CN108280852B - Door and window point cloud shape detection method and system based on laser point cloud data - Google Patents
Door and window point cloud shape detection method and system based on laser point cloud data Download PDFInfo
- Publication number
- CN108280852B CN108280852B CN201810042774.9A CN201810042774A CN108280852B CN 108280852 B CN108280852 B CN 108280852B CN 201810042774 A CN201810042774 A CN 201810042774A CN 108280852 B CN108280852 B CN 108280852B
- Authority
- CN
- China
- Prior art keywords
- point cloud
- shape
- fitting
- door
- window
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20048—Transform domain processing
- G06T2207/20061—Hough transform
Abstract
The invention provides a door and window point cloud shape detection method and system based on laser point cloud data, which comprises the steps of preprocessing original point cloud data to obtain point cloud data with rough differences removed, and extracting edges; judging the shape of the point cloud according to the result of edge extraction, projecting the three-dimensional point cloud to two dimensions in the direction vertical to the plane of a door frame, dividing a grid, extracting an internal grid at the middle position, judging whether the number of points of the internal grid is greater than a preset corresponding threshold value, if so, judging the shape of the point cloud to be L-shaped, otherwise, judging the shape of the point cloud to be other shapes, dividing the edge extraction result of the point cloud into an upper part, a lower part, a left part, a right part and a lower part, and correspondingly judging the fitting of the upper part and fitting the edge of the point cloud; and calculating fitting errors of the fitting data and the point cloud data point by point, then calculating a decision coefficient, and extracting fitting parameters of which the decision coefficients meet corresponding threshold values. The invention has the characteristics of simplicity, effectiveness, high precision and easy realization, and has important market value.
Description
Technical Field
The invention belongs to the field of laser point cloud three-dimensional reconstruction, and particularly relates to a technical scheme for realizing shape detection and extraction of doors and windows by using point cloud data.
Background
The laser radar (Lidar) directly and quickly acquires high-precision dense dot matrix data by measuring the propagation time of laser pulses and combining positioning attitude data provided by an airborne positioning system, and is a novel active remote sensing technology. One of the remarkable features of the Lidar technology is that 3D coordinates of ground objects can be directly obtained, and the points are unevenly and discretely distributed in space and appear as "clouds", so that the method is called point cloud. The point cloud-based three-dimensional reconstruction technology is a research field of intersection of multiple subjects such as computer graphics, virtual reality, computer vision and the like. The method mainly researches how to restore the geometric information of the three-dimensional object recorded by the point cloud into a figure and an image, and displays the figure and the image through a computer, so that the object can be conveniently and quickly subjected to quantitative analysis, display, processing and the like. The point cloud-based three-dimensional reconstruction technology is widely applied to the following fields: computer graphics, such as movie and television special effects, three-dimensional animation, establishment of three-dimensional game models and the like; the medical field comprises medical repair, medical detection and simulation, medical bionics, plastic cosmetology, orthodontic simulation and evaluation and the like; the reverse engineering field comprises CAD model reconstruction, rapid modeling, finite element analysis and the like; the criminal investigation field comprises bullet marks, footprint, tool trace data acquisition, modeling and the like; in the field of industrial inspection, there are part inspection, product analysis, and the like.
The basic task of point cloud-based three-dimensional reconstruction is to perform curved surface reconstruction on point clouds acquired by measuring equipment to generate a three-dimensional solid model of an object. However, the measured point cloud data may be scattered or uniform, and the quality of the target grid is divided into good and bad, so that the three-dimensional reconstruction based on the point cloud has no uniform modeling process. But generally has the following basic flow: carrying out point cloud data acquisition on an object to be reconstructed by using measuring equipment; then, preprocessing the collected point cloud data, including denoising and simplification, to obtain a point cloud model of the object; then, gridding and optimizing the spatial point cloud to obtain a reconstructed grid model; and finally, carrying out texture processing on the grid to obtain a three-dimensional solid model of the object.
