DE102004016829A1 - Determining the colour of opaque and transparent material, comprises taking one or more pictures using a camera, and analysing the colour - Google Patents

Abstract

A process for determining the colour of opaque and transparent material, especially plastic granules or powder, comprises taking one or more pictures of the material and then analysing the colour. In order to determine the colour of the opaque material, it is indirectly illuminated with diffuse light using an Ulbricht sphere (14). A sensor is arranged at an angle of up to 8 degrees to the axis perpendicular to the measurement table (15). The sensor consists of a camera, and illumination is by several groups of LED's of different colours.

Description

field of use

The The present invention relates to a method and an apparatus to accurately determine the color of both opaque as well transparent bulk material, in particular of plastic granules or powdery bulk material.

The Determination of the color of bulk material is mainly used in the context of production processes or quality controls needed. An important field of application of the invention is the production of Plastics in which a base material of polymers or resins a number of additives are included, in particular Color pigments. The molten homogeneous mass is extruded and subsequently granulated. The color of the plastic is generally from the client exactly given and may vary only within small tolerances. To the desired To ensure color during compounding, you have to measure it. The measurement can, in principle, be done in different places within the Manufacturing process done: in the liquid phase of the plastic, on the extruded strand, on the granules or on measuring platelets, which are made from the granules. It is desirable to have a measurement on the granules. If this measurement succeeds, the time-consuming production can of measuring plates eliminated and the correction values for controlling the color mixture are faster to disposal.

For an accurate Color measurement is by methods known in the art a color homogeneous, flat test specimen needed. Bulk goods, in particular plastic granules or powder, however, has a non-planar and color inhomogeneous surface on. To the desired To ensure color, therefore, are used in the production of plastics as well-known standard method from the granules in an injection molding machine shim shaped, which are then measured using a conventional spectrometer. The primary The measurement result of the spectrometer is the reflectance spectrum of the sample. This will generally be the color perception of the Derived people's relevant parameters, usually the standardized color values L * a * b * (CIE). This procedure has the disadvantage that from taking the sample (the granules) to to provide the measurement results (on Prüfflättchen) a relatively long Time goes by. While but this time must the continuously operating production plant continues to run and possibly produces Committee. The problem is the dead time determined by the measurement process in the regulation of color mixing. It can only be solved if one find a method that manages without measuring plate, so that one the color of the plastic already at an earlier stage of the production process can determine.

To solve the problem with the dead time, various ways have been proposed. In DE 195 07 750 A1 discloses a method by which one can measure the color of the plastic with the aid of a spectrometer directly after the extruder on the cylindrical strands. This is not possible under normal conditions, because a plane sample of a few millimeters in diameter is required for measurement with a spectrometer. In DE 195 07 750 A1 The plastic strands are run in a water bath, so that the cylindrical sample has a uniform color from the perspective of the spectrometer. In US 00552628A Also, a method of measuring the color of the cylindrical extrudates is described. Here, however, a conventional color camera (RGB) is used as a sensor, which is trying first to compensate for the unwanted inhomogeneities in the appearance of the plastic strand and secondly to perform a mapping of the RGB values in L * a * b *.

All mentioned methods for measuring the color of plastics in the Production has not been successful in practice so far. A measurement the color of plastics on the granules has not been known.

Task of invention

task It is the object of the invention to provide an apparatus and a method to provide an accurate measurement of the color of both opaque as well as transparent bulk material, in particular of plastic granules or pulverulent bulk material (or at least of Color differences of the respective sample compared to a reference sample) despite the non-planar and e.g. through shadows or breakages colored inhomogeneous surface allows.

presentation the invention

The The object is achieved by the method of the main claim 1. Of the Claim 18 gives an apparatus for carrying out the method according to the Claim 1 to. Advantageous embodiments of the method and the Device are the subject of the dependent claims.

For a precise color measurement you need a defined illumination, a defined measurement geometry and a color homogeneous, even test ling. Accordingly, commercially available spectrometers are usually constructed as follows: the light source used is a lamp which emits a sufficiently high power over the entire spectrum of visible light (eg xenon lamp); the specimen is illuminated with diffused light (eg via an integrating sphere); The sensor is directed at an angle of preferably 8 ° degrees to the specimen and the specimen has a color-homogeneous, flat surface. The spectrometer measures the reflectance spectrum of the device under test, averaged over a measuring spot of eg 10 mm diameter.

