阅读说明：本技术 一种在曲板上动态投射视觉测量匹配用图案的方法 (Method for dynamically projecting patterns for visual measurement matching on curved plate ) 是由 胡勇 鲁恩顺 毛翔 赵章焰 于洋 于 2020-04-10 设计创作，主要内容包括：本发明公开了一种在曲板上动态投射视觉测量匹配用图案的方法,采用计算机预先生成规则图案,将图案传输到投影仪上,投影仪将图案投射到待测的曲板上,摄像机拍摄带有图案的图片,计算机根据特征提取算法自动提取图案中的特征点,用于特征点匹配,以便视觉测量能快速、准确还原出特征点处的三维坐标,从而实现对曲板上特征点的三维检测。能克服曲板测量时人工粘贴标志点的不足,且能根据需求实时变更不同图案,其操作简单、快速、经济。可应用于曲板的视觉测量。(The invention discloses a method for dynamically projecting patterns for matching visual measurement on a curved plate, which adopts a computer to generate regular patterns in advance, transmits the patterns to a projector, projects the patterns on the curved plate to be measured by the projector, shoots pictures with the patterns by a camera, and automatically extracts feature points in the patterns by the computer according to a feature extraction algorithm for matching the feature points so that the three-dimensional coordinates of the feature points can be quickly and accurately restored by the visual measurement, thereby realizing the three-dimensional detection of the feature points on the curved plate. The method can overcome the defect that the mark points are manually pasted during the measurement of the curved plate, and can change different patterns in real time according to requirements, and the method is simple, rapid and economical to operate. The method can be applied to the visual measurement of the curved plate.)

1. A method of dynamically projecting patterns for visual measurement matching on a curved plate, characterized by: the method comprises the following steps of projecting a regular pattern on a curved plate by using projection equipment, obtaining feature points in the projected pattern by using a feature point extraction algorithm, and restoring three-dimensional space coordinate information of the feature points by stereo matching of the feature points, wherein the method comprises the following specific steps:

firstly, preparing a projection pattern, and automatically generating a regular checkerboard matrix or a regular circular matrix through a program;

secondly, projecting a pattern;

thirdly, extracting characteristic points in the pattern;

the feature points in the projected pattern include intersection points of checkerboards, centers of circles or ellipses, edge points of curved plates. Wherein the intersection points of the checkerboard can be obtained by a corner point algorithm. The center of the circle or the ellipse can be firstly detected to form the boundary of the circle or the ellipse, and then the center point of the ellipse is obtained by fitting an ellipse equation.

And fourthly, acquiring three-dimensional space coordinate information of the characteristic points on the curved plate. Three-dimensional coordinates of the pattern feature points on the curved plate can be restored and calculated through the stereo matching of the corresponding feature points in the left image and the right image in the visual measurement.

Through the steps, the three-dimensional detection of the curved plate is completed.

2. A method of dynamically projecting patterns for visual measurement matching on curved plates according to claim 1, wherein: if the feature points on the measuring bent plate need to be increased, the effect can be realized by replacing a denser pattern.

Technical Field

The invention belongs to the field of mechanical manufacturing such as ship manufacturing, airplane manufacturing, automobile manufacturing and the like by using vision measurement. In particular to a method for dynamically projecting patterns for visual measurement matching on a curved plate.

Background

The curved plate with the free-form surface is widely applied to the industries of ships, aerospace, automobiles, high-speed rails, buildings and the like, and generally a straight plate is required to be formed through processing. The shape of the bent plate needs to be detected at any time in the machining process, and the shape is compared with the theoretical shape to judge whether the machining precision meets the requirement.

Common detection methods are classified into contact methods and non-contact methods. Computer (machine) -based vision measurement is the most widely used non-contact measurement method, and has the characteristics of high measurement speed, large information amount, high measurement precision and the like.

The machine vision measurement is based on parallax error, and three-dimensional information is acquired by a trigonometry principle. In machine vision measurement, in addition to calibration of a camera, stereo matching of feature points in an image is a key core technology for ensuring measurement accuracy.

At present, the industrial industry widely uses a handheld 3D laser scanner and a binocular vision laser structured light scanning system, the working principle of the system is the same, 2 cameras which are arranged on the left and the right acquire images of a single or a plurality of scanning laser lines, characteristic points on the laser lines are extracted through an algorithm to perform stereo matching, and three-dimensional coordinate information of the points on the laser lines is obtained through calculation.

The handheld 3D laser scanner is fast in single measurement speed and high in measurement accuracy, but the single measurement range is small and is generally within 100mm x 100 mm. When a large curved plate workpiece needs to be measured, a plurality of circular targets are usually manually pasted on the workpiece in advance, and the scanned small curved surfaces are automatically spliced to form large curved surface three-dimensional information. However, in the actual production of the curved plate, the manual target pasting is complicated, and the overall measuring speed is greatly reduced. And the workpiece needs to be processed for many times due to the factors of springback and the like. The target previously attached may be destroyed in the next process. When the measurement is carried out again, the target needs to be manually re-attached. After the workpiece is detected, the target needs to be manually removed. Therefore, the method for manually pasting the target is only suitable for laboratory research and is not suitable for industrial production.