In a building scene, the most critical requirement is the measurement of doors and windows. In door and window's measurement field, the current mode of mainly adopting traditional manual work to measure consumes manpower and materials and time, and especially to special-shaped door and window, the efficiency and the precision of traditional measuring method are very low. In the field of application of the existing laser radar, an algorithm flow aiming at the shape detection of the door and the window does not exist. The method applies the laser radar point cloud to the shape detection of doors and windows, mainly extracts the line characteristics of the point cloud, simplifies the modeling of daily life scenes, constructs a function model from actual scenes, and realizes the shape detection and extraction of point cloud data.
Disclosure of Invention
The invention realizes a technical scheme for detecting the shape of the point cloud of the door and window based on the laser point cloud data, and can realize the fitting of complex shapes and extract corresponding shape parameters.
The invention provides a door and window point cloud shape detection method based on laser point cloud data, which comprises the following steps:
step 3.1, projecting the three-dimensional point cloud to two dimensions in the direction vertical to the plane of the door frame, then dividing the two-dimensional point cloud image into grids, and extracting an internal grid at an intermediate position;
step 3.2, judging whether the number of points of the internal grid is larger than a preset corresponding threshold value or not, if so, judging the internal grid to be L-shaped, finishing the judgment, fitting 6 edges of the door and window by using 6 straight lines, and entering step 5; if not, the shape is determined to be other shapes, and the step 3.3 is carried out;
step 3.3, dividing the edge extraction result of the point cloud into an upper part, a lower part, a left part and a right part;
step 3.4, fitting a straight line by using the upper part to obtain a correlation coefficient of the fitted straight line and the upper part, and when the fitted correlation coefficient R is greater than a preset corresponding threshold value, considering that the shape of the door and window is rectangular, ending the judgment, and entering step 4; if not, the shape is determined to be other shape, and the step 3.5 is carried out;
step 3.5, removing the fitting upper part by using an arc to obtain a fitting correlation coefficient R1; fitting the upper part by using two straight lines to obtain a fitted correlation coefficient R2, if R1> is R2, considering that the upper part of the door and window is in a circular arc shape, otherwise, considering that the upper part of the door and window is in a triangular shape; step 4, fitting the left, right and lower parts of the edge of the point cloud by using straight lines;
and 5, error control, namely calculating fitting errors of the fitting data and the point cloud data point by point, then calculating a decision coefficient, and extracting fitting parameters of which the decision coefficient meets a corresponding threshold value.
In step 1, the isolated points, outliers, and burrs are removed.
And in step 2, fitting the edges by using a Candy algorithm, and then realizing the integrity extraction of the edges through endpoint detection, intersection tracking and closed edge tracking.
Furthermore, in step 4, the hough transform is used to fit the left, right and lower portions of the point cloud edges.
In step 5, R is set2Is to determine the coefficients, as calculated below,
wherein, wiIs the weight, yiIs the value of the observed value and is,
is an estimated value of the amount of time,
is the average of the observed values, R2Has a normal value range of [0, 1 ]]Closer to 1 indicates a better fit of the model to the data.
The invention also provides a door and window point cloud shape detection system based on the laser point cloud data, which comprises the following modules:
the data preparation module is used for preprocessing the original point cloud data to obtain the point cloud data without gross errors;
the edge extraction module is used for extracting the edge of the point cloud data with the gross error removed, which is obtained by the data preparation module;
a shape judging module for judging the shape of the point cloud according to the result of the edge extraction, comprising the following units,
the first unit is used for projecting the three-dimensional point cloud to two dimensions in the direction vertical to the plane of the door frame, then dividing the two-dimensional point cloud image into grids and extracting the internal grids at the middle positions;
the second unit is used for judging whether the number of points of the internal grid is larger than a preset corresponding threshold value or not, if so, the internal grid is in an L shape, the judgment is finished, 6 edges of the door and window are fitted by 6 straight lines, and the error control module is commanded to work; if not, the shape is considered to be other shapes, and the third unit is instructed to work;
the third unit is used for dividing the edge extraction result of the point cloud into an upper part, a lower part, a left part and a right part;
the fourth unit is used for fitting the upper part to obtain a straight line and obtaining the correlation coefficient of the fitted straight line and the upper part, and when the fitted correlation coefficient R is larger than a preset corresponding threshold value, the shape of the door and window is considered to be rectangular, the judgment is finished, and the fitting module is instructed to work; if not, the shape is considered to be other shape, and the fifth unit is instructed to work;
a fifth unit, for fitting the upper part with the circular arc to obtain a fitted correlation coefficient R1; fitting the upper part by using two straight lines to obtain a fitted correlation coefficient R2, if R1> is R2, considering that the upper part of the door and window is in a circular arc shape, otherwise, considering that the upper part of the door and window is in a triangular shape;
a fitting module for fitting the left, right and lower portions of the edge of the point cloud with a straight line;
and the error control module is used for calculating the fitting error of the fitting data and the point cloud data point by point, then calculating a decision coefficient, and extracting the fitting parameters of which the decision coefficient meets the corresponding threshold value.