The Problem with the color measurement at the bulk material lies in the fact that the surface the examinees neither is it even, nor is it homogeneous in color. Inhomogeneous is the color e.g. in plastic granules due to the different surface properties at the cut surfaces and the lateral surfaces the granules or as a result of white fracture. There are also color deviations in shadow and border areas. The problem solved by the invention is thus in overcoming the obstacles to a precise color measurement: the non-level and the color inhomogeneous surface the examinees. The present invention is based on the finding that based of one or more shots of the bulk material with a sufficiently high Resolution first the Pixels (in the following selected pixels can be determined) which belong to the color-homogeneous partial surfaces of the bulk material. Prefers will do this first a color image, in particular an RGB image, either directly by measurement obtained or calculated from multiple images and the determination the selected one Pixels based on an agglomeration analysis in one- or multi-dimensional Histogram of the color image performed. With an averaging of Readings of these selected Pixels can then be the color of the bulk material, especially in one to the human eyesight adjusted unit like for example, Lab, and / or the color deviation between the bulk material and a reference sample can be determined from the averaged measured values.

With of the invention realize an automated color determination, e.g. during the regular production with small time delay controls whether the products or intermediates such as plastic granules have the given color or not.

preferred embodiments The present invention will be described below with reference to FIG the accompanying drawings explained in detail. Show it:

1 a measurement structure of a device according to the invention for determining the color of bulk material;

2 a one-dimensional histogram of a color image;

3 Emission spectra of light emitting diodes (LEDs);

4 Construction of an imaging spectrometer.

1 shows the measuring arrangement used. It is the same geometry as commercially available spectrometers: the illumination is via an integrating sphere 14 indirectly with diffused light produced by lamps 12 and the sensor 11 is at an angle greater than or equal to 0 ° degrees with respect to the perpendicular to the measuring table 15 , preferably of 8 degrees, on the test piece 13 directed, which is on the measuring table 15 located. For measuring the color of transparent samples, in particular transparent plastic granules, the in 1 to replace incident light illumination shown by a corresponding transmitted light illumination.

Of the most important difference commercial Spectrometers is that of the specimens a high-resolution image is recorded, e.g. with the help of a color-capable line scan camera or a Area camera. The spatial resolution is while e.g. at 60μm per pixel. Then account for example, a granule of 3mm edge length about 2500 pixels; with 100 received grains one has about 250,000 to Granules belonging Pixels. In the image obtained, the cylindrical surfaces of the Granules due to the diffuse lighting largely independent of the inclination of the surface elements in the same color; the background, the cut surfaces, white break, shadow and border areas appear in different colors.

The color of the bulk material is determined according to the invention by evaluation of the obtained color image. For this, the one- or multi-dimensional histogram of the color values of the image is first determined. Preferably, an RGB color image is used, so that a three-dimensional histogram with the three dimensions of the color separations R, G and B results. 2 shows a histogram in one-dimensional form. In the color space, an agglomeration analysis is then carried out that essentially looks for two distinct agglomerations: the background cluster 22 and the cluster belonging to the color-homogeneous partial surfaces of the bulk material 24 , Further values result from shadows 21 and border areas 23 , As a result of the congestion analysis, all the images in the color image become selected points that belong to the color-homogeneous partial surfaces of the bulk material. By averaging the measured values over all selected pixels, mean values for R, G and B are determined. These averages represent the color of the granules. Because of the large number of measurements taken in the averaging, the averages obtained are very accurate and well reproducible. The combination of agglomeration and averaging over a large number of measurements is an integral part of the invention.

For the labeling of colors are usually the units adapted to human perception L * a * b * (CIE). An accurate determination of the Lab values is not even glowing bodies generally requires a measurement of the reflectance spectrum of the sample over the entire wavelength range of visible light. Depending on the application and requirements However, approximate solutions may suffice for the accuracy of the color measurement. Accordingly, the o.g. Principle of color measurement according to the invention in different ways realized. In the following three embodiments will be described. For all described realizations, that is not necessarily the absolute Color values must be determined in Lab. In practice, it usually suffices if the color deviation between a sample and a reference sample can be specified in Lab.

