DE19638727A1 - Method for increasing the significance of the three-dimensional measurement of objects - Google Patents

Abstract

A method for increasing the significance of tridimensional measuring of objects trough optical recordings, projected samples and triangulation calculation, whereby a coded sample is projected on the object to prevent multiplicities during the evaluation of the image data is disclosed. This method is useful specially for medical diagnostics, therapy and documentation.

Description

The invention relates to a method for increasing the significance of the three-dimensional measurement of objects by optical recordings, projected pattern and triangulation calculation, in which a coded pattern on the object to avoid manifolds when evaluating the image data is projected.

For the purpose of measurement offers the use of methods based on optical Working from a variety of advantages. A measurement can be done quickly and done without contact. The prior art includes electronic image converters,  for example CCD arrays, the output signals immediately after one Digitization can be saved or evaluated.

Methods and devices for display and optical are known three-dimensional measurement of spatial surfaces. They are based on Triangulation methods in which point, line or any other patterns are projected onto the surface under consideration and the projected pattern from a different perspective than the projection angle with an optical system and an image converter. The well-known The geometry between the projection direction and the recording direction allows the three-dimensional calculation of surface support points.

If, for example, a uniform line pattern is projected onto an object with spatial extensions, a distorted line pattern results when the viewing direction is different from the projection direction, depending on the surface shape of the object. If the object is imaged onto a CCD array using suitable optics, the image signals are digitized and made available to a data processing unit, it is possible to identify the lines at various points in the image and with knowledge of the optical beam path and to calculate 3 D coordinates using triangulation, taking into account the geometric design of the projected line pattern.

With a simple spatial training of the object can due to relative distance between the lines shown incremental coordinates be calculated.

However, it is known that elevation levels are below in the surface to be measured certain geometric conditions can lead to relevant measurement errors. Directs the object with respect to the projection or the recording direction Undercuts, it is not immediately apparent whether it is neighboring lines in the image actually also around neighboring projected lines acts. If individual lines have not been displayed, it is also not Knows how many lines have been "swallowed". As undercuts are here To understand the formation of the three-dimensional contour, which as a result of  Viewing or projection direction hidden by parts of the object itself and are therefore not accessible for viewing or projection.

In European patent application 0 250 993 a system for the production of Ceramic inlays described, in which one for measuring the cavity of a tooth optical 3D measuring device is used. There the meaning of a "Optimal view" underlined for two reasons: - first - to the Align the recording unit with respect to the viewing and projection direction in such a way that the relevant surface segments of the object with a single exposure can be included and - secondly - so as not to be due to Undercuts to calculate the 3D coordinates incorrectly. In this Context is suggested, the lines based on the partially existing Contrast information in the calculation to "weight". However, this can be the case with a with regard to the contrast different formation of the surface structure of the Object only partially the probability of an incorrect calculation minimize.

It is part of the state of the art that consists of individual images with different Partial information obtained via three-dimensional surfaces based on the 3D coordinates of matching surface segments using Combine so-called "matching algorithms" to provide overall information. It it is obvious that there are manifolds or uncertainties regarding the correctness make a meaningful numerical calculation of the incoming coordinates impossible can, especially if they automatically and online during the Recording process should be made.

The object of the invention is therefore to increase the significance of the method three-dimensional measurement of objects of the type described above create with which each area of the projected pattern in the captured image for itself considered can be clearly identified without a continuity of the surface with regard to the viewing and projection direction in the manner described above and how must be assumed.  

The object is achieved in the method in that the for Purposes of three-dimensional measurement of objects with spatial Extents projected according to the triangulation method in such a way are coded that the assignment to the corresponding area of the projected pattern. From knowledge the beam path, the geometric design of the projected pattern and the Knowledge of the position of the corresponding sample section in the image acquisition can calculate absolute 3D coordinates using a simple triangulation calculation be determined.

If parts of the projected pattern now fall into a shadowed area, so can still partially identify the pattern and an exact 3D calculation be carried out. Although the three-dimensional measurement of the surface not possible with regard to the shaded area, but for the other areas correct due to the clear assignment of the respective sample lots.

It is conceivable that the formation of the encoded pattern against one simple stripe pattern, the number of predictable 3D information reduced. However, if you move the at low speed Acquisition unit opposite the object and takes a sequence different individual images, so by a suitable combination of the Information associated with single images condenses the 3D information as well to be gradually completed; be it due to undercuts to add missing information in single images; be it larger objects to represent as this due to the limitations of the field of view of the imaging unit is possible in a single shot.

By the method according to the invention can be described in the above Avoid manifolds in the triangulation calculation and the The significance of these calculations can be increased.  

For the purposes of this invention, “coded” are areas of the projected pattern that with regard to their geometric formation and / or extension and / or their other, for example colored design, are significantly distinguishable.

If a line pattern is used as a pattern for the projection, one obtains for the 3D calculation relevant information when the lines with an essential Component perpendicular to that between the direction of view and projection strung plane are striped.

When using a line pattern, the lines can be advantageous according to the invention e.g. B. can be encoded by varying the line width within the respective line. Binary, other discrete and analog encodings can be used here come.

For example, if - first - bright lines are projected onto the object, lie - second - the contrast data of the digitized image information in the memory of one Data processing unit flat, so are organized in rows and columns third - represent the column data, for example, the image information in Direction of the plane spanned between the direction of recording and projection the automated evaluation of the data can be carried out in a suitable algorithm for the purpose of calculating the absolute 3D coordinates in such a way that the Contrast data along an image column to match pattern sequence dark / light / dark can be checked. The corresponding statistical are for this and numerical methods such as cross correlation, Wiener filter etc. also for the Subpixel evaluation known.

