CN-116071393-B - Time sequence multidimensional point cloud deformation identification method, system, electronic equipment and storage medium

CN116071393BCN 116071393 BCN116071393 BCN 116071393BCN-116071393-B

Abstract

The invention discloses a time sequence multi-dimensional point cloud deformation identification method, a system, electronic equipment and a storage medium, wherein the time sequence multi-dimensional point cloud deformation identification method comprises the steps of constructing a time sequence multi-dimensional point cloud based on multi-frame point clouds on the surface of a measured object subjected to deformation, meshing the time sequence multi-dimensional point clouds, carrying out point-by-point deformation tracking on the meshed time sequence multi-dimensional point clouds, establishing a point-to-point matching relation of the time sequence multi-dimensional point clouds based on deformation tracking results, and constructing a local deformation descriptor and a global deformation descriptor based on the point-to-point matching relation so as to identify the deformation process of the surface of the measured object. The method is applied to small scene fine full-field deformation identification, a comprehensive and complete identification method is provided for deformation of the surface of the object by constructing local and global deformation descriptors in the time sequence multidimensional point cloud, the whole process of deformation of the surface of the object to be measured is accurately depicted from the local and global angles, the deformation rule is reflected better, and the deformation mechanism and characteristics are revealed.

Inventors

LIU CHUN
ZHAO SHUANG
ZHANG JING
Akronmu Akbar

Assignees

同济大学

Dates

Publication Date: 20260512
Application Date: 20230209

Claims (8)