image capture with colorado camera

When using color-capable cameras as sensor in 1 Two types come into question: area cameras and line scan cameras. In both cases, the lighting is provided by a white light source with the widest possible spectrum. Line scan cameras are advantageous because of their high spatial resolution (eg 2000 colored pixels on the line) and because of the simple measurement geometry. When using line scan cameras, the material to be measured is moved on a carriage under the sensor and the image is recorded synchronously with the movement. When using surface-mounted cameras, the material to be measured rests during image acquisition. Both sensor types directly deliver a color image (in RGB) and the method described above for determining the color of the bulk material can be used directly. The Lab values are determined directly from the RGB values obtained for the color of the bulk material. The conversion is based on a calibration that can be done either globally (using samples that cover the entire color space) or locally (using samples that cover the color gamut in the vicinity of the color for the given bulk material) is expected).

advantage this version is the simple and inexpensive Realization. It is also advantageous that in a very short time Time a big one Measure the number of readings, e.g. 20 million RGB values per second. Disadvantage is that converting the RGB values into Lab values is not unique in principle and therefore always critical colors where the measurement gives unsatisfactory results. The reason for this lies in the poor spectral resolution of the measurements (only three integral values: R, G and B).

Image capture with gray-tone camera, Illumination by flashed LEDs

In this realization, the sensor is in 1 a gray-scale camera (area camera or line camera) and the lighting consists of several groups of LEDs of different colors; eg 8 groups with the spectra after 3 , The LEDs are selected so that their emission spectra cover the spectrum of visible light as completely as possible. The image is advantageously taken with a surface camera. When the measurement object is stationary, the camera takes one after the other as many images as there are groups of LEDs. For each of the recordings, a different group of LEDs will turn on in turn. Thus, in this example, one scene produces eight images, each image representing the reflectance of the sample in the spectral range of the associated group of LEDs. From these eight images, an RGB image is first calculated again, and according to the method described above, those pixels are determined which represent the color of the bulk material. These pixels are marked in the eight spectral excerpts and from the marked pixels are then calculated by averaging eight vector components for the determination of the color of the bulk material. The link for calculating the Lab values from the vector components is either a calibration (as in the first implementation) or an estimate for the course of the remission spectrum of the sample (see third implementation).

These Realization is a compromise between the above first and third implementation with regard to spatial resolution (number the pixels captured per unit of time) and spectral resolution.

image capture with an imaging spectrometer

In this realization, the sensor is in 1 an imaging spectrometer. From the user's point of view, an imaging spectrometer is a line scan camera with, for example, 680 or 1000 pixels per line. For each pixel on the surface of the test object, however, not only one measured value (as in the case of a gray-scale camera) or three measured values (as with a color-capable camera) is produced, but, for example, 120 measured values. Each measured value represents the energy portion of the measured light in a narrow spectral range. At 120 measured values, distributed over the spectrum of visible light between 400 nm and 700 nm, the spectral resolution is 2.5 nm per pixel. The functional principle of an imaging spectrometer is in 4 shown. The imaging spectrometer becomes a line of the scene 41 via a lens 42 on the entrance slot 43 a spectrograph 44 displayed. The spectrograph forms the entrance slot 43 on a flat sensor and in such a way that in the x direction 45 the location information is retained and in y-direction 46 a spectral decomposition of the incident of each line element light takes place. As the light source, as in the first embodiment, a lamp (or a combination of lamps) is used, which provides as uniform and temporally constant luminous flux (flat spectrum) over the entire range of visible light. The image acquisition takes place as in a line camera synchronous to the movement of the measuring table.

Analogous to proceed in the second realization is derived from the primary Spectral images first calculates an RGB image. From this (as described above) by Constraint analysis determines those pixels that represent the color of the bulk material. These pixels are marked in the spectral images and off The marked spectra are calculated using a mean spectrum. With in the middle spectrum are the characteristic values for the color of the bulk material in Lab uniquely determined (see definition of L * a * b * color space in CIE 1964 and DIN 5033 DIN 6174).

advantage This third realization is the high spectral resolution of the Readings. This makes it possible for every Any color to clearly determine the key parameters in Lab.

From Disadvantage is the higher apparative effort for the imaging spectrometer and a relatively long time for data acquisition. At the same data rate of the sensor, much larger amounts of data must be available be recorded. Compared to the first realization, e.g. 120 measured values per surface element instead of three readings.