In addition, in the data preparation module, the isolated points, the outliers and the burr points are removed.
And in the edge extraction module, edges are fitted by using a Candy algorithm, and then the integrity extraction of the edges is realized through endpoint detection, intersection tracking and closed edge tracking.
Furthermore, in the fitting module, hough transform is used to fit the left, right and lower parts of the edges of the point cloud.
In the error control module, R is set2Is to determine the coefficients, as calculated below,
wherein, wiIs the weight, yiIs the value of the observed value and is,
is an estimated value of the amount of time,
is the average of the observed values, R2Has a normal value range of [0, 1 ]]Closer to 1 indicates a better fit of the model to the data.
The invention provides a technical scheme for detecting and extracting the shape of a door and a window based on laser point cloud data, which can construct a function model of a complex scene by only operating a small-size and light-weight handheld device without traditional manual measuring equipment, can detect the shape of the door and the window of the scene and extract parameters, and provides a new way for further improving the surveying efficiency and reducing the loss of manpower and material resources. The method has the characteristics of simplicity, effectiveness, high precision and easiness in implementation, and has important market value.
Drawings
FIG. 1 is an overall flow chart of an embodiment of the present invention.
Fig. 2 is a schematic view showing a shape result of a door window according to an embodiment of the present invention, wherein fig. 2(a) is a schematic view of a rectangular door window, fig. 2(b) is a schematic view of a rectangular door window with a lower portion and a triangular door window with an upper portion, fig. 2(c) is a schematic view of a rectangular door window with an upper portion and an arc with a lower portion, and fig. 2(d) is a schematic view of an L-shaped door window.
FIG. 3 is a flowchart of step 3 according to an embodiment of the present invention.
FIG. 4 is a schematic diagram of mesh partitioning and internal meshes according to an embodiment of the present invention.
FIG. 5 is a schematic view of upper, lower, left and right portions of an embodiment of the present invention.
FIG. 6 is a schematic view of an arc fitting upper portion of an embodiment of the present invention.
FIG. 7 is a partially schematic view of a two-segment line fit according to an embodiment of the present invention.
Detailed Description
The technical scheme of the invention can adopt a computer software mode to support the automatic operation process. The technical scheme of the invention is explained in detail in the following by combining the drawings and the embodiment.
Fig. 1 is a flow chart of a shape detection and extraction technique based on laser point cloud data, and the method of the present invention is described in further detail below with respect to each step in the flow of an embodiment.
(1) And (3) data preparation, namely preprocessing the original point cloud data to obtain the point cloud data with the gross errors removed.
In an embodiment, the orphan, outlier and burr points are removed: for one point Pg in the point cloud, if the distance between the point Pg and the nearest point is far larger than the average point distance of the point cloud, the Pg point is called a solitary point; for one point Pl in the point cloud, if only k adjacent points can be found when the adjacent points are searched according to a certain preset distance threshold, and the distances between the (k + 1) th nearest point and the Pl and the k preceding nearest points are far greater than the distance threshold, the Pl and the k nearest points are called outliers, the difference between the outliers and the outliers is that the outliers are clustered, and a single outlier is the outlier; for a point Ps in the point cloud, if the point cloud surface of the point cloud where the point Ps is away from the point is not isolated, but the smoothness of the local surface where the point cloud is located is influenced by the existence of the point Ps, the point Ps is called a bur point (non-smooth point). In specific implementation, the isolated points are detected by using a weighted average distance leaving division method, and the outliers are detected by using a grid detection method.
(2) And (3) edge extraction, which comprises the step of carrying out edge extraction on the point cloud data obtained in the step (1) after gross errors are removed, fitting edges by using a Candy algorithm, and then realizing the integrity extraction of the edges by algorithms such as endpoint detection, intersection tracking, closed edge tracking and the like.
