CN102289784A - Method for simplifying digital geometric image based on point cloud model - Google Patents

Abstract

The invention discloses a digital geometric image simplification method based on a point cloud model, comprising the following steps: step 1: selecting the geometric center of the point cloud model as the origin of polar coordinates; step 2: converting the Cartesian coordinate value of the point cloud model into polar coordinates Coordinate value; Step 3: Normalize and quantize the obtained polar coordinate value and map it to a gray value; Step 4: Delete some data points based on the simplification of the gray value; Step 5: Randomly sample and restore some deleted data points . It combines the two methods of geometric image and random sampling. First, the geometric image is simplified for the point cloud data. The simplified point cloud model can maintain the main geometric features in the original model. Then, the idea of random sampling is introduced, and the deleted point set Random sampling is performed, and the sampled points are added to the simplified model, so as to avoid the impact of holes on the subsequent surface reconstruction method, and realize a fast feature-preserving point cloud model simplification.

Description

A Method of Simplifying Digital Geometric Image Based on Point Cloud Model

technical field

The invention relates to the field of simplified processing of digital geometric images, in particular to a method for simplifying digital geometric images based on point cloud models.

Background technique

With the rapid development of digital technology and electronic technology, digital geometric media has received more and more attention from related industries. Large-scale real-time processing of digital geometric media, complex types of data mining and knowledge extraction, interactive graphical modeling and 3D digital reconstruction of dynamic objects pose new challenges to information technology. In recent years, with the significant development of laser scanning technology, the accuracy of data acquired by 3D laser scanning equipment is getting higher and higher, and the amount of data is also increasing. Here comes some questions. Not all data points are useful for subsequent modeling. Redundant data will not only reduce the calculation speed and increase memory overhead, but also may make it difficult to judge the feature part, thereby affecting the accuracy of subsequent modeling. In order to solve this outstanding problem, it is necessary to simplify the measured point cloud data under the premise of maintaining the geometric characteristics of the measured object.

The existing point cloud model simplification method based on geometric images is mainly implemented by a simple spherical polar coordinate plane indifference method, which is mainly divided into two steps. First, the Cartesian coordinates of the point cloud model are converted to polar coordinates, and then the spherical polar coordinates are resampled to convert them into grayscale images. The main steps include: (1) select the geometric center of the point cloud model as the origin of polar coordinates; (2) convert the Cartesian coordinate values of the point cloud model into polar coordinate values; (3) normalize and quantize the obtained polar coordinate values . However, this simplification method has the following technical disadvantages: First, by analyzing the above-mentioned geometric image production process, it is found that points with close spatial distances appear as adjacent pixels on the geometric image, so that the result can be further simplified; in addition, the above-mentioned The geometric image simplification process is a simplification based on the image resolution and a simplification based on the spatial distance, which does not preserve the characteristic, and it is easy to cause a partial vacuum phenomenon for more complex graphics.

Therefore, it is necessary to provide a digital geometric image simplification method based on point cloud model to make up for the above defects.

Contents of the invention

The purpose of the present invention is to provide a digital geometric image simplification method based on the point cloud model, which can further simplify the results based on the results obtained by the simplification method based on the geometric image resolution and restore the geometric image resolution based on the geometric image resolution to a certain extent. The simplification creates a vacuum region that preserves the characteristics of the simplification process.

In order to achieve the above object, the present invention provides a method for simplifying a digital geometric image based on a point cloud model, comprising the following steps: Step 1: select the geometric center of the point cloud model as the origin of polar coordinates; Step 2: convert the Cartesian point cloud model Coordinate values are converted to polar coordinate values; Step 3: Normalize and quantize the obtained polar coordinate values and map them to gray values; Step 4: Delete some data points based on the simplification of gray values; Step 5: Randomly sample and restore the part The data point to be removed.

Specifically, the step 4 further includes: step 41: calculating the value sumVal of the sum of the differences between a point and all pixels in its area; step 42: comparing sumVal with the preset threshold maxVal, and if sumVal is less than maxVal, Delete the point; if not established, save the point; Step 43: Repeat steps 41 and 42 until all points are traversed.

Wherein, the size of the threshold maxVal is set in the experiment according to the simplicity.

In order to enable the point cloud simplification method to obtain a better balance point in terms of simplification accuracy, simplicity, and speed, the present invention proposes a digital geometric image simplification based on a point cloud model that combines geometric image simplification and random sampling. method. It combines the two methods of geometric image and random sampling. First, the point cloud data of the geometric image is simplified. The simplified point cloud model can maintain the main geometric features in the original model. For the problem of holes in the complex model, by introducing The idea of random sampling is to randomly sample the deleted point set, and add the sampled points to the simplified model, so as to avoid the influence of holes on the subsequent surface reconstruction method, and realize a fast feature-preserving point cloud model simplification.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. Those skilled in the art can also obtain other drawings based on these drawings without creative work.

