Algorithm for improving plate surface point cloud density calculated by 3D contour sensor

Abstract

The invention relates to the technical field of intelligent manufacturing, in particular to an algorithm for improving the plate surface point cloud density calculated by a 3D contour sensor, and thealgorithm comprises the steps: 1, collecting ultrahigh-density point cloud data of the same type of plates; 2, calculating a depth value mean value and a depth value standard deviation of the ultrahigh-density point cloud data, judging the data of which the depth value is located in an interval as conventional data, and judging the data of which the depth value is located outside the interval as extreme data; 3, training a set of IT2 fuzzy logic system through the conventional data, and training a set of T1 fuzzy logic system through the extreme data; and 4, scanning sparse point cloud data ofa to-be-detected plate, setting required interpolation coordinate points, and calculating depth estimation values of the interpolation coordinate points in combination with the IT2 fuzzy logic systemand the T1 fuzzy logic system. According to the method, the use cost of the 3D contour sensor can be greatly reduced, the applicability is wide, and the use effect is good.

Description

technical field [0001] The invention relates to the technical field of intelligent manufacturing, in particular to an algorithm for improving the point cloud density of a plate surface calculated by a 3D profile sensor. Background technique [0002] 3D profile sensors are also called 3D scanners and 3D cameras in the industry. Depending on the accuracy of application scenarios and project cost requirements, design schemes based on different principles including triangular distance measurement, structured light projection, and time-of-flight can be used. Among them, the principle of 3D profile sensor based on laser triangulation ranging is as follows: figure 1 Shown: the semiconductor laser focuses on the measured object through lens 1, and the reflected light of the measured object is collected through lens 2 and projected onto the CCD array. The signal processor calculates the position of the light spot on the array through trigonometric functions to obtain the distance of ...

AI Technical Summary

Problems solved by technology

For example, a piece of wood floor has a dark crack with a width of about 0.3mm along the y-axis, and the interval of the point cloud on the x-axis is 0.5mm. If the crack happens to be in the middle of two columns of points, it cannot be detected

On the other hand, the price of 3D profile sensors is high. For example, the price of domestic modules that meet the scanning width of about 200mm and the x-axis accuracy of about 0.3mm is more than 100,000 yuan. It is unacceptable. If the width of the plate is greater than 200mm, multiple sets of modules are required to be used together, which not only doubles the price, but also increases the cycle of mechanism design. Otherwise, the scanning width can only be increased by greatly sacrificing accuracy.

Similarly, the y-axis accuracy will also be challenged when the pipeline speed is high, which may further increase the manufacturing cost

Images