1. The time sequence multidimensional point cloud deformation identification method is characterized by comprising the following steps of: constructing a time sequence multidimensional point cloud based on the multi-frame point cloud of the surface of the measured object subjected to deformation; gridding the time sequence multidimensional point cloud; carrying out point-by-point deformation tracking on the time sequence multidimensional point cloud after grid meshing; Establishing a point-to-point matching relationship of the time sequence multidimensional point cloud based on a deformation tracking result; The point pair matching relationship is a matching relationship between a previous frame of multi-dimensional point cloud and a later frame of multi-dimensional point cloud of the time sequence multi-dimensional point cloud; Constructing a local deformation descriptor and a global deformation descriptor based on the point pair matching relationship so as to identify the deformation process of the surface of the measured object; the step of constructing the local deformation descriptor based on the point-to-point matching relationship comprises the following steps: Based on the point pair matching relationship, discrete space-time coordinate information, discrete displacement vectors, discrete strain vectors and discrete deformation rate vectors are obtained; Based on the discrete space-time coordinate information, the discrete displacement vector, the discrete strain vector and the discrete deformation rate vector, constructing a local deformation descriptor through the following formula: ; Wherein, LDD represents a local deformation descriptor, a represents discrete space-time coordinate information, delta represents discrete displacement vector, e represents discrete strain vector, v represents discrete deformation rate vector, and T represents transpose symbol; The step of constructing a global deformation descriptor based on the point-to-point matching relationship comprises the following steps: Interpolation is carried out on the discrete space-time coordinate information, the discrete displacement vector, the discrete strain vector and the discrete deformation rate vector to obtain a continuous displacement field, a continuous strain field and a continuous speed field of a preset time point and a preset space position; the displacement field is a function of the displacement state of the surface of the measured object, the strain field is a function of the strain state of the surface of the measured object, and the speed field is a function of the deformation rate of the surface of the measured object; based on the continuous displacement field, continuous strain field and continuous velocity field, a global deformation descriptor is constructed by the following formula: ; Wherein the said The global deformation descriptor is characterized in that, In order to be a displacement field, In order to be able to act as a strain field, For the velocity field, x, y, z represent three-dimensional coordinate information of x axis, y axis and z axis of a point of a multi-dimensional point cloud of a certain frame, T represents time information of a point of a multi-dimensional point cloud of a certain frame, and T represents transposed symbols.
2. The method for identifying time-series multidimensional point cloud deformation according to claim 1, wherein the step of constructing the time-series multidimensional point cloud comprises the following steps: collecting multi-frame point clouds on the surface of the deformed measured object according to a preset time interval; the step of constructing the time sequence multidimensional point cloud comprises the following steps: constructing a time sequence multidimensional point cloud based on the acquired multi-frame point cloud according to the acquisition time sequence; Or alternatively, the first and second heat exchangers may be, The step of constructing the time sequence multidimensional point cloud comprises the following steps: collecting multi-frame point clouds on the surface of a deformed measured object in real time; the step of constructing the time sequence multidimensional point cloud comprises the following steps: And constructing a time sequence multidimensional point cloud in real time based on the multi-frame point cloud acquired in real time.
3. The method of time-series multidimensional point cloud deformation identification as recited in claim 1, wherein prior to the step of meshing the time-series multidimensional point cloud, the method comprises: preprocessing the time-series multidimensional point cloud, wherein the preprocessing comprises at least one of removing background data, removing noise data and homogenizing.
4. The method for identifying time-series multidimensional point cloud deformation according to claim 1, wherein the step of performing point-by-point deformation tracking on the time-series multidimensional point cloud after meshing comprises the steps of: extracting key points and multi-dimensional features of the key points from the time sequence multi-dimensional point cloud after grid connection, wherein the multi-dimensional features comprise rotation scaling invariant features and/or texture features; performing initial point pair matching on the key points based on the multi-dimensional features; optimizing the initial point pair matching result through a correlation model and an iterative optimization method to establish a lattice point pair matching relationship, wherein the iterative optimization method comprises a Gaussian-Newton iterative method, and the correlation model is used for judging whether the point pair matching is accurate or not.
5. The method for identifying time-series multidimensional point cloud deformations according to claim 4, wherein said step of establishing a point-to-point matching relationship of said time-series multidimensional point cloud based on deformation tracking results comprises: And establishing a point pair matching relationship of the time sequence multidimensional point cloud according to the grid point pair matching relationship.
6. A time-series multidimensional point cloud deformation identification system, characterized in that the time-series multidimensional point cloud deformation identification system comprises: The acquisition module is used for constructing a time sequence multidimensional point cloud based on the multi-frame point cloud of the surface of the measured object which is deformed; The grid module is used for grid the time sequence multidimensional point cloud; the deformation tracking module is used for carrying out point-by-point deformation tracking on the time sequence multidimensional point cloud after the grid connection; the matching module is used for establishing a point-to-point matching relation of the time sequence multidimensional point cloud based on the deformation tracking result; The point pair matching relationship is a matching relationship between a previous frame of multi-dimensional point cloud and a later frame of multi-dimensional point cloud of the time sequence multi-dimensional point cloud; the construction module is used for constructing a local deformation descriptor and a global deformation descriptor based on the point-to-point matching relationship so as to identify the deformation process of the surface of the measured object; The construction module is further used for acquiring discrete space-time coordinate information, discrete displacement vectors, discrete strain vectors and discrete deformation rate vectors based on the point pair matching relationship, and constructing a local deformation descriptor based on the discrete space-time coordinate information, the discrete displacement vectors, the discrete strain vectors and the discrete deformation rate vectors through the following formula: ; Wherein, LDD represents a local deformation descriptor, a represents discrete space-time coordinate information, delta represents discrete displacement vector, e represents discrete strain vector, v represents discrete deformation rate vector, and T represents transpose symbol; The construction module is further used for interpolating the discrete space-time coordinate information, the discrete displacement vector, the discrete strain vector and the discrete deformation rate vector to obtain a continuous displacement field, a continuous strain field and a continuous speed field of a preset time point and a preset space position; the displacement field is a function of the displacement state of the surface of the measured object, the strain field is a function of the strain state of the surface of the measured object, and the speed field is a function of the deformation rate of the surface of the measured object; based on the continuous displacement field, continuous strain field and continuous velocity field, a global deformation descriptor is constructed by the following formula: ; Wherein the said The global deformation descriptor is characterized in that, In order to be a displacement field, In order to be able to act as a strain field, For the velocity field, x, y, z represent three-dimensional coordinate information of x axis, y axis and z axis of a point of a multi-dimensional point cloud of a certain frame, T represents time information of a point of a multi-dimensional point cloud of a certain frame, and T represents transposed symbols.
7. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the time-sequential multidimensional point cloud deformation identification method of any of claims 1-5 when the computer program is executed by the processor.
8. A computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the time-sequential multidimensional point cloud deformation identification method of any of claims 1-5.

