WO2008087104A1

WO2008087104A1 - Method for determining the quality of a measuring point, in particular for determination of edge location, and optical precision measuring device

Info

Publication number: WO2008087104A1
Application number: PCT/EP2008/050314
Authority: WO
Inventors: Susanne TÖPFER; Karina Weissensee; Gerhard Linss; Uwe Nehse
Original assignee: Technische Universität Ilmenau; Mahr Okm Gmbh
Priority date: 2007-01-15
Filing date: 2008-01-12
Publication date: 2008-07-24
Also published as: DE102007003060A1; EP2115695A1

Abstract

The invention relates to a method and a device for determining the quality of a measuring point using image sensors, in particular for the detection of edges for optical length measurement. According to the invention, an optical image is recorded by means of an image sensor, and a measuring signal I with image data I (x,y) for the pixels of the image sensor is detected. Through the evaluation of the image data I (x,y), the location of a measuring point P (x,y) is determined in a sensor coordinate system. Parallel to this, the determination of the quality of the measuring signal I takes place independent of the determination of the location of the measuring point P(x,y), and also the determination of a quality identifier QK which represents the quality of the measuring signal I at the measuring point P (x,y). Finally, the location data and the quality identifier are consolidated in the sensor coordinate system to define a quality-evaluated measuring point P (x,y,QK). The method can be used for all coordinate measuring devices which have at least one image sensor as part of the measuring sensors.

Description

Method for determining the quality of a measuring point, in particular for edge location determination, and optical

precision measuring instrument

The invention relates to a method for determining the quality of a measuring point in edge location determination (edge detection), in particular in optical length measuring technology. Furthermore, the invention relates to a precision measuring device, in particular a coordinate measuring machine for the dimensional detection of workpieces.

Coordinate measuring devices are frequently used in optical length measuring technology. For the purposes of the present invention, a coordinate measuring machine is understood to mean a device with a plurality of drives whose respective position can be determined by means of scales, wherein the measuring sensor system can be moved relative to the workpiece in the 3 coordinates x, y and z. The measuring sensor system can consist of one sensor or of several different sensors (multi-sensors). The present method can be used on all coordinate measuring machines in which at least one image sensor is present as part of the measuring sensor system. In addition, the present method can be used on all measuring devices that have at least one image sensor, but no movement axes (image sensor as a stand-alone solution), for example in automation technology.

WO 02/084215 A1 discloses a method for optimizing the target variables during optical precision measurement. In this case, auxiliary parameters are first obtained from image information of a workpiece to be measured, from which subsequently control information for the influencing variables of these target variables be derived. The auxiliary parameters are determined such that they have a similar extreme of the functional dependence of the influencing variables.

DE 196 54 067 A1 describes a method for measuring edges on workpieces. For this purpose, a scanning point is moved over an edge to be measured in such a way that the edge is cut in a plurality of adjacent points. From the measured distance values, the intersections of the trajectory and the measured edge are determined by interpolation.

WO 2004/103181 A1 relates inter alia. a method for recording the movement of the internal organs of the body, wherein an image of the organ is generated and from this the movement parameters of the organ are determined.

When solving a measurement task with an optical coordinate measuring machine, the operator must select an edge location criterion before starting the measurement. The edge location criterion is a mathematical calculation rule that determines the associated edge location from an intensity transition in the image. Depending on the nature of the intensity transition in the image, different edge location criteria are suitable for generating the measuring point. In edge detection in optical length metrology, however, measuring points are also generated which, for example, were recorded in image areas with a low contrast, and therefore have a higher probing uncertainty. Measuring points that were recorded with very good contrast, have a low Antastunsicherheit. In known optical coordinate measuring devices, the quality of a measuring point is not recorded. Thus, all points in the subsequent calculation of the geometry elements are considered equivalent. WO 2005/050133 A1 relates to a method for calculating geometry elements and links between these measurement points measured on the basis of a coordinate measuring machine. The type of the geometric elements and the type of links should be found automatically by determining a match with mathematical models. Here, however, measurement points with a high uncertainty with the same priority are taken into account in the geometry element calculation as measuring points with a low uncertainty. Consequently, not all the information available by the image sensor is used to solve the measurement task. Depending on the calculation method, the calculated geometric element may have positional and shape errors that would not occur or only to a limited extent if the different quality of the measuring points were taken into account.

From the prior art, no methods are known in which the quality of the measuring points in the edge detection in the optical length measurement technique is determined.

The object of the present invention is therefore to provide a method and a device for determining the quality of a measuring point, in particular for edge detection for optical length measuring technology. A subtask consists of using not only the measuring points but also their respective quality for the calculation of geometric elements within such a measuring method in order to be able to calculate the geometric elements with a lower uncertainty.