The binocular vision laser structure light scanning system can scan the whole workpiece by adopting a long laser line, has a large measuring range, does not need to stick a target for curved surface splicing, but has a slow measuring speed (needs to scan one line by one line), and uses points on the scanning laser line for stereo matching.

In order to overcome the defects of the two measurement methods, a target is not required to be pasted, but characteristic points for visual measurement and three-dimensional matching are arranged on the curved plate, a new method for generating the characteristic points on the curved plate is needed, so that the matching precision is improved, and conditions are created for improving the subsequent measurement precision.

Disclosure of Invention

The invention aims to provide a method for dynamically projecting a pattern for matching visual measurement on a curved plate.

In order to achieve the purpose, the invention adopts the technical scheme that:

a method for dynamically projecting patterns for visual measurement matching on a curved plate comprises the steps of projecting regular patterns on the curved plate by using projection equipment, obtaining feature points in the projected patterns by using a feature point extraction algorithm, and restoring three-dimensional space coordinate information of the feature points by means of feature point stereo matching, wherein the method comprises the following specific steps:

firstly, preparing a projection pattern, and automatically generating a regular checkerboard matrix or a regular circular matrix through a program;

secondly, projecting a pattern;

thirdly, extracting characteristic points in the pattern;

the feature points in the projected pattern include intersection points of checkerboards, centers of circles or ellipses, edge points of curved plates. Wherein the intersection points of the checkerboard can be obtained by a corner point algorithm. The center of the circle or the ellipse can be firstly detected to form the boundary of the circle or the ellipse, and then the center point of the ellipse is obtained by fitting an ellipse equation.

And fourthly, acquiring three-dimensional space coordinate information of the characteristic points on the curved plate. Three-dimensional coordinates of the pattern feature points on the curved plate can be restored and calculated through the stereo matching of the corresponding feature points in the left image and the right image in the visual measurement.

Through the steps, the three-dimensional detection of the curved plate is completed.

If the feature points on the measuring bent plate need to be increased, the effect can be realized by replacing a denser pattern.

The invention has the following beneficial effects:

(1) patterns for matching of visual measurement are dynamically projected on the curved plate, characteristic points for matching can be added on the curved plate, and the characteristic points can achieve sub-pixel positioning through an algorithm, so that the precision of three-dimensional matching is improved, and the visual measurement precision is greatly improved.

(2) The shape, brightness, quantity and distribution of the projected patterns can be changed according to requirements, and the method is simple and convenient and can meet different requirements on site.

(3) When the projection equipment works, the projection pattern on the curved plate exists; when not working, the pattern disappears automatically. And manual removal is not needed.

Drawings

FIG. 1 is a flow chart of the present invention.

FIG. 2 is a schematic diagram of a checkerboard pattern projected on a curved plate according to the present invention.

Fig. 3 is a schematic diagram of left images of feature points in the pattern projected on a curved plate of the present invention in left and right cameras.

Fig. 4 is a schematic diagram of a right image of a feature point in the pattern projected on the curved plate of the present invention in left and right cameras.

Detailed Description

A method for dynamically projecting patterns for matching visual measurement on a curved plate comprises the steps of projecting regular patterns generated in advance by a computer on the curved plate through a projector, shooting two curved plate pictures with patterns by using a camera for visual measurement, obtaining feature points on the projected patterns of the curved plate by using a feature extraction algorithm, and performing stereo matching, so that three-dimensional coordinates of the feature points on the curved plate are obtained through reduction calculation, and the curved plate is quickly and accurately measured.

The specific implementation of the pattern method for dynamic projection of visual measurement matching on a curved plate is as follows:

fig. 1 shows a flow chart of a method for dynamically projecting a pattern for visual measurement matching on a curved plate.

The specific pattern projection steps are as follows:

step 1: automatically generating a regular pattern, such as a checkerboard, or a circular pattern, as shown in FIG. 2, using computer programming;

step 2: connecting a computer and a projector, and projecting the pattern onto a curved plate to be measured by using the projector;

and step 3: two pictures of the curved plate with patterns are taken by a vision measuring camera, wherein the picture is taken from the left and the right, as shown in figures 3 and 4;

and 4, step 4: extracting characteristic points in the pattern by adopting a visual algorithm;

and 5: and carrying out stereo matching on the characteristic points in the left image and the right image to obtain the three-dimensional coordinates of the characteristic points on the curved plate. And finishing the detection of the bent plate.

If the feature points on the measuring bent plate need to be increased, the effect can be realized by replacing a denser pattern.