CN-121999134-A - Automatic engineering measurement point cloud data processing method and system

CN121999134ACN 121999134 ACN121999134 ACN 121999134ACN-121999134-A

Abstract

The invention relates to the technical field of three-dimensional reconstruction, in particular to an automatic processing method and system for engineering measurement point cloud data. According to the invention, the incident angle characteristic and the intensity attenuation characteristic in single-point observation are converted into the range error variance and the credibility weight constraint is formed, so that coordinate compensation correction can be differentiated and converged along with the change of observation conditions, the misjudgment influence of fixed threshold value elimination on complex terrains is weakened, residual deviation after multi-station splicing is restrained, geometric attributes such as a main curvature direction and Gaussian curvature are obtained through neighborhood quadric surface fitting based on corrected coordinates, curvature gradient is synthesized by combining curvature difference and space displacement, and texture feature vector histogram is formed through projection statistics, synchronous expression of surface morphology continuity and texture stability is realized, and stable expression of a point cloud model in terms of slope detail, boundary definition and engineering interpretation consistency is improved.

Inventors

Li Liane
ZHAO SHANSHAN
ZHAO KETAO
Ma Caigu
LIU GUOQING
LIU JIANHUA
SUN JIAN
ZHOU ZHUO

Assignees

云南省测绘地理信息科技发展有限公司

Dates

Publication Date: 20260508
Application Date: 20260123

Claims (10)

