Point cloud simplifying and filtering method

Abstract

A point cloud simplifying and filtering method provided by the present invention comprises the steps of firstly establishing an angle distance coordinate system oxyz, then dividing cylinders, and simplifying the points in each minimum cylinder into a three dimensional angular distance point; and then utilizing the three dimensional angular distance points of the upper and lower and the right and left cylinders to filter the three dimensional angular distance point of the current cylinder; after the filtered three dimensional angular distance points in the angle distance coordinate system oxyz are transformed into the three dimensional angular distance points in a three dimensional position coordinate system o'x'y'z', furthering simplifying into the three dimensional distance points by the cylinder segmentation; and then in the three dimensional position coordinate system o'x'y'z', utilizing the three dimensional distance points of the upper and lower and the right and left cylinders to further filter the three dimensional distance point of the current cylinder. The method of the present invention is high in point cloud simplifying and filtering precision, fast in operation speed and small in memory occupation space.

Description

Technical field [0001] The invention relates to a point cloud simplified filtering method, which belongs to the technical field of image processing. Background technique [0002] The three-dimensional laser scanner uses a laser beam to continuously scan along the solid surface, and the diffuse reflection light spot formed on the measured solid surface is imaged on the sensor. These dense point data are vividly called "point clouds", which can truly describe the overall structure and morphological characteristics of the measured entity. [0003] In the point cloud acquisition process, the distance L between the scanning head and the object to be measured can be calculated according to the time difference between the launch and the reception of the laser beam. The scanning data of the pendulum mirror can be used to obtain the laser emission azimuth, namely the α angle and the β angle. The combination of distance and orientation can measure the three-dimensional coordinates of each ...

AI Technical Summary

Problems solved by technology

In engineering applications, point clouds are mostly processed based on embedded processors. Due to the low computing power of embedded processors, existing methods take a long time to run and are difficult to meet the needs of engineering applications.

In the point cloud obtained by practical engineering applications, the distance noise is relatively large, because of the error caused by the distance noise, there is no effective filtering method

Images