Fig. 1 is the flow chart of the digital geometric image simplification method based on point cloud model of the present invention;

FIG. 2 is a flowchart of a method for simplifying a digital geometric image based on a point cloud model according to an embodiment of the present invention.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

As mentioned above, the present invention provides a digital geometric image simplification method based on the point cloud model, which can further simplify the results obtained by the simplification method based on the geometric image resolution and restore the resolution based on the geometric image resolution to a certain extent. The vacuum area generated by the simplification of the rate makes the simplification process characteristic-preserving.

With reference to Fig. 1, the digital geometric image simplification method based on the point cloud model of the present invention comprises the following steps: Step S001: select the geometric center of the point cloud model as the polar coordinate origin; Step S002: convert the Cartesian coordinate value of the point cloud model is the polar coordinate value; step S003: normalize and quantize the obtained polar coordinate value and map it to a gray value; step S004: delete some data points based on the simplification of the gray value; step S005: randomly sample and restore some deleted data point.

Specifically, the step S004 further includes:

Step 41: Calculate the value sumVal of the sum of the differences between a point and all pixels within its domain;

Step 42: Compare sumVal with the preset threshold maxVal, if sumVal is less than maxVal, delete the point; if not, save the point;

Step 43: Repeat steps 41 and 42 until all points are traversed.

Wherein, the size of the threshold maxVal is set in the experiment according to the simplicity.

After coordinate conversion and normalization quantization, points with close spatial distances appear as adjacent pixels on the geometric image. The change in pixel gray value corresponding to the area with obvious curvature change is relatively large, while the change in pixel gray value corresponding to the area with gentle curvature change is relatively small, so the change of image gray value reflects the change information of the sampling surface curvature. Based on the above analysis, a simplification step based on the gray value can be further introduced. By comparing the sum of sumVal of the gray value difference between a certain pixel point and its surrounding pixels in the grayscale image, if the sumVal is less than the threshold value maxVal given by the user, the point is deleted. The size of the threshold maxVal can be set in experiments according to the simplicity.

The above-mentioned simplification method is only applicable to relatively simple models, and it is difficult to guarantee some graphic features of relatively complex models. Due to the nature of its multi-point mapping to one point, the simplification result is likely to produce a vacuum area. Therefore, the present invention also introduces a random The sampling method makes up for the above shortcomings to a certain extent. The random sampling method is to randomly select some data points from the deleted data points and add them to the simplified model. Practice has proved that this method is simple and fast, and it can effectively fill the local vacuum area.

With reference to Fig. 2, the specific implementation steps of the digital geometric image simplification method based on the point cloud model of the present invention are as follows:

θ_i＝cos^-1(z_i/r_i)得到

其中r_i是p_i到坐标原点的距离，

是p_i在xy平面上从x轴正向旋转得到的角度，

又叫方位角，

θ_i是p_i在zy平面上，从z轴正向旋转得到的角度，θ_i又叫极角，θ_i∈[0，π]。Step S001: Select the geometric center point of the point cloud model as the coordinate origin, and use the polar coordinate transformation formula for any p _i (xi _, y _i , zi ₎ in the point cloud model

θ _i ＝cos ^-1 (z _i /r _i ) gives

where r _i is the distance from p _i to the origin of coordinates,

is the angle obtained by the positive rotation of p _i from the x-axis on the xy plane,

Also called azimuth angle,

θ _i is the angle obtained from the positive rotation of the z-axis on the zy plane, θ _i is also called the polar angle, _{θ i} ∈ [0, π].

的范围：r_i∈[r_max，r_min]，

θ_i∈[θ_max，θ_min]。本发明实施例的简化方法将角度

对应到分辨率为M×N的图像上，r_i对应到图像的灰度值G上。对任意

对应横坐标θ_i对应纵坐标j＝(θ_i-θ_min)×N/(θ_max-θ_min)，r_i对应灰度值g＝(r_i-r_min)×(G-1)/(r_max-r_min)。M和N的大小根据点云模型的数据量及简度来定。G值取为256，其中256作为灰度图像的背景色。由灰度图像生成的过程可知，由

和θ_min+jΔθ＜θ≤θ_min+(j+1)Δθ组成的空间角中的点都将投影到[i，j]上。如果该区域中包含多个点，而一幅图像的一个像素只能保留一个灰度值，那么所有对应到[i，j]上的点只能保留一个，其余的点都丢失，在此会简化掉一部分空间距离较近的点云数据。Step S002: Perform normalization and quantization on the basis of the spherical polar coordinates of the point cloud generated in step S001, and convert Mapped into a grayscale image [i, j, g]. Find out separately

The range of: r _i ∈ [r _max ，r _min ],

θ _i ∈ [θ _max , θ _min ]. In the simplified method of the embodiment of the present invention, the angle

Corresponds to the image with a resolution of M×N, r _i corresponds to the gray value G of the image. to any

Corresponding abscissa θ _i corresponds to the vertical coordinate j=(θ _i -θ _min )×N/(θ _max -θ _min ), r _i corresponds to the gray value g=(r _i -r _min )×(G-1)/(r _max -r _min ). The size of M and N is determined according to the data volume and simplicity of the point cloud model. The G value is taken as 256, and 256 is used as the background color of the grayscale image. From the process of grayscale image generation, we can see that by

and θ _min + jΔθ<θ≤θ _min + (j+1)Δθ The points in the space angle will be projected onto [i, j]. If the area contains multiple points, and one pixel of an image can only retain one gray value, then all points corresponding to [i, j] can only retain one, and the rest of the points will be lost. Simplify part of the point cloud data with close spatial distance.