1. The automatic engineering measurement point cloud data processing method is characterized by comprising the following steps of: s1, scanning and collecting original laser point cloud data of an engineering terrain area, marking each engineering terrain laser foot point in the original laser point cloud data, extracting space coordinates and space characteristics of each engineering terrain laser foot point, and constructing a terrain single-point observation data set; S2, calculating the range error variance of each engineering terrain laser foot point based on the terrain single-point observation data set; S3, constructing an observation residual error objective function by taking the space coordinates of engineering terrain laser foot points in the terrain single-point observation data set as iteration initial values, calculating a weighting coefficient of observation reliability based on the ranging error variance, and superposing the weighting coefficient into the observation residual error objective function, and compensating and correcting the space coordinates to obtain corrected space coordinates; S4, obtaining the corrected space coordinates and the corresponding space neighborhood points corresponding to each engineering terrain laser foot point, and performing local quadric surface fitting to obtain a terrain surface curvature geometrical attribute set of the local quadric surface at the engineering terrain laser foot point; And S5, extracting texture features of engineering survey terrain according to the terrain surface curvature geometric attribute set, and constructing an engineering terrain mapping point cloud model.
2. The automatic engineering survey point cloud data processing method according to claim 1, wherein the terrain single-point observation data set comprises spatial coordinates and corresponding spatial features of engineering terrain laser foot points, the ranging error variance is specifically obtained by calculating a weighted sum of a tangent parameter of a laser incidence angle and an inverse intensity parameter, the corrected spatial coordinates are specifically obtained by compensating and correcting the spatial coordinates of the engineering terrain laser foot points by using an optimal coordinate adjustment amount, the terrain surface curvature geometric attribute set comprises a maximum principal curvature direction vector, a minimum principal curvature direction vector and a terrain gaussian curvature, and the engineering terrain mapping point cloud model comprises a geometric position field of the corrected spatial coordinates and a texture attribute field of a terrain texture feature vector histogram.
3. The automatic engineering measurement point cloud data processing method according to claim 1, wherein step S1 specifically comprises: s101, scanning an engineering terrain area to collect an original laser point cloud data stream, identifying a plurality of discrete laser echo signals formed by laser pulse reflection in the original laser point cloud data stream, marking each discrete laser echo signal as an engineering terrain laser foot point, reading the space coordinate, the laser echo intensity and the corresponding laser emission center space coordinate of each engineering terrain laser foot point, and generating original laser point cloud space attribute data; S102, calculating Euclidean distance between the current engineering terrain laser foot point and other engineering terrain laser foot points in an engineering terrain area based on the space coordinates of the engineering terrain laser foot points in the original laser point cloud space attribute data, screening the engineering terrain laser foot points with the distance smaller than a preset neighborhood radius threshold value, marking the engineering terrain laser foot points as corresponding space neighborhood points, fitting a local surface tangent plane together according to the coordinates of the engineering terrain laser foot points and the corresponding space neighborhood points, and calculating a surface normal vector perpendicular to the local surface tangent plane; s103, integrating the laser echo intensity of the engineering terrain laser foot point, the laser emission center space coordinate and the earth surface normal vector in the original laser point cloud space attribute data to construct the space feature of the terrain physical attribute as a terrain single-point observation data set.
4. The automatic engineering measurement point cloud data processing method according to claim 1, wherein step S2 specifically includes: S201, extracting the space coordinates of a laser emission center and the space coordinates of engineering terrain laser foot points associated with each engineering terrain laser foot point in the terrain single-point observation data set, calculating the vector difference of the space coordinates of the laser emission center pointing to the space coordinates, obtaining a laser beam transmission direction vector, calculating the cosine value of an included angle between the laser beam transmission direction vector and a ground surface normal vector, and performing back calculation to obtain the laser incidence angle when the laser beam contacts the ground surface; S202, calculating a tangent function value of the laser incidence angle, defining the tangent function value as a tangent parameter of the laser incidence angle, reading laser echo intensity of the terrain single-point observation data set, calculating the reciprocal of a laser echo intensity value, and defining the reciprocal of the intensity parameter as an intensity reciprocal parameter representing signal attenuation characteristics; and S203, respectively distributing corresponding weight coefficients to the laser incidence angle tangent parameters and the intensity reciprocal parameters, calculating the sum of the weighted laser incidence angle tangent parameters and the weighted intensity reciprocal parameters, and quantifying the distance deviation degree of each engineering terrain laser foot point in the measuring process to generate a range error variance.
5. The automatic engineering measurement point cloud data processing method according to claim 1, wherein step S3 specifically includes: S301, constructing an observation residual error objective function by taking the space coordinates of engineering terrain laser foot points in the terrain single-point observation data set as reference iteration initial values, calculating a weighting coefficient of observation credibility based on the ranging error variance, introducing the weighting coefficient into the observation residual error objective function, and carrying out weighting treatment on the observation residual error items in the observation residual error objective function to obtain a weighted observation residual error objective function; s302, continuously minimizing residual values in the weighted observation residual objective function in an iterative process, extracting correction vectors aiming at the current engineering terrain laser foot points, quantifying the direction and distance of the engineering terrain laser foot points to be moved in space, and generating optimal coordinate adjustment quantity; And S303, superposing the optimal coordinate adjustment quantity on the space coordinates of the original engineering terrain laser foot points in the terrain single-point observation data set, carrying out point-by-point compensation correction on the original coordinates, updating the space coordinate information of all the engineering terrain laser foot points in the terrain single-point observation data set, and generating corrected space coordinates.