If the column position of the pattern match is known, that is, you have A line identified can be followed by edge tracking of the course and training the line in the adjacent columns. With the status of The line itself can use technology-related methods of pattern recognition be examined with regard to their training and thus their coding.

As a result - firstly - the column position at which the line with its The focus of the sample was included and - secondly - the corresponding one  Line segment decoded; thus the line segment can be clearly corresponding sample in the undistorted and projected Assign patterns. This information can be used to calculate the triangulation 3D coordinate for a corresponding point on the observed surface of the Object can be determined.

Now identify the other lines in the manner described in the same column and this procedure is also carried out in other columns you get a number of coordinates, all of which are individual points on the Surface of the object. Such so-called "point clouds" with which the formation of surfaces is described are in the art known. The further data processing of point clouds, for example for formation of grid structures or of surface segments also belongs to the state of the Technology.

Especially with medical applications, the one with the above enables Procedure possible exact three-dimensional digital description the waiver spatial impressions (such as plaster casts) to document findings three-dimensional content or for computer-aided production of therapeutic agents (such as braces, dentures, implants).

It is advantageous according to the invention, along the line under consideration alternately arrange segments with single and double line width. If, for example, short and long line segments are also used, can from the variation of short and long line segments with simple or double the line width along a line analogous to the Morse code the lines be coded differently. If now lines in a shadowed area can still fall for each visible line segment that the contains complete coding, a calculation of absolute 3D coordinates without any information regarding other lines or Line segments are required.

It is advantageous according to the invention, in certain pattern sections, geometrically clear order, for example, to vary the black and white portion  the pattern ratio of others depending on the respective ratio black to white to be able to distinguish. According to the invention, the Segments with different black levels arranged linearly or concentrically will.

It is advantageous according to the invention in certain pattern sections Arrange breaks in the lines and different designs these interruptions to vary the sample parts.

Alternatively or in addition to a black and white training of the pattern can color coding of the pattern is particularly advantageous according to the invention be made. When using lines as a pattern, according to the invention each line advantageously has a different color, for example will. As a result, the information density per line is completely preserved; yet each line must be clearly identified, provided that the color content of the recorded Surface allows such a distinction.

"Color" in the sense of this invention is a selection of spectral lines and Spectral ranges of visible, infrared and / or ultraviolet light Understood. In the sense of this definition it is irrelevant whether the color by Generation of certain spectra or spectral ranges and possibly by additive color mixture formed or by absorption and optionally subtractive Color mixture is filtered from a wider range.

For the purposes of this invention, "black" becomes synonymous with essentially unlit and / or for essentially no retro-or Transmission component used due to absorption.

For the purposes of this invention, “white” becomes synonymous with having a broad spectral band illuminated and for with retroreflective or transmission component with wide Spectral band used.  

All of the embodiments of the method described in this invention can be particularly advantageous according to the invention in a variety of ways can be combined with each other.

Further advantages, features and possible uses of the present Invention emerge from the following description more preferred Exemplary embodiments in connection with the attached drawing.

It shows the accompanying drawings in the figures Fig. 1 and 2 embodiments of coded patterns for carrying out optical pick up for the purpose of three-dimensional measurement of objects using triangulation aufprojizierte patterns and according to the invention, partially in schematic form.

In the line pattern shown in Fig. 1, the line width along the respective line is carried out in a regularly repeated sequence alternately in single or double width (line width) such that in a defined line section the length ratio of the pieces with single and double line width varies from line to line is trained. In an analogous application of the expression customary in signal processing, in other words one can speak in this context of a different black / white duty cycle within the periodicity of the pattern.

The embodiment shown in FIG. 1 is characterized in addition to the coding of the sample parts in that, due to the special design of the pattern, a high density of the calculable support points can be achieved.

An arrangement of concentric pattern elements is shown in FIG. 2, the individual elements being coded in such a way that - firstly - the inner circle is blackened differently in 90 ° segments, - secondly - the outer concentric ring is blackened differently in 90 ° segments and - thirdly - the angular position of the outer concentric ring pattern with respect to the higher-level element arrangement is designed differently. The embodiment shown in Fig. 2 is characterized in addition to the coding of the sample parts in that

• a) through the three possibilities of different training of the high significance of the coding can be achieved for each sample element can and insofar
• b) a large number of elements can be safely distinguished, respectively
• c) redundant information can be displayed for test purposes,
• d) the concentric formation of the pattern elements for the automatic Pattern recognition is suitable and
• e) the inner circle in its different training for the exact Determination of the position in the distorted image of the projected pattern using Pattern recognition algorithms are suitable.

Claims (9)

1. A method for increasing the significance of the three-dimensional measurement of objects by optical recordings, projected patterns and triangulation calculation, characterized in that different areas of the projected pattern have a different design and are therefore coded. 2. The method according to claim 1, characterized in that the pattern is composed of at least two geometric elements by different training are coded. 3. The method according to claim 2, characterized in that it is in the differently formed geometric elements around elements with im essentially linear extension. 4. The method according to claim 2, characterized in that it is in the differently formed geometric elements around elements with im essentially concentric extension. 5. The method according to any one of claims 1 to 4, characterized in that the differently trained areas of the pattern differences have with regard to the formation of the line width and are thereby coded. 6. The method according to any one of claims 2 to 5, characterized in that the differently designed geometric elements of the pattern Have differences due to interruption and are therefore coded. 7. The method according to any one of claims 1 to 6, characterized in that the differently trained areas of the pattern differences with regard to the color design and / or the black / white content and are encoded by it.   8. The method according to any one of claims 1 or 7, characterized in that 3D coordinates of at least one point each from the image information from at least two areas of the surface of the recorded object calculated on the differently coded during recording Areas of the pattern have been projected. 9. Use of the method according to one of claims 1 to 8 for the medical diagnostics, therapy or documentation.