According to the invention, the solution of this problem succeeds with the features of patent claim 1 and of subordinate claim 8, respectively. The invention is characterized in particular by the fact that not only the per se known determination of the position of a measuring point P (x, y) in a sensor coordinate system by evaluating a measuring signal with image data I (x, y) is carried out to determine the quality of a measuring point. In addition, independently of the position determination of the measuring point P (x, y), the determination of the quality of the measuring signal I and the determination of a quality index Q _k , which represents the quality of the measuring signal I at the measuring point P (x, y), take place according to the invention. Finally, the position data and the quality index are combined to form a "complete" quality-valued measuring point P (x, y, Q _κ ).

Preferred embodiments of the method according to the invention are specified in the subclaims.

The invention will be explained in more detail below with reference to preferred embodiments with reference to the drawing. Show it:

1 shows a basic structure of a known coordinate measuring machine, with which the inventive method can be performed;

2 shows a typical intensity profile I (x _s ) along a search beam s with registered quality indices, recorded on an edge by incident light illumination;

Fig. 3 calculation rules for quality indicators;

4 shows a flow chart of the method according to the invention for the determination of a measuring point and the evaluation of the quality of the measuring point. 1 shows by way of example the construction of a coordinate measuring apparatus (CMM) with which the method according to the invention can be carried out. The coordinate measuring machine comprises a plurality of drives with slides 5, 6 and 8, whose respective position can be determined by means of linear scales Ia, Ib and Ic. The carriages can be moved linearly along the coordinate axes x, y, z with the aid of the drives. In this case, the position of the respective carriages 5, 6 and 8 via the reading of an unillustrated reading heads, the associated

Scan scale, determined. Optionally, the CMM may have an axis of rotation 4, the position of which is determined with respect to an angle coordinate α via an angular measuring standard Id. The drives and their carriages 5, 6, 8 are fastened to a base frame 2 or, in the case of the z-slide 8, to a z-pillar 3. A workpiece 9 to be measured is either clamped on the axis of rotation 4 or fixed on a measuring table 10. The measuring sensor system is an image sensor 7, which additionally has lighting devices and a corresponding imaging system. The measuring sensor system can be moved relative to the workpiece 9 in the four coordinates x, y, z and α. The measuring sensor can consist of a single sensor or several different sensors (multi-sensor). For the control of the CMM and the evaluation of the read by the scales Ia, Ib, Ic and Id positions is a computer 11. This can be used simultaneously for the evaluation of the image information supplied by the image sensor 7.

It should be remembered that the method according to the invention can be applied to all coordinate measuring machines in which at least one image sensor is present as part of the measuring sensor system. Instead of a CMM, simpler tasks such as acquiring the geometry of flattening In mass production, a much simpler gauge with only two or one axis of motion or without any relative movement between the image sensor and the workpiece can be used. The method according to the invention therefore does not depend on the use and special structure of the coordinate measuring device.

The term "quality of a measuring point" includes the evaluation of the suitability of the optical intensity profile at an edge of the workpiece to be measured for the precise location of the edge An edge contains a change of the average intensity in the image stochastic errors. Systematic errors are not considered, as they can be detected and corrected by appropriate correction or calibration procedures.

The analysis of the intensity profile at an edge in the image serves as the basis for the determination of the quality of the measurement point, which is determined from this intensity profile. The evaluation of the quality of the measuring point with the help of various quality indicators is shown in FIG. The formulas for determining the quality indices are shown in the appended FIG. 3 and will be explained in more detail below.

The edge location is determined search-beam-based in the example explained in more detail below. Therefore, in all given formulas and equations (FIG. 3), the intensity I of the pixels always depends only on one spatial coordinate, namely the search beam coordinate x _s . However, the edge location can not be determined search-beam-based. The type of edge location determination in the sense of search-beam-based (evaluation of I (x _s )) or not search-beam-based (evaluation of I (x, y)) is irrelevant to the applicability of the present method.

FIG. 4 shows the basic sequence of the method. The method initially starts in a manner known per se with the determination of the measurement scene and the acquisition of an image, wherein an optical image is generated on the image sensor. The image data I (x, y) of each individual pixel recorded by the image sensor are transmitted as measurement signal I to a processing unit, of which the following

Image processing is carried out, performing the method according to the invention. The edge location determination initially takes place in a manner known per se, which need not be explained in more detail here. For this purpose, a measurement field (AOI) in the recorded image is expediently selected before executing the edge location determination. Within the measuring field, there are usually numerous search beams which intersect the edge to be found (ideally orthogonal). The signal curve recorded by the image sensor, i. the measurement signal I, for example, corresponds to the intensity profile along a search beam (FIG. 2).