In specific implementation, the existing methods for edge detection can be referred to, and the methods can be roughly classified into two types: based on the search and based on the zero crossing.
Search-based edge detection methods first compute the edge strength, usually expressed in terms of a first derivative, such as a gradient mode, and then compute to estimate the local direction of the edge, usually the direction of the gradient, and use this direction to find the maximum of the local gradient mode.
The zero crossing based approach finds the zero crossing points of the second derivative derived from the image to locate the edges. Usually with the laplace operator or the zero crossing of a non-linear differential equation.
Filtering is usually necessary as a pre-processing for edge detection, and gaussian filtering is usually used.
(3) And judging the shape, namely judging that the shape of the door and window point cloud sign is rectangular, circular, triangular or L-shaped according to the result of edge extraction.
The shape judgment is a very critical step, and only if the shape judgment is correct, the correct function model can be used for fitting the discrete point cloud data.
Common door frame (window frame) shapes are: rectangle, lower part rectangle upper portion circular arc, lower part rectangle upper portion triangle, L shape, refer to respectively FIG. 2: fig. 2(a) is a schematic view of a rectangular door window, fig. 2(b) is a schematic view of a rectangular door window on the lower portion, and a triangular door window on the upper portion, fig. 2(c) is a schematic view of a rectangular door window on the lower portion, and fig. 2(d) is a schematic view of an L-shaped door window. Therefore, the most efficient and reasonable distinguishing mode is designed, whether the shape of the upper part is L-shaped or not is judged firstly from the number of the lattice points of the middle part, then the upper part, the lower part, the left part and the right part are divided, and the shape of the upper part is judged through fitting.
Referring to fig. 3, the technical route for distinguishing the shape of the door frame is as follows:
(3.1) firstly, projecting the three-dimensional point cloud to two dimensions in the direction vertical to the plane of the door frame, then dividing the two-dimensional point cloud image into grids, and extracting the internal grids at the middle positions. For example, a grid of 30 by 30, the 24 by 24 region of the central region is considered to be the inner grid, with the central region in the middle as in fig. 4.
(3.2) judging whether the number of points of the internal grid is larger than a corresponding threshold value or not according to the grid dividing result, if so, judging the grid to be L-shaped, finishing the judgment, fitting 6 edges of the door and window by using 6 straight lines, and entering the step (5); if not, the shape is determined to be other shape, and the process proceeds to (3.3) to continue the determination. Referring to fig. 4, the number of points of the mesh inside the L-shaped door window is greater than the threshold value. Wherein, the setting of the threshold value can be obtained by the optimal result of the experiment: setting the maximum value and the minimum value of the threshold value, setting the change amount of the threshold value in each experiment, and then changing the threshold value to repeat the experiment to obtain the threshold value under the optimal experimental result. That is, the threshold value is set in consideration of the actual situation.
(3.3) then, the edge extraction result of the point cloud is divided into four parts, namely, an upper part, a lower part, a left part, a right part and a left part, as shown in fig. 5, wherein the parts 12, 23, 34 and 41 are the left part, the lower part, the right part and the upper part respectively.
(3.4) fitting a straight line by using the upper part (14 part) to obtain a correlation coefficient of the fitted straight line and the upper part, wherein the formula is as follows:
wherein, R (X, Y) is a correlation coefficient, Y is a value of a straight line point obtained by fitting, and X is a value of an original upper part point. Cov (X, Y) is the covariance of X and Y, Var [ X ] is the variance of X, and Var [ Y ] is the variance of Y.
When the fitted correlation coefficient R is larger than a preset threshold value, the shape of the door and window is considered to be a rectangle, the judgment is finished, and the step (4) is carried out; if not, the shape is determined to be another shape, and the next determination is made. Wherein, the setting of the threshold value in the concrete implementation can be obtained by the optimal result of the experiment.
