology, which provide a kind of can fast, high precision, obtain the technological means of earth's surface three-dimensional information in real time, different from traditional observation technology, what LiDAR obtained is a large amount of earth's surfaces unique point in three dimensions, the spatial positional information of ground and atural object can be provided exactly, be widely used in now the many aspects such as Digital Mapping, forestry monitoring, resource investigation.What obtain due to LiDAR is the cloud data of whole observation area, so the attribute of LiDAR point cloud is resolved and classification distinguishes that the effect for LiDAR has great significance, therefore Objects extraction becomes the important content in LiDAR subsequent treatment, and the extraction of (making somebody a mere figurehead) line of electric force and analysis are one of them study hotspots.Accurately and intactly obtain power line data and have very important status at many subject neighborhoods, it is widely used in all many-sides such as estimation, Ecological in Urban assessment of biomass and carbon amounts, so how effectively to obtain line of electric force point accurately and efficiently to have important theory value and actual value from raw LiDAR data.
At present, the research of LiDAR line of electric force acquisition of information focuses mostly in the following aspects: (1) utilizes the spatial data such as remote sensing, aviation image as supplementary means, merges with LiDAR point cloud the extraction carrying out line of electric force mutually; (2) be grating image by LiDAR point cloud interpolation, then utilize the method process of classification of remote-sensing images, indirectly reach the effect of LiDAR point cloud line of electric force information extraction; (3) using some characteristic of LiDAR as foundation, carry out terrain classification by the method for machine learning, or utilize the special nature of all-wave, many echoes LiDAR, excavate it obtains foundation feasibility as line of electric force.The core concept of above most of power line extraction method is used for reference the sorting technique of remote sensing image in fact, then be applied on this discrete point set of LiDAR, these methods all have ignored the feature of the special three-dimensional spatial distribution that LiDAR itself has, and most of LiDAR data does not often have the characteristic of Full wave shape and many echoes, special nature cannot be utilized to extract LiDAR point cloud line of electric force, so the acquisition that existing method is used for LiDAR line of electric force exists many defects, the requirement of practical application cannot be met.
Summary of the invention
The technical problem to be solved in the present invention is: overcome above-mentioned the deficiencies in the prior art, proposes a kind of method from airborne laser point cloud data identification overhead power transmission line, can accurately and from LiDAR data, extract overhead power transmission line efficiently.
In order to solve above technical matters, a kind of method from airborne laser point cloud data identification overhead power transmission line provided by the invention, comprises the following steps:
1st step, morphologic filtering is carried out to LiDAR data, removably obtain culture point data acquisition after millet cake;
2nd step, the method utilizing region to increase split described culture point data acquisition, obtain the atural object point set that several are independent;
3rd step, obtain the spatial shape index of each atural object point set in accordance with the following methods:
A, use measure-alike cube with superposition mode remove mulched ground object point collection;
B, the cube that first use larger skirt is long remove mulched ground object point collection;
C, constantly reduce cubical size, until only hold a LiDAR point in each cube;
D, the cube quantity containing LiDAR point according to inside and cubical length of side r, calculate the spatial shape index of logN/logr as this atural object point set;
4th step, the spatial shape index atural object point set between 1.08 to 1.23 is judged to be overhead power transmission line point, and extracts.
Innovative point of the present invention is, first by the filter preprocessing of LiDAR data, original LiDAR data is separated into the set of topocentric set and culture point, in original three-dimensional point set, in order to extract specific atural object, need from cloud data, distinguish topographic(al) point subset sums culture point subset.By LiDAR filtering exclusively millet cake, the data acquisition only comprising culture point can be obtained, extraction and analysis is carried out to specifically object point in the basis of culture point set, then 3D region division and spatial shape is adopted to describe the mode combined, the three dimensions feature intrinsic to LiDAR point cloud atural object itself is evaluated, region segmentation is first utilized to isolate each atural object, obtain object point agglomerated masses one by one, then calculate " the spatial shape index " of atural object, analyze the three-dimensional space shape of atural object and then determine that the classification of atural object tells the some cloud of overhead power transmission line type, thus realize extracting overhead power transmission line terrestrial object information fast and more adequately from LiDAR point cloud.