An essential core idea of the invention is the completely independent evaluation of the signal curve I (x _s ) on the one hand for the determination of the edge location and on the other hand for the determination of the quality information. The edge location determination (left branch in FIG. 4) yields as a result in each case a measuring point P (x, y) in the sensor coordinate system (SCS). The measuring point P (x, y) denotes a point on the search beam which has been detected as an edge location. In parallel, the quality of the

Measurement signal along the search beam determines (right branch in Fig. 4). The quality of the signal curve is determined for each individual measuring point. Parallel processing is not necessarily to be understood as simultaneous, but in the sense of a determination of quality independent of the edge location determination. This corresponds to the separation of the measurement from the quality evaluation of the measurement signal. As a result of this processing branch, there is a quality index Q _K for each individual measuring point, which provides information about the quality of the measuring signal representing it. Finally, the quality information on the suitability of the signal curve for precise edge location determination is combined with the coordinates of the measured edge location and yields the quality-evaluated measurement point P (x, y, Q _κ ) in the local sensor coordinate system (SCS) of the image sensor. Q _{K is} the quality index which contains the quantitative information about the "quality of the measuring signal." Since this value is assigned directly to the respective measuring point, as illustrated in Fig. 4, this value is also referred to as "quality of the measuring point" ,

Based on the I ++ / DME specification, the value range of quality codes is between 0 and 100. The higher the quality index, the poorer the existing quality. Consequently, the quality of a measurement point is excellent if its quality index Q _{κ is} equal to zero.

When measuring a workpiece, determining the edge location or detecting the geometry, the explained method of determining the quality of individual measuring points can be used in the evaluation of each individual search beam. The quality indicators obtained can be evaluated statistically, so that ultimately a statement about the quality of the found / measured edge can be made.

In modified embodiments, the determination of the quality of the measurement points can also be limited to selected search beams, for example in order to reduce the computational effort. The methods for calculating the quality index Q _K for the quality of a measuring point are based on the known signal-theoretical relationships in the image recording with areal image sensors. In the calculation of the quality index Q _κ for the quantitative determination of the quality of the measurement signal or of the intensity profile, no default values are used with regard to the variable to be measured.

The quality index Q _κ for the quality of a measuring point can be composed of any number of individual quality indicators. In a preferred embodiment of the method according to the invention, five individual criteria (quality indicators) are used for the evaluation of the quality of a measuring point. The formulas for the calculation of the individual quality indices are shown in FIG. This quality score system does not contain redundancies, that is, one and the same feature of the intensity history is not scored twice.

The quality index Q _R is used to evaluate the noise of the intensity curve S ₁ . Here, only the areas to the right and left of the intensity transition of the edge in the image are evaluated. A possible extension of this quality index is to refer the noise not to the quantization levels Q _{n of} the image sensor but to the contrast K along the search beam. This allows a much sharper evaluation of the intensity profile. Furthermore, this extension corresponds to an evaluation of the signal-to-noise ratio instead of the pure evaluation of the noise as indicated in FIG.

The quality index for the edge increase Q _A is in the simplest case from the two intensity measurements above and determined below the mean level I _E. The larger the increase, the more accurate the edge location can be calculated.

The quality index for the form Q _{F of} the intensity transition provides information about the deviation from a jump signal. It is not redundant to Q _A. For example, the quality index Q _{A can be} excellent, but the quality index Q _F is comparatively poor. This is true if the intensity transition is very steep in the middle intensity range but extremely flat in the upper or lower intensity range; for example, for measurements on height-extended measuring objects. Depending on the applied edge location criterion, this can lead to a shift of the detected edge location.

The quality index edge width Q _B evaluates the width of the intensity transition. Compared to Q _B and Q _A , only Q _F provides the possibility to identify disturbed intensity transitions. These can occur, for example, in reflected light illumination and are frequently characterized by unwanted direct reflections on the surface of the measurement object.

The quality index Q _E is used for the assessment of uniqueness. This quality indicator is used to detect strong disturbances of the intensity profile. For measurements in incident light, the reflection pattern overlaps the surface of the measurement object with the diffraction effects at the edge. As a result, intensity traces can occur which intersect the mean level I _E more than once. The precise edge location tuning on such a waveform is very difficult. A possible extension of the measure Q _E is the consideration of the distance between the probable edge location in the interval [x _S i..Xs2] and the location of the second or third pass. As a general rule, the overall quality index for the quality of a measuring point for intensity profiles recorded in reflected light is, on average, significantly worse than for intensity traces recorded in transmitted light.