Step S003: From the analysis of the above steps, it can be seen that points with close spatial distances appear as adjacent pixel points on the geometric image. The change in pixel gray value corresponding to the area with obvious curvature change is relatively large, while the change in pixel gray value corresponding to the area with gentle curvature change is relatively small, so the change of image gray value reflects the change information of the sampling surface curvature. Based on the above analysis, a simplification step based on the gray value can be further introduced. By comparing the sum of the difference between a certain pixel in the grayscale image and the gray value of its surrounding pixels, if the sum is less than the threshold value given by the user, the point is deleted. The size of the threshold can be set in the experiment according to the simplicity.

Step S004: Step S001 describes a simplification of the resolution of a geometric image, which is a simplification method based on spatial distance, which does not preserve characteristics, and for complex models, a local vacuum will be caused by mapping one point with multiple points; and based on Although the simplification of the gray value of the geometric image is a feature-preserving simplification method, it will also cause a regional vacuum phenomenon due to the deletion of too much data in the area with a small curvature change. In this regard, the simplified method of the present invention introduces a random sampling process, and randomly selects some data points from the deleted data points to add to the simplified model. Practice has proved that the method is simple and fast, and effectively fills the local vacuum area.

Simplified compression of digital geometry data is the key to solving data storage and transmission efficiency. The present invention starts from the perspective of the user's subjective visual habits (paying more attention to the contour features of objects in motion, in the distance or with small actual size), and aims at different screens (such as high-definition television, PC, mobile phone, PDA) platform Efficiency display requirements, combined with the geometric error of the outline area in different projections to display different characteristics on the screen, starting from the perspective of display screen resolution accuracy, and aiming at the resolution characteristics of the target screen, research the digital geometric image simplification method based on screen perception measurement .

In order to enable the point cloud simplification method to obtain a better balance point in terms of simplification accuracy, simplicity, and speed, the present invention proposes a digital geometric image simplification based on a point cloud model that combines geometric image simplification and random sampling. method, which combines the two methods of geometric image and random sampling. First, the geometric image is simplified for the point cloud data. The simplified point cloud model can maintain the main geometric features in the original model, but holes will be generated for complex models. By introducing the idea of random sampling, the deleted point set is randomly sampled, and the sampled points are added to the simplified model, thereby avoiding the influence of holes on the subsequent surface reconstruction method, and realizing a fast feature-preserving point cloud model simplification .

The beneficial effects brought by the technical solution of the present invention:

After the geometric data simplification method, the number of points in the data set, the density threshold of the points in the data set, and the normal error caused by deleting a point are considered to simplify the point cloud data from the two aspects of simplicity and accuracy; at the same time, for massive data Simplification, speed is also the first issue. Therefore, the advantages and disadvantages of a point cloud data simplification method can be measured from the following three aspects:

(1) Accuracy: That is, the error between the surface fitted by the streamlined point cloud data and the real surface should be small. It is necessary to ensure that the error value is within an acceptable range and retain the characteristics of the original point cloud as much as possible.

(2) Simplicity: the percentage of the simplified point cloud relative to the original point cloud. The simplification should reduce the data as much as possible while ensuring a certain accuracy, but sometimes too few data points will also bring difficulties to subsequent modeling. Therefore, the appropriate degree of simplification should be selected according to actual needs.

(3) Speed: On the premise of ensuring accuracy and simplicity, faster speed and efficiency should be pursued. The present invention proposes a point cloud data simplification method combining geometric image simplification and random sampling, just to make the point cloud simplification method obtain a better balance point in the performance of the above three aspects.

The method for simplifying a digital geometric image based on a point cloud model provided by the embodiment of the present invention has been introduced in detail above. In this paper, specific examples are used to illustrate the principle and implementation of the present invention. The description of the above embodiment is only used To help understand the method of the present invention and its core idea; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation and scope of application. In summary, this specification The content should not be construed as a limitation of the invention.

Claims (3)

1. A digital geometric image simplification method based on point cloud model, is characterized in that, comprises the following steps: Step 1: Select the geometric center of the point cloud model as the polar coordinate origin; Step 2: Convert the Cartesian coordinate values of the point cloud model into polar coordinate values; Step 3: Normalize and quantize the obtained polar coordinate values and map them to gray values; Step 4: Based on the simplification of the gray value, delete some data points; Step 5: Randomly sample to restore some of the deleted data points. 2. The method according to claim 1, wherein said step 4 further comprises: Step 41: Calculate the value sumVal of the sum of the differences between a point and all pixels within its domain; Step 42: Compare sumVal with the preset threshold maxVal, if sumVal is less than maxVal, delete the point; if not, save the point; Step 43: Repeat steps 41 and 42 until all points are traversed. 3. The method according to claim 2, characterized in that, the size of the threshold maxVal is set in experiments according to the degree of simplicity.

Images