(3.5) fitting the upper part with a circular arc, as shown in FIG. 6; and two straight lines are used to fit the upper part as shown in fig. 7. Namely, an arc is fitted by the upper part (14 part), and the fitting can be realized by a matlab function in specific implementation to obtain a correlation coefficient between the fitted arc and the upper part; namely, two straight lines (two side waistlines of a triangle, usually an isosceles triangle) are fitted by using the upper part (part 14), and the correlation coefficient between the fitted straight lines at the two ends and the upper part is obtained. And (4) respectively obtaining fitted correlation coefficients R1 and R2, wherein the calculation formula of the correlation coefficients is shown above. Compared with the prior art, if R1> is R2, the fitting degree of the circular arc is better, the upper part of the door and window is considered to be circular arc, and otherwise, the upper part of the door and window is triangular. And (4) entering.
The shape of the door and window is obtained based on the specific shape fitting method provided by the invention, the precision is judged in the size of the fitting relation coefficient, and if the relation coefficient is large, the fitting effect is good, and the shape is considered to be the shape. By adopting the analysis process provided by the invention, the common door and window shapes can be better distinguished through reasonably setting the threshold value. Based on the method, the method is a novel and efficient door and window point cloud shape classification detection method.
(4) Straight lines are used to fit the left, right and lower portions of the edges of the point cloud. During specific implementation, straight line detection can be realized by using Hough transform (hough transform) and other modes, the point cloud shape is fitted, shape parameters are obtained, and fitting results of the left and right sides of the door and window are obtained.
Except the step, 6 straight lines are used for fitting 6 edges of the door and window in the step (3.2), an upper part (14 part) is used for fitting a straight line in the step (3.4), two sections of straight lines are used for fitting the upper part in the step (3.5), and hough transformation can be adopted in the same way: a point in the original image coordinate system corresponds to a straight line in the parametric coordinate system, a straight line in the same parametric coordinate system corresponds to a point in the original coordinate system, and then all points of the straight line in the original coordinate system have the same slope and intercept, so they correspond to the same point in the parametric coordinate system. Thus, after each point in the original coordinate system is projected under the parameter coordinate system, whether the gathering point exists under the parameter coordinate system or not is seen, and the gathering point corresponds to a straight line in the original coordinate system.
Where k denotes a slope of a straight line, b denotes an intercept of the straight line, (x, y) is a coordinate in a rectangular coordinate system, and (ρ, θ) denotes a coordinate in a polar coordinate system.
(5) And error control, namely calculating the fitting error point by point for the fitting data and the point cloud data, and then calculating a decision coefficient. To ensure the accuracy of the method results, a decision coefficient R is calculated2。R2Is a decision coefficient that characterizes how well a fit is by the change in data.
Wherein, wiIs the weight, yiIs the value of the observed value and is,
is an estimated value of the amount of time,
is the average of the observations. From the above expression, R is known2Has a normal value range of [0, 1 ]]Closer to 1 indicates a better fit of the model to the data. In specific implementation, the method can be adopted when the decision coefficient is larger than the corresponding threshold value, the shape parameters and the type are output, otherwise, the fitting result is abandoned, and the accuracy is further ensured.
In specific implementation, the method provided by the invention can realize automatic operation flow based on software technology, and can also realize a corresponding system in a modularized mode. The embodiment provides a door and window point cloud shape detection system based on laser point cloud data, which comprises the following modules:
the data preparation module is used for preprocessing the original point cloud data to obtain the point cloud data without gross errors;
the edge extraction module is used for extracting the edge of the point cloud data with the gross error removed, which is obtained by the data preparation module;
a shape judging module for judging the shape of the point cloud according to the result of the edge extraction, comprising the following units,
the first unit is used for projecting the three-dimensional point cloud to two dimensions in the direction vertical to the plane of the door frame, then dividing the two-dimensional point cloud image into grids and extracting the internal grids at the middle positions;
the second unit is used for judging whether the number of points of the internal grid is larger than a preset corresponding threshold value or not, if so, the internal grid is in an L shape, the judgment is finished, 6 edges of the door and window are fitted by 6 straight lines, and the error control module is commanded to work; if not, the shape is considered to be other shapes, and the third unit is instructed to work;
the third unit is used for dividing the edge extraction result of the point cloud into an upper part, a lower part, a left part and a right part;
the fourth unit is used for fitting the upper part to obtain a straight line and obtaining the correlation coefficient of the fitted straight line and the upper part, and when the fitted correlation coefficient R is larger than a preset corresponding threshold value, the shape of the door and window is considered to be rectangular, the judgment is finished, and the fitting module is instructed to work; if not, the shape is considered to be other shape, and the fifth unit is instructed to work;
a fifth unit, for fitting the upper part with the circular arc to obtain a fitted correlation coefficient R1; fitting the upper part by using two straight lines to obtain a fitted correlation coefficient R2, if R1> is R2, considering that the upper part of the door and window is in a circular arc shape, otherwise, considering that the upper part of the door and window is in a triangular shape;
a fitting module for fitting the left, right and lower portions of the edge of the point cloud with a straight line;
and the error control module is used for calculating the fitting error of the fitting data and the point cloud data point by point, then calculating a decision coefficient, and extracting the fitting parameters of which the decision coefficient meets the corresponding threshold value.
