EP2110187B1 - Method for identifying, classifying and sorting objects and materials and a recognition system for carrying out this method - Google Patents

Description

The invention relates to a method for identifying, classifying and sorting objects, objects and materials, wherein in one method step the spectral data, such as color, chemical properties, and spatial data, such as shape, size, position and / or structure of relative to Measurement unit of moving or moving objects, objects and materials in both planar dimensions with high spatial and full spectral resolution in real time and detected, and the objects, objects and materials thus detected are classified by this data in real time and then sorted, and wherein the a sorting criterion corresponding items are withdrawn from the process cycle by means of a sorting device, preferably by means of a controllable air flow, without interruption of the overall process. The invention further relates to an arrangement for carrying out this method.

In a number of industrial processes, methods for identifying or classifying objects, objects, and materials are an important tool for process control.

Thus, according to the DE-A1 19751862 an automated sorting method is provided in which the identification of the objects to be sorted is performed by means of spectral data analysis. Depending on the control of the spectrometer upstream scanning device in the identification process, the spectral data for a given amount of objects to be sorted are determined and forwarded for subsequent classification to a computer unit. This method is complicated in terms of apparatus technology, since such a detection system consists of three subsystems, namely an image analysis system, a scanning device and an NIR (near infrared) spectrometer.

Another known method for classifying objects, objects and materials is the measurement of spectral properties on a measurement object in one or more optical wavelength ranges and the evaluation of the spectral absorption, emission or reflection patterns thus obtained. However, these methods assume that the measuring spot is representative of the entire object to be measured. If this is not guaranteed, for example, if there is a possibility that the object is made by other materials is contaminated or covered, inhomogeneous, or if the measurement objects have a relevant surface structure to be detected, these methods are insufficient. For such cases, the use of systems is required with which the required spectral information can be detected spatially resolved.

It is known to use filter systems in this field, optionally using discrete filters when observing a few, selected wavelengths, or by using tunable optical filters to acquire a multispectral image in combination with a capture unit, usually a camera. Such a system using acoustically tunable optical filters is disclosed in US Pat US-A-5,216,484 described. The limitation when using filter systems with more than one observed wavelength is that the object to be measured may not move relative to the measuring device during the duration of the measurement. When a relative movement occurs, a series of images are obtained which are individual images of the measurement object at individual wavelengths but at different positions, whereby a complete image of a moving measurement object can not be obtained. This limitation can not meet the requirement of high resolution measurement in both planar dimensions and in real time without interrupting the process of such systems.

From the WO-A-01/67073 a method for quality control of fruit is known, using as the analytical method, the near infrared spectroscopy. However, this method is limited to subareas of the analysis object, so that only partial spectroscopic data are determined from the analysis object. The invention aims to remedy this situation.

It is therefore an object of the present invention to eliminate the disadvantages of the prior art methods for identifying and classifying objects, objects and materials, at the same time ensuring a simple procedure.

According to the invention, a method of the type mentioned is proposed, which is characterized in that a recognition system, consisting of a detection unit for the simultaneous determination of spatially resolved image and spectral data in real time, which is above the on a transport system, preferably a conveyor track, distributed to be sorted objects to be positioned, and one with this unit connected evaluation unit for evaluating the determined multispectral image data is used, and that as the detection unit, an imaging spectrograph is used in combination with a recording unit, whereby lines which are transverse to the direction of relative movement between the observed object, object or material and the detection unit, with Mapped by an appropriate, spectrally matched optics and fed through an entrance slit the imaging spectrograph, which with a wavelength-dispersive element, preferably a transmission grating, for the extraction of the spect alinformation from a light beam is equipped, so that at the exit side of the spectrograph, while preserving the spatial information of the imaged line, the light beams of the imaged lines are spectrally decomposed and imaged on the recording unit.

The invention further relates to an arrangement for carrying out the method according to the invention, comprising a detection system consisting of a detection unit, which is positioned above on a transport system, and an evaluation unit connected to this unit, wherein the detection unit used is an imaging spectrograph in combination with a recording unit is provided, and being provided as Transportanalge a continuous conveyor track and at least one means operatively connected to the detection unit for sorting out of the sorting criteria corresponding detected objects, objects or materials. Further advantageous embodiments of this arrangement are the subject of the dependent claims.

For carrying out the method according to the invention, the following ranges and / or subregions of the ranges and / or combinations of the ranges are used as spectral ranges: ultraviolet (UV, 200-380 nm), visible light (380-780 nm), near infrared ( NIR, 780 - 2500 nm) and mid-infrared (> 2.5 μm). The objects, objects and materials thus captured can be classified on the basis of this data in real time and then sorted automatically, for example.