Accordingly, with the method according to the invention, in addition to the edge detection, the signal course in the structure transition area is also evaluated with reference to an ideal signal course according to the five different criteria explained above (quality indicators).

The ideal waveform corresponds to a jump input, which is imaged by the diffraction-limited optical imaging system on the image sensor and there depending on the

Bit depth of the image sensor is detected. The more the actual waveform resembles that of the ideal waveform, the lower the probing uncertainty of the measuring point or the higher its quality.

From the five individual evaluations, the overall quality of the measuring point is formed by weighted addition. The five individual criteria are e.g. Noise, edge width, edge slope, edge uniqueness and edge shape. Depending on the wishes of the user, the weighting of the individual criteria can be varied with each other or even criteria can be disregarded. This can be useful for computations that do not run on a PC, for computational reasons.

The method according to the invention has a number of advantages, since it uses previously unused information that is always present in such optical measurements for the improved solution of the measuring task. The parameters Among other things, the parameters of a measuring point in the I ++ DME specification include the parameter Q for the quality of the measuring point. In the case of optical measurements with image sensors, this parameter can be filled with a value by means of the present inventive method (each measuring point is provided with an indication of its quality), so that a higher accuracy can be achieved in the calculation of the geometry elements. The detection and elimination of outliers for the high-precision geometry measurement is greatly simplified and the computational outlay of the outlier filters is reduced.

LIST OF REFERENCE NUMBERS

Ia, Ib, Ic linear standards

Id angle dimension

2 base frame

3 z-pillar

4 rotation axis

5, 6, 8 sledges

7 image sensor

9 workpiece

10 measuring table

11 computers

x _s - search beam coordinate

I (x _s ) - Intensity of the search beam

Xp ₈ - Pixels center distance in the search beam n - Number of pixels in the search beam

Q _n - number of quantization levels of the image sensor Si - standard deviation of the intensity values

Si, int - Si of the intensity values in the interval Int

Iint ~ mean intensity in the interval Int δ (x _s ) - Dirac momentum b _k - measured edge width b _k , ideal ^~ ideal edge width

I _A - measured maximum edge increase t - edge number p - threshold value function

I _E - Threshold Ψ _RM ~ Cross correlation between edge signal and M

M - reference pattern

Δ - shift between M and I (x _s )

Claims

claims

A method for determining the quality of a measuring point in an image acquisition system, comprising the following steps: - taking an optical image with an image sensor and detecting a measurement signal I with image data I (x, y) for the pixels of the image sensor;

- Determining the position of a measuring point P (x, y) in a sensor coordinate system by evaluating the image data I ( ^χ , y);

Determining the quality of the measuring signal I independently of the position determination of the measuring point P (x, y) and determining a quality index Q _k which represents the quality of the measuring signal I at the measuring point P (x, y); - merging the position data and the quality index for defining a quality-valued measuring point P (x, y, Q _κ ) in the sensor coordinate system.

2. The method according to claim 1, characterized in that it is for execution within a search beam based

Edge location determination is adapted, wherein after taking the image, a search beam is set in a measuring field (AOI), which crosses the edge to be determined; and wherein an edge location determination is carried out for determining the position of the measuring point P (x, y) representing the edge location, with evaluation of the measurement signal obtained along the search beam.

3. The method according to claim 1 or 2, characterized in that the quality index Q _κ of several different

Quality individual criteria is composed, which evaluate the measurement signal with respect to its suitability for a precise image data evaluation or edge location determination.

4. The method according to claim 3, characterized in that the

Quality index Q _κ , which represents the quality of the measuring point, is determined from the weighted addition of several quality individual criteria.

5. The method according to any one of claims 1 to 4, characterized in that the quality index Q _κ (quality of the measuring point) based on a single or any combination of several quality individual criteria, such as noise, edge width, edge slope, edge uniqueness and edge shape is determined.

6. The method according to any one of claims 1 to 5, characterized in that a grayscale camera is used for image recording.

7. The method according to any one of claims 1 to 5, characterized in that a color camera is used for image recording.

8. Precision measuring device with at least one image sensor for detecting a workpiece, characterized in that means for carrying out a method according to one of claims 1 to 7 are present.

9. Precision measuring device according to claim 8, characterized in that it is designed as a coordinate measuring machine for the dimensional detection of workpieces.

10. Precision measuring device according to claim 8 or 9, characterized in that further sensors (multi-sensor) are provided for the detection of the workpiece.

11. Precision measuring device according to one of claims 8 to 10, characterized in that one or more axes of movement for the relative movement between sensors and workpiece are present.