In traditional LiDAR overhead power transmission line extracting method, often need the support of the auxiliary data such as remote sensing, aerial images or require that LiDAR point cloud data have the characteristics such as Full wave shape, limit the scope of application, and in computation process, these methods usually need to build the data structure such as the triangulation network, grid graticule mesh and to reorganize a cloud and by machine learning assisted extraction such as SVM, computation process is comparatively complicated, make huge, the consuming time length of operand, be difficult to the needs meeting real data process.
And the present invention is split by 3D region just and design a kind of segmentation is described after the mode of " spatial shape index " of atural object morphological feature carry out the differentiation of type of ground objects thus extract the LiDAR point of overhead power transmission line.The spatial shape feature of various kind atural object more adequately can be stated by this " spatial shape index ", the spatial shape desired value of different atural object has the distribution range of oneself substantially, and this matches with space distribution form specific to such atural object.Between the obvious atural object of feature, its spatial shape desired value interval is non-cross, has good Division identification condition, and it is larger to distribute between more complicated its spatial shape Index areas of atural object of form.Overhead power transmission line is as a kind of atural object of form more complicated, its spatial shape index has comparatively fixing interval range, can comparatively accurately also intactly distinguish, and this method only employs the intrinsic three-dimensional coordinate attribute of LiDAR point cloud itself, do not utilize extra attribute and information, greatly widen the scope of application in LiDAR point cloud, there is stronger applicability.
In the present invention the 1st step, the morphologic filtering algorithm being separated LiDAR ground point and culture point is: first morphology erosion operation a little, does to the point in the window ranges centered by this point in traversal institute; If the elevation of the point after erosion operation is less than its original elevation, then carry out a morphological dilations computing with onesize window; Calculate the elevation of every bit after excessive erosion and dilation operation and the difference of original elevation, if difference is greater than predetermined threshold value, then judge that this point is as culture point; Continuous repetition above-mentioned steps is until tell all ground points and culture point.
In the present invention the 2nd step, the method step that region increases is: first three-dimensional cube subdivision is carried out according to the length of side of setting in entirely object point collection region, forms the bounding volume hierarchy (BVH) of atural object point set; The bounding box of certain non-NULL selected, as seed, is then expanded to its field, if neighborhood bounding box is also non-NULL, then the point thinking in neighborhood bounding box is also the point of single atural object, and adds the some set of this atural object; Using the bounding box newly added as new seed, the repetition above-mentioned steps of circulation, until do not have new point to add.
The concrete methods of realizing of the present invention the 3rd step is as follows:
I, set the cubical initial length of side and equal atural object point set Outside Dimensions, and use this cube mulched ground object point collection;
II, all cubes are divided into 8 measure-alike cubes, statistics includes the cube quantity of LiDAR point, if the cube quantity including LiDAR point does not change, then goes to step III; Otherwise repeat this step;
III, calculate the spatial shape index of atural object point set according to the cube radius r before the last time segmentation and the inner cube quantity containing LiDAR point, spatial shape index is logN/logr.
To sum up, the present invention utilizes 3D region to split, and devises " spatial shape index " for describing atural object spatial shape feature, and then proposes a kind of method from airborne laser point cloud data identification overhead power transmission line according to spatial shape index.The method utilizes the means such as some cloud filtering, region growing segmentation, spatial shape metrics evaluation, the extraction of overhead power transmission line Objects recognition, comparatively complete and acquisition LiDAR overhead power transmission line point exactly.Experimental result shows, no matter this method is extracting the needs that integrity degree, precision or speed all can meet preferably reality.
Accompanying drawing explanation
Below in conjunction with accompanying drawing, the present invention is further illustrated:
Fig. 1 is that the present invention uses morphologic filtering millet cake and culture point discretely, obtains culture point set schematic diagram.
Fig. 2 be the present invention over the ground object point data set carry out the schematic diagram of region growing segmentation.
Fig. 3 is that the present invention carries out the schematic diagram of morphological assessment to atural object computer memory morphological index.
Fig. 4 is the schematic diagram that the present invention identifies overhead power transmission line atural object and extracts.
Embodiment
Describe the present invention in detail with reference to the accompanying drawings below, object of the present invention and effect will become more obvious.
The present embodiment, from the method for airborne laser point cloud data identification overhead power transmission line, comprises the following steps:
1st step, raw LiDAR data is carried out LiDAR point cloud filtering exclusively millet cake by morphologic filtering algorithm, obtain the data acquisition (see figure 1) only comprising culture point.