Description

Time sequence multidimensional point cloud deformation identification method, system, electronic equipment and storage medium Technical Field The present disclosure relates to the field of point cloud deformation recognition technologies, and in particular, to a method, a system, an electronic device, and a storage medium for recognizing time-sequence multidimensional point cloud deformation. Background Along with the rapid development of sensor technology and an observation platform, point clouds are a common way for representing three-dimensional information of the surface of an object, and the deformation identification technology of the surface of the object is gradually developed from the deformation of identification mark points of the surface of the object to full-field deformation identification based on the point clouds of the surface of the object. Full-field deformation identification methods based on object surface point clouds can be generally divided into two categories, point cloud-based models and surface model-based models. The former mainly includes a C2C (Cloud-to-Cloud) and M3C2 (Multiscale Model-to-Model Cloud) method, and the latter mainly includes a C2M (Cloud-to-Mesh) and M2M (Mesh-to-Mesh, model-to-Model) method, the difference between the former is that the deformation is recognized by directly comparing the point clouds, and the latter is based on surface modeling of the point clouds. However, at present, deformation between point clouds is identified based on geometric information of the point clouds whether the point clouds are based on a point cloud model or a surface model, the deformation identification method is widely used for identifying deformation of a large scene, and because the deformation identification method is used for identifying deformation of the surface of an object based on calculation principles of nearest neighbors or adjacent points, the deformation identified by the two methods is generally smaller than actual deformation of the surface of the object, the identification precision is lower, and the deformation identification method is not suitable for fine deformation identification of a small scene. Disclosure of Invention The invention aims to overcome the defect of low recognition precision of small scene refinement full-field deformation in the prior art, and provides a time sequence multidimensional point cloud deformation recognition method, a system, electronic equipment and a storage medium. The invention solves the technical problems by the following technical scheme: the invention provides a time sequence multidimensional point cloud deformation identification method, which comprises the following steps: constructing a time sequence multidimensional point cloud based on the multi-frame point cloud of the surface of the measured object subjected to deformation; gridding the time sequence multidimensional point cloud; carrying out point-by-point deformation tracking on the time sequence multidimensional point cloud after grid meshing; Establishing a point-to-point matching relationship of the time sequence multidimensional point cloud based on a deformation tracking result; The point pair matching relationship is a matching relationship between a previous frame of multi-dimensional point cloud and a later frame of multi-dimensional point cloud of the time sequence multi-dimensional point cloud; and constructing a local deformation descriptor and a global deformation descriptor based on the point pair matching relationship so as to identify the deformation process of the surface of the measured object. Preferably, the step of constructing the time sequence multidimensional point cloud includes: collecting multi-frame point clouds on the surface of the deformed measured object according to a preset time interval; the step of constructing the time sequence multidimensional point cloud comprises the following steps: constructing a time sequence multidimensional point cloud based on the acquired multi-frame point cloud according to the acquisition time sequence; Or alternatively, the first and second heat exchangers may be, The step of constructing the time sequence multidimensional point cloud comprises the following steps: collecting multi-frame point clouds on the surface of a deformed measured object in real time; the step of constructing the time sequence multidimensional point cloud comprises the following steps: And constructing a time sequence multidimensional point cloud in real time based on the multi-frame point cloud acquired in real time. Preferably, before the step of meshing the time-series multidimensional point cloud, the step of meshing the time-series multidimensional point cloud includes: preprocessing the time-series multidimensional point cloud, wherein the preprocessing comprises at least one of removing background data, removing noise data and homogenizing. Preferably, the step of performing point-by-point deformation tracking on