6. The automatic engineering measurement point cloud data processing method according to claim 1, wherein step S4 specifically includes: S401, redetermining a space neighborhood point corresponding to each engineering terrain laser foot point based on the corrected space coordinates, fitting a local quadric surface by utilizing coordinates of the engineering terrain laser foot points and the space neighborhood points, calculating second derivative information and curvature tensor of the local quadric surface at the engineering terrain laser foot points, and quantifying the bending degree and geometric form of the earth surface at the engineering terrain laser foot points to generate surface curvature geometric feature quantity; S402, performing feature decomposition operation on the surface curvature geometric feature quantity, solving a maximum principal curvature feature value and a minimum principal curvature feature value, and obtaining a maximum principal curvature feature vector corresponding to the maximum principal curvature feature value and a minimum principal curvature feature vector corresponding to the minimum principal curvature feature value; S403, calculating the terrain Gaussian curvature according to the maximum principal curvature characteristic value and the minimum principal curvature characteristic value, and combining the maximum principal curvature characteristic vector and the minimum principal curvature characteristic vector to construct a terrain surface curvature geometric attribute set.
7. The automatic engineering measurement point cloud data processing method according to claim 6, wherein step S5 specifically includes: s501, extracting a terrain Gaussian curvature from the terrain surface curvature geometric attribute set, calculating a difference value of the terrain Gaussian curvature between an engineering terrain laser foot point and a corresponding space neighborhood point by combining the corrected space coordinate, and combining a space displacement vector between the engineering terrain laser foot point and the corresponding space neighborhood point to synthesize a local curvature gradient vector representing curvature change trend; S502, respectively projecting the local curvature gradient vector onto the maximum principal curvature characteristic vector and the minimum principal curvature characteristic vector, calculating a projection component value of the local curvature gradient vector on the maximum principal curvature characteristic vector and a projection component value of the local curvature gradient vector on the minimum principal curvature characteristic vector, counting cumulative frequency of the projection component falling in each preset interval, and constructing a terrain texture characteristic vector histogram according to the cumulative frequency distribution condition; S503, constructing a point cloud data structure comprising a geometric position field and a texture attribute field, writing the corrected space coordinates into the geometric position field, writing the topographic texture feature vector histogram into the texture attribute field, integrating information of all engineering topographic laser foot points, and generating an engineering topographic mapping point cloud model.
8. The automatic engineering measurement point cloud data processing method according to claim 5, wherein the process of constructing the observation residual objective function is specifically: extracting the space coordinates of engineering terrain laser foot points in the terrain single-point observation data set, and setting the space coordinates of the engineering terrain laser foot points as position variables to be optimized; Acquiring a laser emission center space coordinate corresponding to an engineering terrain laser foot point and an original laser ranging observation value; calculating the geometric distance between the space coordinates of the laser emission center and the space coordinates of the engineering terrain laser foot points; Calculating the numerical value difference between the geometric distance and the original laser ranging observation value, and defining the numerical value difference as a single-point distance observation residual error; square operation is carried out on the single-point distance observation residual error, and a residual error square term is obtained; multiplying the weighted coefficient of the observation reliability as a weight factor by a residual square term to obtain a weighted residual square term; and executing accumulation summation on weighted residual error square items corresponding to all engineering terrain laser foot points in the terrain single-point observation data set, and determining an accumulation summation result as an observation residual error objective function.
9. The automatic engineering measurement point cloud data processing method according to claim 6, wherein the process of fitting the local quadric is specifically: Extracting corrected space coordinates corresponding to engineering terrain laser foot points and space neighborhood points in the terrain single-point observation data set; Constructing a local coordinate system by taking the corrected space coordinates of the engineering terrain laser foot points as an original point, and calculating relative coordinate data of the space neighborhood points under the local coordinate system; Constructing a polynomial equation comprising a quadratic term, a cross term and a first term of the plane coordinate component, and establishing a mathematical mapping relation between the height Cheng Fenliang in the relative coordinate data and the plane coordinate component; Substituting the relative coordinate data into a polynomial equation to establish an overdetermined equation set of the deviation between the actual observation position and the theoretical curved surface position; based on a least square optimization criterion, solving an overdetermined equation set with the square sum of the minimized deviation as a target; and obtaining undetermined parameter values in a polynomial equation, and determining a local quadric surface approaching to the spatial distribution characteristics of the engineering terrain laser foot points and the spatial neighborhood points.
10. An automatic engineering measurement point cloud data processing system, wherein the automatic engineering measurement point cloud data processing method according to any one of claims 1 to 9 is executed, and the system comprises: The terrain point cloud single-point construction module scans and collects original laser point cloud data of an engineering terrain area, marks each engineering terrain laser foot point in the original laser point cloud data, extracts space coordinates and space characteristics of each engineering terrain laser foot point, and constructs a terrain single-point observation data set; The distance measurement error variance estimation module is used for calculating the distance measurement error variance of each engineering terrain laser foot point based on the terrain single-point observation data set; The weighted coordinate correction module is used for constructing an observation residual objective function by taking the space coordinates of engineering terrain laser foot points in the terrain single-point observation data set as iteration initial values, calculating the weighted coefficient of the observation reliability based on the ranging error variance, superposing the weighted coefficient into the observation residual objective function, and compensating and correcting the space coordinates to obtain corrected space coordinates; The local curved surface curvature calculation module is used for obtaining the corrected space coordinates and the corresponding space neighborhood points corresponding to each engineering terrain laser foot point, carrying out local quadric surface fitting, and obtaining a terrain surface curvature geometric attribute set of the local quadric surface at the engineering terrain laser foot point; and the terrain texture modeling module extracts texture features of engineering survey terrains according to the terrain surface curvature geometric attribute set, and builds an engineering terrain mapping point cloud model.