An exemplary application for the method according to the invention is the sorting of articles in terms of material or type, Objects and materials, such as waste materials, but also the treatment of recycled materials. It is known to sort by hand any kind of items, such as waste and recycled material. It is based on criteria such as color, shape and composition of the items to be sorted. Although presorting by hand, as is done, for example, with smaller amounts of waste in households, is a relief for further waste disposal, yet manual sorting are consuming to perform. In order to simplify the sorting or to enable the sorting of health-endangering or harmful substances, the method according to the invention can be used here.

Another application example is the use of the method according to the invention for in-line quality control of products during the production process, for example of surface coatings. The use of the method according to the invention makes it possible to control the distribution, homogeneity and quality of a coating process in real time for this application. In addition, by applying image processing algorithms to the classification result, it is possible to automatically and in real time make statements, for example about the spatial and / or the size distribution of irregularities, which is an essential control and / or quality assurance parameter for the process control of essential importance for efficient and low-interference production , Compared to the state of the art, namely (i) off-line sampling tests, (ii) non-spatially resolved on-line sum testing, in the example sheet resistance and / or capacity, or (iii) the costly use of a plurality of apparatus, and equipment, distributed over the width of the conveyor individual measuring heads, the inventive method thus represents a significant advance.

With the method according to the invention, lines transverse to the direction of the relative movement of the objects to be classified can be imaged through a gap in the recognition system. In this case, this recognition system consists of an imaging spectrograph in combination with a recording unit. This system is characterized in that lines which extend transversely to the direction of relative movement between the object to be measured and detection unit, with Mapped by an appropriate, spectrally matched optics and fed via an entrance slit an imaging spectrograph, which is equipped with a wavelength dispersive element for spectral dissection of the lines, so that decomposed at the exit side of the spectrograph, the incoming light into the corresponding spectral regions and a receiving unit is shown.

The data determined as explained above are subsequently forwarded to the evaluation unit, which consists for example of an electronic data processing system. The data transmitted at high speed from the detection unit are first read in by means of a corresponding computer software program. For further processing, the computer software includes a number of mathematical algorithms for processing in real time the data streams determined using the recognition system (s).

Due to the relative movement, lines are sequentially recorded and assembled by means of computer software into a multidimensional information space. This consists of the two spatial coordinates in the longitudinal and transverse direction to the direction of movement and from a spectral dimension, which describes the spectral information as a spectral intensity as a function of the wavelength. By evaluating this data, it is possible to determine both the shape, size and position of the objects to be classified and, at the same time, color and / or other material properties, e.g. the chemical composition, to determine.

In general, this classification serves to associate the detected objects with groups having similar or comparable properties, such as in the case according to the invention, shape, size, color, structure and / or chemical constitution, as well as combinations of these and / or similar parameters. Although colors can also be captured with the help of conventional color cameras, these are limited to three color channels, so that a color classification can be made only with great technical effort or reduced precision. However, the method according to the invention enables a spectral decomposition which is typical for 100 to 200 wavelengths, so that a precise identification and classification can be achieved the items to be sorted is guaranteed.

By limiting the unit-detectable spectral range, it may be advantageous for specific applications to equip the overall system with two or more detection units, each consisting of spectrally tuned optics, imaging spectrograph, and camera. This makes it possible to broaden the information base of the classification and thus to increase the classification reliability. The detection units are preferably integrated into the overall system in such a way that they run synchronously with each other and detect the same observation field. The detection units can optionally be connected to a common or individual recognition and evaluation units.

• Fig. 1 eine mögliche Vorrichtung zur Durchführung des erfindungsgemäßen Verfahrens;
• Fig. 2 einen in dieser Vorrichtung vorgesehenen bildgebenden Spektrographen mit integrierter Aufnahmeeinheit;
• Fig. 3 ein in dieser Vorrichtung vorgesehenes Förderband; und
• Fig. 4 den Querschnitt dieses Förderbandes darstellen.

The invention will be described below by way of example with reference to FIG Fig. 1 to 4 explains

• Fig. 1 a possible device for carrying out the method according to the invention;
• Fig. 2 an imaging spectrograph having an integrated receiving unit provided in this device;
• Fig. 3 a conveyor belt provided in this apparatus; and
• Fig. 4 represent the cross section of this conveyor belt.