LiDAR point cloud is the gathering of the three-dimensional point spatially presenting stochastic distribution, by LiDAR filtering exclusively millet cake, the data acquisition only comprising culture point can be obtained, extraction and analysis is carried out to specifically object point in the basis of culture point set, substantially increase accuracy and the integrality of extraction, so the pre-service of LiDAR point cloud data is very important for the extraction of LiDAR atural object.There are many filtering algorithms at present, and morphologic filtering algorithm is more typical one, morphologic filtering algorithm can be separated effectively to rough error point and most of atural object, so adopt the filtering based on Morphological Gradient can meet the requirement of LiDAR point cloud data prediction.By obtaining the culture point set that step below needs after Morphological Gradient filtering process.
The morphologic filtering algorithm being separated LiDAR ground point and culture point is: first morphology erosion operation a little, does to the point in the window ranges centered by this point in traversal institute; If the elevation of the point after erosion operation is less than its original elevation, then carry out a morphological dilations computing with onesize window; Calculate the elevation of every bit after excessive erosion and dilation operation and the difference of original elevation, if difference is greater than predetermined threshold value, then judge that this point is as culture point; Continuous repetition above-mentioned steps is until tell all ground points and culture point.
2nd step, the method utilizing region to increase split described culture point data acquisition, obtain the atural object point set that several are independent, as shown in Figure 2.
The method step that region increases is: entirely will carry out three-dimensional cube subdivision according to the setting length of side (2 times that get LiDAR point cloud equalization point distance in this example) in object point collection region, then the bounding box of certain non-NULL is selected at random as seed, then expand to its 26 field, if neighborhood bounding box is also non-NULL, then think that the point in neighborhood bounding box is also the point belonging to this single atural object, constantly carry out expanding until do not have new point to add, now just obtain an independent atural object, this atural object is given an independent numbering.Segmentation is increased with regard to stop area after all atural object has all identified.
3rd step, indicate each atural object by the region growing segmentation of second step after, each atural object is calculated its " spatial shape index " respectively, differentiates that the type attribute of each atural object is sorted out according to the spatial shape index calculated.
In this step, when calculating the spatial shape index of each atural object, in order to calculate facility, that the cube utilizing size identical goes to cover atural object, first go to cover atural object with the long cube of larger skirt, then constantly reduce cubical size, until each cube has just held a three-dimensional LiDAR point, now recording not is empty cube number N and cube length of side r, calculates the spatial shape index of logN/logr as this atural object.Concrete methods of realizing in the present embodiment is as follows:
I, set the cubical initial length of side and equal atural object point set Outside Dimensions, and use this cube mulched ground object point collection;
II, all cubes are divided into 8 measure-alike cubes, statistics includes the cube quantity of LiDAR point, if the cube quantity including LiDAR point does not change, then goes to step III; Otherwise repeat this step;
III, calculate the spatial shape index of atural object point set according to the cube radius r before the last time segmentation and the inner cube quantity containing LiDAR point, spatial shape index is logN/logr.
4th step, according to each atural object spatial shape index, atural object point set spatial shape index be greater than between 1.08 to 1.23 is labeled as overhead power transmission line, by the Objects extraction being designated overhead power transmission line out thus obtain overhead power transmission line point, sees Fig. 4.
The spatial shape index of dissimilar atural object has different interval ranges, different classifications is belonged to according to respective desired value, in this example, just overhead power transmission line is thought for the atural object of spatial shape index between 1.08 to 1.23, and is not in and is just designated non-overhead power transmission line atural object in this span.
Pass through previous step, each atural object is identified the classification belonging to it, travel through in all atural object, once find that there is the atural object being designated overhead power transmission line, just the LiDAR in this atural object is exported, the complete overhead power transmission line culture point of arriving after traversal.
Through above-mentioned process, final LiDAR overhead power transmission line point cloud can be obtained.Fig. 1-Fig. 4, for using this method, carries out to one piece of LiDAR data each step design sketch that overhead power transmission line extraction process obtains.Can see, each step process is extracted around LiDAR overhead power transmission line all well and is launched, and does not occur obviously departing from.Therefore, finally obtain result and meet there is better integrality and accuracy, meet the requirement that LiDAR overhead power transmission line extracts.
In addition to the implementation, the present invention can also have other embodiments.All employings are equal to the technical scheme of replacement or equivalent transformation formation, all drop on the protection domain of application claims.