The specific implementation of each module can refer to the corresponding step, and the detailed description of the invention is omitted.
The specific examples described herein are merely illustrative of the spirit of the invention. Various modifications or additions may be made or substituted in a similar manner to the specific embodiments described herein by those skilled in the art without departing from the spirit of the invention or exceeding the scope thereof as defined in the appended claims.
Claims (10)
1. A door and window point cloud shape detection method based on laser point cloud data comprises the following steps:
step 1, data preparation, which comprises the steps of preprocessing original point cloud data to obtain point cloud data with gross errors removed;
step 2, edge extraction, including edge extraction of the point cloud data obtained in the step 1 and subjected to gross error removal;
step 3, shape judgment, which comprises the steps of judging the shape of the point cloud according to the result of edge extraction,
step 3.1, projecting the three-dimensional point cloud to two dimensions in the direction vertical to the plane of the door frame, then dividing the two-dimensional point cloud image into grids, and extracting an internal grid at the middle position;
step 3.2, judging whether the number of points of the internal grid is larger than a preset corresponding threshold value or not, if so, judging the internal grid to be L-shaped, finishing the judgment, fitting 6 edges of the door and window by using 6 straight lines, and entering step 5; if not, the shape is determined to be other shapes, and the step 3.3 is carried out;
step 3.3, dividing the edge extraction result of the point cloud into an upper part, a lower part, a left part and a right part;
step 3.4, fitting a straight line by using the upper part to obtain a correlation coefficient of the fitted straight line and the upper part, and when the fitted correlation coefficient R is greater than a preset corresponding threshold value, considering that the shape of the door and window is rectangular, ending the judgment, and entering step 4; if not, the shape is determined to be other shape, and the step 3.5 is carried out;
step 3.5, removing the fitting upper part by using an arc to obtain a fitting correlation coefficient R1; fitting the upper part by using two straight lines to obtain a fitted correlation coefficient R2, if R1> is R2, considering that the upper part of the door and window is in a circular arc shape, otherwise, considering that the upper part of the door and window is in a triangular shape;
step 4, fitting the left, right and lower parts of the edge of the point cloud by using straight lines;
and 5, error control, namely calculating fitting errors of the fitting data and the point cloud data point by point, then calculating a decision coefficient, and extracting fitting parameters of which the decision coefficient meets a corresponding threshold value.
2. The method for detecting the shape of the laser point cloud based on the window and door point cloud data as claimed in claim 1, wherein: in step 1, the isolated points, outliers and burs are removed.
3. The method for detecting the shape of the laser point cloud based on the door and window point cloud data as claimed in claim 1 or 2, wherein: in step 2, fitting the edge by using a Candy algorithm, and then realizing the integrity extraction of the edge through endpoint detection, intersection tracking and closed edge tracking.
4. The method for detecting the shape of the laser point cloud based on the door and window point cloud data as claimed in claim 1 or 2, wherein: in step 4, fit the left, right and lower parts of the point cloud edge with hough transform.
5. The method for detecting the shape of the laser point cloud based on the door and window point cloud data as claimed in claim 1 or 2, wherein: in step 5, R is set2Is to determine the coefficients, as calculated below,
wherein, wiIs the weight, yiIs the value of the observed value and is,
is an estimated value of the amount of time,
is the average of the observed values, R2Has a normal value range of [0, 1 ]]Closer to 1 indicates a better fit of the model to the data.