• Die zu sortierenden Gegenstände, Objekte und Materialien werden in Richtung des Pfeils 7 auf das Förderband 4 der Vorrichtung 1 gut verteilt aufgebracht. Nach einer kurzen Beschleunigungsstrecke gelangen diese Gegenstände mit einer Geschwindigkeit von typischerweise 1 bis 3 m/s zur Detektionseinheit 2, welche oberhalb des Förderbandes angeordnet ist. Der zu betrachtende Bereich des Förderbandes ist durch wenigstens eine optische Strahlungsquelle beleuchtet, welche sich bevorzugt durch eine über den Beobachtungsbereich räumlich und spektral homogene Intensitätsverteilung auszeichnet. In dieser Anordnung ist es möglich, Linien, welche quer zum Förderband verlaufen, durch eine Optik 8 der Detektionseinheit 2, wie sie in Fig. 2 dargestellt ist, zu erfassen. Das von dieser Beobachtungslinie ausgehende Licht gelangt über den Eintrittsspalt 9 - siehe Figur 2 - in einen abbildenden Spektrographen 10, der mit einem wellenlängendispersiven Element 11, bevorzugt einem Transmissionsgitter, ausgerüstet ist, so dass an der Austrittseite des Spektrographen unter Erhalt der räumlichen Information der abgebildeten Linie die Lichtstrahlen der abgebildeten Linien spektral zerlegt auf einer Aufnahmeeinheit 12 abgebildet werden. Diese Aufnahmeeinheit 12 ist vorteilhafterweise ein zweidimensionales Array, wie eine CCD-(charge-coupled device) oder CMOS- (complementary metal-oxide semiconductor) Kamera oder eine sonstige Kamera, welche für den jeweiligen Spektralbereich besonders geeignet ist.

An advantageous embodiment of the method according to the invention for use as a detection, classification and control system for the sorted material sorting will be described in the following with reference to the exemplary illustrations Fig. 1 to 4 explained in more detail:

• The objects, objects and materials to be sorted are applied well distributed in the direction of the arrow 7 on the conveyor belt 4 of the device 1. After a short acceleration section, these objects reach the detection unit 2 at a speed of typically 1 to 3 m / s, which is arranged above the conveyor belt. The area of the conveyor belt to be considered is illuminated by at least one optical radiation source, which preferably passes through the observation area distinguishes spatially and spectrally homogeneous intensity distribution. In this arrangement, it is possible, lines that are transverse to the conveyor belt, through an optical system 8 of the detection unit 2, as shown in FIG Fig. 2 is shown to capture. The light emanating from this observation line passes over the entrance slit 9 - see FIG. 2 in an imaging spectrograph 10, which is equipped with a wavelength-dispersive element 11, preferably a transmission grating, so that the light beams of the imaged lines are spectrally dissected on a receiving unit 12 on the exit side of the spectrograph while preserving the spatial information of the imaged line. This recording unit 12 is advantageously a two-dimensional array, such as a CCD (charge-coupled device) or CMOS (complementary metal-oxide semiconductor) camera or other camera, which is particularly suitable for the respective spectral range.

The data is transmitted to the evaluation unit 3 and further processed there. Two categories of data are simultaneously determined with the recognition system used according to the invention, namely on the one hand the spectral data for determining the color and / or the chemical properties and on the other hand the image data for determining the shape, position, size and / or structure of the objects to be classified. The shape of the objects is captured by capturing and evaluating the lines across the conveyor belt with a typical resolution of between 128 points and 1600 points. Due to the sequential recording of the lines, a two-dimensional image emerges from the aforementioned lines. From this data set, the geometric properties of the objects to be classified can be determined.

For typical, purely qualitative classification, the data processing includes a dark current correction, a referencing to a standard, a normalization of the spectra, a filter stage to reduce the noise, the calculation of the 1st or 2nd derivative as well as the actual classification according to a specific application and optimized algorithm. Further functions can be, for example, those for determining the center of gravity and / or the object or class edges and / or other object properties by means of image processing algorithms. This makes it possible to precise the objects to be separated recognize, classify in real time and in the further example to make a corresponding sorting.

In the case of waste separation, for example, for the recycling of plastic bottles, such as polyethylene terephthalate (PET) - bottles, not only the classification by material, components, but also by color and / or the nature of the screw cap of the bottle for further processing advantageous. Likewise, standard sorting criteria that are used in other classification methods, such as length, width and shape or contour can be extracted directly from the data and used for sorting. By combining the required data in the software evaluation module, for example by comparison with stored target specifications, the classification result can be adapted to the process-specific requirement. The data, the so-called output data of the evaluation module, is transmitted via a user-defined interface, another model of the computer software, to the corresponding control components, for example a higher-level process control system or, as in this exemplary application, to an automated sorting unit , about controllable exhaust nozzles, forwarded. In this application for material sorting, an automated sorting, for example by blowing out or by means of controlled flaps, can thus take place after the determination of the aforementioned data.

Particularly preferred is the sorting with the aid of a controllable nozzle row 5, in which air nozzles 14 are arranged such that the air streams generated thereby vortex the items auszusortierenden laterally into corresponding receptacle so that they can be deducted from the conveyor belt 4. The nozzle row 5 is as out FIGS. 3 and 4 can be seen, advantageously arranged below the strands of the strips 13 carrying the articles.