Claims (21)

Method for accurately determining the color of both opaque and transparent bulk material, in particular of plastic granules or powdered bulk material, the following steps having: - Admission one or more high-resolution images of the bulk material, which has a non-planar and color-inhomogeneous surface; and - Detection the color of the bulk material by evaluating the measured values of the high resolution image (s). A method according to claim 1, characterized in that for the accurate determination of the color of opaque bulk material, the illumination for the recording of the high-resolution images of the bulk material via an integrating sphere ( 14 ) takes place indirectly with diffused light in incident light. A method according to claim 1, characterized in that for the accurate determination of the color of transparent bulk material, the illumination for the recording of the high-resolution images of the bulk material via an integrating sphere ( 14 ) in transmitted light. Method according to one of claims 1 to 3, characterized in that the sensor at an angle greater than or equal to 0 ° degrees relative to the perpendicular to the measuring table ( 15 ), in particular of 8 ° degrees, is directed to the bulk material to be measured. Method according to one of claims 1 to 4, characterized the evaluation of the measured values has the following steps: - Determination the pixels belonging to the color-homogeneous partial surfaces of the bulk material (selected Pixels); and - averaging the one selected Associated pixels Measurements; and - Determination the color of the bulk material, in particular in a unit of measurement adapted to human vision such as Lab, and / or the color deviation between the bulk and a reference sample from the averaged readings. Method according to claim 5, characterized in that that a color image, in particular an RGB image, either directly through Measurement is obtained or is calculated from multiple images and the determination of the selected Pixels based on an agglomeration analysis in one- or multi-dimensional Histogram of the color image is performed. Method according to one of claims 1 to 6, characterized in that as a sensor ( 11 ) for the high-resolution image of the bulk material, a color-capable camera and for lighting a white light source with the widest possible spectrum are used. Method according to claim 7, characterized in that that as color-able Camera an area camera is used and the bulk material resting during the recording. A method according to claim 7, characterized gekenn records that as a color-capable camera, a line scan camera is used and the bulk material is moved when shooting under the sensor and the recording takes place synchronously to the movement of the bulk material. Method according to one of claims 7 to 9, characterized that the measured color values, in particular RGB values, of the selected pixels averaged and from these averaged color values the color of the bulk material, in particular the Lab values, based on a calibration, the can be calculated either globally or locally. Method according to one of claims 1 to 6, characterized in that as a sensor ( 11 ) are used for the recording of high-resolution images of the bulk material, a gray-scale camera and lighting several groups of light emitting diodes (LEDs) of different colors. Method according to claim 11, characterized in that that the LEDs are so selected are that their emission spectra are the spectrum of visible light preferably gapless cover. Method according to one of claims 11 or 12, characterized that a gray-tone area camera is used and the dormant bulk corresponding to the often Number of groups of LEDs is recorded, taking for each shot another LED group is used for lighting; and from these Shooting a color image, in particular an RGB image, for the agglomeration analysis is calculated. Method according to one of claims 11 to 13, characterized that over the selected one Pixels the measured values of the images with different LED illumination Averaging and from the averaged measured values the calculation the color of the bulk material, in particular the Lab values, based on a calibration or by an estimate for the Course of the remission spectrum of the bulk material takes place. Method according to one of claims 1 to 6, characterized in that as a sensor ( 11 ) for the recording of high-resolution images of the bulk material, an imaging spectrometer and for illuminating a white light source with flat as possible spectrum are used and from the obtained spectral images, a color image, in particular an RGB image, is calculated for the agglomeration. Method according to claim 15, characterized in that that the bulk material when recording under the sensor is moved and recording synchronously to move the bulk material he follows. Method according to one of claims 15 or 16, characterized that the measured spectral values of the selected pixels are averaged and from the averaged spectrum the color of the bulk material, in particular the Lab values, be determined. Device for accurately determining the color of both opaque and transparent bulk material, in particular of plastic granules or bulk powder, with: - one Device for taking one or more high-resolution images of the bulk material, which has a non-planar and color-inhomogeneous surface; and - one Device for determining the color of the bulk material by evaluating the Readings of the high resolution image (s). Device according to claim 18, characterized in that as sensor ( 11 ) are provided for the recording of the high-resolution image of the bulk material a color-capable camera and lighting a white light source with a flat spectrum as possible. Device according to claim 18, characterized in that as sensor ( 11 ) are provided for the recording of high-resolution images of the bulk material, a gray-scale camera and lighting several groups of light emitting diodes (LEDs) of different colors, the LEDs are preferably selected so that their emission spectra cover the spectrum of visible light as completely as possible. Device according to claim 18, characterized in that as sensor ( 11 ) are provided for the recording of high-resolution images of the bulk material an imaging spectrometer and for lighting a white light source with a flat spectrum as possible.