6. A door and window point cloud shape detection system based on laser point cloud data comprises the following modules:
the data preparation module is used for preprocessing the original point cloud data to obtain the point cloud data without gross errors;
the edge extraction module is used for extracting the edge of the point cloud data with the gross error removed, which is obtained by the data preparation module;
a shape judging module for judging the shape of the point cloud according to the result of the edge extraction, comprising the following units,
the first unit is used for projecting the three-dimensional point cloud to two dimensions in the direction vertical to the plane of the door frame, then dividing the two-dimensional point cloud image into grids, and extracting an internal grid at the middle position;
the second unit is used for judging whether the number of points of the internal grid is larger than a preset corresponding threshold value or not, if so, the internal grid is in an L shape, the judgment is finished, 6 edges of the door and window are fitted by 6 straight lines, and the error control module is commanded to work; if not, the shape is considered to be other shapes, and the third unit is instructed to work;
the third unit is used for dividing the edge extraction result of the point cloud into an upper part, a lower part, a left part and a right part;
the fourth unit is used for fitting the upper part to obtain a straight line and obtaining the correlation coefficient of the fitted straight line and the upper part, and when the fitted correlation coefficient R is larger than a preset corresponding threshold value, the shape of the door and window is considered to be rectangular, the judgment is finished, and the fitting module is instructed to work; if not, the shape is considered to be other shape, and the fifth unit is instructed to work;
a fifth unit, for fitting the upper part with the circular arc to obtain a fitted correlation coefficient R1; fitting the upper part by using two straight lines to obtain a fitted correlation coefficient R2, if R1> is R2, considering that the upper part of the door and window is in a circular arc shape, otherwise, considering that the upper part of the door and window is in a triangular shape;
a fitting module for fitting the left, right and lower portions of the edge of the point cloud with a straight line;
and the error control module is used for calculating the fitting error of the fitting data and the point cloud data point by point, then calculating a decision coefficient, and extracting the fitting parameters of which the decision coefficient meets the corresponding threshold value.
7. The laser point cloud data-based door and window point cloud shape detection system of claim 6, wherein: and in the data preparation module, removing the isolated points, the outliers and the burr points.
8. The laser point cloud data-based door and window point cloud shape detection system of claim 6 or 7, wherein: in the edge extraction module, an edge is fitted by using a Candy algorithm, and then the integrity extraction of the edge is realized through endpoint detection, intersection tracking and closed edge tracking.
9. The laser point cloud data-based door and window point cloud shape detection system of claim 6 or 7, wherein: in the fitting module, hough transformation is used for fitting the left, right and lower parts of the edges of the point cloud.
10. The laser point cloud data-based door and window point cloud shape detection system of claim 6 or 7, wherein: in the error control module, set R2Is to determine the coefficients, as calculated below,
wherein, wiIs the weight, yiIs the value of the observed value and is,
is an estimated value of the amount of time,
is the average of the observed values, R2Has a normal value range of [0, 1 ]]Closer to 1 indicates a better fit of the model to the data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810042774.9A CN108280852B (en) | 2018-01-16 | 2018-01-16 | Door and window point cloud shape detection method and system based on laser point cloud data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810042774.9A CN108280852B (en) | 2018-01-16 | 2018-01-16 | Door and window point cloud shape detection method and system based on laser point cloud data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108280852A CN108280852A (en) | 2018-07-13 |
| CN108280852B true CN108280852B (en) | 2021-08-03 |
Family
ID=62803861
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810042774.9A Active CN108280852B (en) | 2018-01-16 | 2018-01-16 | Door and window point cloud shape detection method and system based on laser point cloud data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108280852B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110533677A (en) * | 2019-08-30 | 2019-12-03 | 广东博智林机器人有限公司 | A kind of edge extracting method and system based on 3D vision |