Another preferred method is the use of an additional cross conveyor belt 6, which is mounted at a distance of, for example, 30 cm across the transport conveyor belt 4. The articles to be discharged are in this embodiment swirled through the discharge nozzles on this conveyor belt and then transported away on the cross conveyor belt. This sorting process is based on the representation according to Fig. 3 explained in more detail. It's too see that the conveyor belt 4 in the longitudinal direction of a plurality of spaced apart, for example, about 50 strips 13 consists. The nozzle row 5 is mounted so that the individual nozzles 14 between the strips 13 of the conveyor belt come into effect. At these Ausblasstationen the objects to be sorted are guided over the mounted under the conveyor belt, controllable nozzle row 5, which is carried out by the special design of the conveyor belt in the form of strips 13 no interruption of the belt and thus no interruption in the flow. The shape of the nozzles 14 and the airfoil produced thereby are adjusted so that the items to be sorted out are swirled onto the cross conveyor 6 or into a side receptacle (not shown) according to their weight and airflow generated. By using the special conveyor belt 4, such as a belt band, and the attachment of the nozzle row 5 under this band, those objects that are not sorted out in this separation stage, continue to remain in the same position and location on the conveyor belt for transport to the next separation stage , As a result, several separation steps can be carried out depending on the requirement profile. As a result, the automated sorting of different objects is made possible so that a specific sorting by composition, such as according to different plastics, metals and the like, but also by environmental damage, sales value or calorific value is made possible.

In summary, it can be said that the method according to the invention for identifying, classifying and sorting objects can therefore be carried out in a simple manner, since geometric and spectral data, for example concerning color and / or chemical data, of a unit simultaneously and with high spatial and spectral resolution in real time Characteristics of the objects to be measured are recorded. The method according to the invention is particularly relevant in the case of measurement objects which are in motion relative to the recognition system for process engineering or other reasons. The method according to the invention can be carried out in a simple manner, in particular with the aid of the recognition system used according to the invention, consisting of a unit for the simultaneous determination of image and spectral data and a unit connected therewith for evaluating the determined image and spectral data.

Claims (8)

1. A process for identifying, classifying and sorting items, objects and materials, wherein in one process step the spectral data, such as colour, chemical properties and spatial data, such as shape, size, position and/or structure of items, objects and materials moved or moving relative to the measuring unit are determined and recorded in real time in both planar dimensions with high spatial and full spectral resolution and the items, objects and materials recorded in this manner are classified in real time using this data and then sorted and wherein the items matching a sorting criterion are removed from the process cycle by means of a sorting mechanism, preferably by means of a controllable air current, without the entire process being interrupted, characterized in that a recognition system comprising a detection unit (2) for the simultaneous determination of spatial resolved image and spectral data in real time, which is positioned above the items to be sorted which are distributed on a transport system, preferably a conveyor belt (4), and an evaluation unit (3) connected to this unit for evaluating the multispectral image data obtained is used and that an imaging spectrograph (10) is used as the detection unit in combination with a recording unit (12), whereby lines which run transversely to the direction of the relative movement between the observed item, object or material and the detection unit are imaged with the help of a suitable spectrally coordinated optical system (8) and supplied to the imaging spectrograph (10) via an entry slot (9), said spectrograph being equipped with a wavelength-dispersive element, preferably a transmission grid (11), for extracting spectral information from a light beam, so that at the outlet side of the spectrograph the light beams of the imaged lines are spectrally fragmented, while retaining the spatial information, and displayed on the recording unit (12).
2. A configuration for implementing the process according to claim 1, comprising an recognition system made up of a detection unit (2), which is positioned above a transport system, and an evaluation unit (3) connected to this unit, wherein an imaging spectrograph (10) is used as the detection unit in combination with a recording unit (12) and wherein a continuous conveyor belt (4) as the transport system and at least one mechanism (5) effectively connected to the detection unit for sorting the items, objects or materials identified as matching the sorting criteria are provided.
3. The configuration according to claim 2, characterized in that the sorting mechanism comprises a row of nozzles (5) and collecting vessels attached to the side.
4. The configuration according to claim 2, characterized in that the sorting mechanism comprises a series of nozzles (5) and a transverse conveyor belt (6) positioned transversely above the conveyor belt (4).
5. The configuration according to one of the claims 2 to 4, characterized in that the conveyor belt (4) is continuous throughout the entire system.
6. The configuration according to claim 5, characterized in that the continuous conveyor belt (4) is formed from a plurality of parallel strips (13) in a longitudinal direction and that the row of nozzles (5) is disposed beneath the offshoot of the strips (13) supporting the items.
7. The configuration according to one of the claims 2 to 6, characterized in that the items being sorted are removed from the material flow by being blown out obliquely to the side.
8. The configuration according to claim 7, characterized in that the items being sorted are spun by blowing out onto a transverse conveyor belt (6) located above the transport belt and thereby removed from the material flow.

Images