Description

Automatic engineering measurement point cloud data processing method and system Technical Field The invention relates to the technical field of three-dimensional reconstruction, in particular to an automatic processing method and system for engineering measurement point cloud data. Background The technical field of three-dimensional reconstruction is a technical field taking the real geometric form restoration of a space target or a scene as core content, and expresses and organizes discrete space information by acquiring space coordinate information and geometric feature data of an object or an environment so as to form a three-dimensional data result for analysis, calculation and modeling. The automatic processing method of engineering measurement point cloud data is a method for processing a large amount of point cloud data acquired in an engineering measurement process according to a set rule by a pointer, and aims at technical matters that original point cloud is tidied into standard data which can be directly used for engineering analysis, three-dimensional coordinates and intensity information recorded by each point are read point by point usually by importing a point cloud file according to a measurement site or a scanning sequence, abnormal points are judged by adopting a fixed threshold value and deleted, then coordinate unification of multiple measurement site clouds is completed according to a preset coordinate conversion relation, classification marking is carried out on the point clouds according to elevation or plane position conditions, and finally a point cloud data set meeting engineering measurement requirements is output. In the prior art, coordinates and intensity information are imported according to a measuring station or scanning sequence and are read point by point, abnormal points are removed and depend on a fixed threshold value, error differences caused by different incident angles and echo attenuation are difficult to adapt, the coordinates are unified, reliability constraint and compensation mechanisms for single-point ranging deviation are mainly lacked according to a preset conversion relation, joint discrimination of local curved surface morphology and curvature change trend is lacked due to classification of marked off-peak elevation or plane position conditions, excessive filtering or mistaken retention is easy to occur at a slope fold transition region and a detail structure, surface relief is smoothed, edges are broken, textures are aliased, and geometric consistency and engineering analysis reliability of a result point cloud are limited. Disclosure of Invention The invention aims to solve the defects in the prior art, and provides an automatic processing method and system for engineering measurement point cloud data. In order to achieve the above purpose, the invention adopts the following technical scheme that the automatic processing method for engineering measurement point cloud data comprises the following steps: s1, scanning and collecting original laser point cloud data of an engineering terrain area, marking each engineering terrain laser foot point in the original laser point cloud data, extracting space coordinates and space characteristics of each engineering terrain laser foot point, and constructing a terrain single-point observation data set; S2, calculating the range error variance of each engineering terrain laser foot point based on the terrain single-point observation data set; S3, constructing an observation residual error objective function by taking the space coordinates of engineering terrain laser foot points in the terrain single-point observation data set as iteration initial values, calculating a weighting coefficient of observation reliability based on the ranging error variance, and superposing the weighting coefficient into the observation residual error objective function, and compensating and correcting the space coordinates to obtain corrected space coordinates; S4, obtaining the corrected space coordinates and the corresponding space neighborhood points corresponding to each engineering terrain laser foot point, and performing local quadric surface fitting to obtain a terrain surface curvature geometrical attribute set of the local quadric surface at the engineering terrain laser foot point; And S5, extracting texture features of engineering survey terrain according to the terrain surface curvature geometric attribute set, and constructing an engineering terrain mapping point cloud model. The invention improves that the terrain single-point observation data set comprises space coordinates and corresponding space features of engineering terrain laser foot points, the distance measurement error variance is specifically obtained by calculating weighted sum of tangent parameters of laser incidence angles and inverse parameters of intensities, the corrected space coordinates are specifically obtained by compensating and correcting the space c