| CN110728753B (en) * | 2019-10-09 | 2022-04-15 | 湖南大学 | Target point cloud 3D bounding box fitting method based on linear fitting |
| CN111739018B (en) * | 2020-07-27 | 2020-12-11 | 中建科技集团有限公司 | Chromatic aberration detection method, device, terminal and medium based on point cloud technology |
| CN111986219B (en) * | 2020-08-10 | 2023-09-19 | 中国科学院光电技术研究所 | Matching method of three-dimensional point cloud and free-form surface model |
| CN112991368B (en) * | 2021-03-16 | 2023-08-15 | 追觅创新科技(苏州)有限公司 | Target object detection method and device, storage medium and electronic device |
| CN113129402B (en) * | 2021-04-19 | 2024-01-30 | 中国航发沈阳发动机研究所 | Cross section data cloud picture drawing method |
| CN115239648B (en) * | 2022-07-08 | 2023-06-02 | 厦门微图软件科技有限公司 | R-angle protrusion point cloud extraction method, device, equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103047943A (en) * | 2012-12-31 | 2013-04-17 | 吉林大学 | Method for detecting vehicle door outer plate shape and size based on single-projection encoding structured light |
| CN105421581A (en) * | 2015-11-29 | 2016-03-23 | 辽宁工程技术大学 | Three-dimension component library of building |
| CN108286951A (en) * | 2018-01-16 | 2018-07-17 | 常景测量科技（武汉）有限公司 | The handheld laser scanner measured for indoor doors or windows |
-
2018
- 2018-01-16 CN CN201810042774.9A patent/CN108280852B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103047943A (en) * | 2012-12-31 | 2013-04-17 | 吉林大学 | Method for detecting vehicle door outer plate shape and size based on single-projection encoding structured light |
| CN105421581A (en) * | 2015-11-29 | 2016-03-23 | 辽宁工程技术大学 | Three-dimension component library of building |
| CN108286951A (en) * | 2018-01-16 | 2018-07-17 | 常景测量科技（武汉）有限公司 | The handheld laser scanner measured for indoor doors or windows |
Non-Patent Citations (1)
| Title |
|---|
| 家居布局的层次化约束及其粒子群优化;陈光明 等;《计算机辅助设计与图形学学报》;20141231;第26卷(第10期);第1603-1613页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108280852A (en) | 2018-07-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108280852B (en) | Door and window point cloud shape detection method and system based on laser point cloud data | |
| CN109544677B (en) | Indoor scene main structure reconstruction method and system based on depth image key frame | |
| Kim et al. | Fully automated registration of 3D data to a 3D CAD model for project progress monitoring | |
| Sun et al. | Aerial 3D building detection and modeling from airborne LiDAR point clouds | |
| Xu et al. | Reconstruction of scaffolds from a photogrammetric point cloud of construction sites using a novel 3D local feature descriptor | |
| CN107767453B (en) | Building LIDAR point cloud reconstruction optimization method based on rule constraint | |
| CN111192284A (en) | Vehicle-mounted laser point cloud segmentation method and system | |
| CN110728753B (en) | Target point cloud 3D bounding box fitting method based on linear fitting | |
| CN103714541A (en) | Method for identifying and positioning building through mountain body contour area constraint | |
| Pound et al. | A patch-based approach to 3D plant shoot phenotyping | |
| CN104063702A (en) | Three-dimensional gait recognition based on shielding recovery and partial similarity matching | |
| CN110009671B (en) | Grid curved surface reconstruction system for scene understanding | |
| CN113378760A (en) | Training target detection model and method and device for detecting target | |
| CN104616349A (en) | Local curved surface change factor based scattered point cloud data compaction processing method | |
| Xiao et al. | Building segmentation and modeling from airborne LiDAR data | |
| Valero et al. | Detection, modeling, and classification of moldings for automated reverse engineering of buildings from 3D data | |
| CN116152697A (en) | Three-dimensional model measuring method and related device for concrete structure cracks | |
| CN103714420A (en) | Object three-dimensional reconstruction method and device | |
| CN104123569B (en) | Video person number information statistics method based on supervised learning | |
| CN116051980B (en) | Building identification method, system, electronic equipment and medium based on oblique photography | |
| Ward et al. | A model-based approach to recovering the structure of a plant from images | |
| CN113129348B (en) | Monocular vision-based three-dimensional reconstruction method for vehicle target in road scene | |
| JP6853150B2 (en) | Sorting equipment, methods and programs | |
| CN101976456B (en) | Characteristic point optimization method for three-dimensional reconstruction of video scene | |
| Tong et al. | A research on 3D reconstruction of building rooftop models from LiDAR data and orthophoto |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |