1. Field of the Invention
The present invention relates generally to systems for separating selected articles from a stream of articles, and more particularly, but not by way of limitation, to a system particularly suited for sorting recyclable materials such as different types of plastic containers and paper or cardboard products, including carrier board, from each other.
2. Description of the Prior Art
Environmental campaigns and recycling efforts in many areas have generated a substantial supply of recyclable waste paper and like materials. These materials need to be sorted before they can be recycled. For instance, plastic and glass articles need to be sorted from the stream itself and then further by plastic resin type, color, etc. Colored paper stock often needs to be separated from white stock, and cardboard and carrier board needs to be removed from newsprint. In addition, it is sometimes necessary or desirable to separate printed materials from blank materials. Further, separation processes such as screens designed to remove cardboard and plastic and metal containers from paper streams, often miss some of those materials, requiring additional separation steps. Unfortunately, sorting of waste paper and paperboard, etc. is still currently performed almost entirely by manual sorting. Manual sorting of such materials can be time consuming and expensive, which can render the use of recycled paper less economical than virgin paper material. This is even more apparent when so-called carrier board is present in the waste stream. So-called carrier board, commonly understood to be as the paperstock used in, e.g., cereal boxes, soda or beer can carriers, frozen food boxes, etc., must be sorted manually, as there is currently no effective automated method for doing do.
Numerous automated waste separation techniques are known. However, these techniques are generally designed for the recovery of metals, alloys, municipal waste, mixed recyclables and plastics. Paper (or, more generally, sheeted material) sorting presents unique problems that cannot be overcome by most prior art separation techniques. For instance, the relatively lightweight and flexible nature of paper presents unique problems when sorting is attempted. Indeed, these problems make it difficult to supply paper to a sorting sensor, especially not at a desirable feed rate (usually defined in terms of feet per minute (fpm), but sometimes also in terms of pieces or objects per minute (ppm) or tons per hour (tph)). Without higher speeds, automated sorting systems do not achieve efficiencies substantially greater than manual sorting. The problem is exacerbated where the waste stream includes paper and non-paper waste.
A number of different sorting systems have been proposed in the prior art for sorting various articles based upon the color of the articles or the characteristics of the reflected or transmitted electromagnetic radiation to which the article is exposed. Such systems have been utilized for sorting glass, plastic, paper, newsprint, fruit and other edible items, and the like. Similarly, a number of arrangements have been provided for carrying the articles through an inspection zone, and for exposing the articles to electromagnetic radiation and then collecting and analyzing the reflected and/or transmitted radiation.
For example, U.S. Pat. No. 4,131,540 to Husome et al. discloses a color sorting system wherein light is reflected off tomatoes and the reflected light is collected and analyzed as the tomatoes fly through an inspection zone.
U.S. Pat. No. 4,657,144 to Martin et al. discloses a system for removing foreign material from a stream of particulate matter such as tobacco as it cascades through an inspection zone.
U.S. Pat. No. 4,919,534 to Reed, discloses a system for determining the color of glass bottles, wherein the light energy is transmitted through the glass bottles.
U.S. Pat. No. 5,085,325 to Jones et al. discloses a system of a very common type wherein articles are examined as they are supported upon a moving conveyor belt.
U.S. Pat. No. 5,297,667 to Hoffman et al. discloses a system of utilizing two light sources and a camera to analyze articles as they fly through an inspection zone.
U.S. Pat. No. 5,314,072 to Frankel et al. discloses a system which analyzes the transmissive characteristics of articles which are exposed to x-ray fluorescence.
U.S. Pat. No. 5,318,172 to Kenny et al. discloses a system which distinguishes different types of plastic materials based upon their reflected electromagnetic radiation.
U.S. Pat. No. 5,333,739 to Stelte discloses another system which transmits light through articles, namely glass articles, and analyzes the transmitted light to determine color.
U.S. Pat. No. 5,443,164 to Walsh et al. discloses a plastic container sorting system which utilizes both transmitted electromagnetic energy and reflected electromagnetic energy to analyze and identify articles.
U.S. Pat. No. 5,675,416 to Campbell et al. discloses an apparatus which looks at the transmissive properties of articles to separate them based upon the material of the article.
U.S. Pat. No. 5,848,706 to Harris discloses a sorting apparatus which examines optical characteristics of the articles against a viewing background.
U.S. Pat. No. 5,966,217 to Roe et al. discloses a system for analyzing articles wherein reflected radiation is split into a plurality of streams which are then filtered and analyzed.
It has also been suggested to separate carrier board from a newspaper stream via a color-based identification system. However, this approach is not very effective or accurate since color is a secondary feature of these materials, not a fundamental characteristic.
In a relatively recent and unique approach, Doak et al., in U.S. Pat. No. 6,497,324, disclose a sorting system utilizing a multiplexer to allow a single analyzer unit to be used to analyze electromagnetic signals from each of a plurality of collector units. Although effective, the Doak et al. system requires the operation of complex and highly sensitive software and mechanical components, which can be difficult to maintain.
In addition, as noted above, another problem encountered by waste sorting systems is the identification and separation of carrier board and coated or waxed board material commonly used as, e.g., beverage cartons, cigarette cartons, etc. from other paper materials. More particularly, the separation of white or printed paper stock from an article stream can be accomplished by recently developed systems, leaving newsprint and carrier board in the article stream. Further separation to provide only newsprint in the stream, however, has proven problematic.
Thus, it is seen that although there have been many arrangements proposed for the examination of a stream of articles by analysis of reflected and/or transmitted electromagnetic radiation from the articles, there is a continuing need for improved systems, which may simplify the analytical mechanism and permit the identification of materials (such as carrier board) heretofore found difficult to process.
A system for sorting articles includes a feed conveyor for conveying the articles toward a first destination. A plurality of sources of narrow bandwidth electromagnetic radiation of differing frequencies are provided for shining electromagnetic energy on the articles in seriatim. The sources are preferably arrayed and actuated such that the individual beams of electromagnetic energy from the sources illuminate the same region of the article as it passes through the sensor region. This can be accomplished spatially or through timing, or both. Each of the sources advantageously has a beam spreader associated with it, for spreading the radiation beam across the width of the conveyor (though preferably not along the length of the conveyor, to avoid overlap with adjoining beams). Additionally, the individual sources may be made up of several sources (arranged perpendicular to the flow direction of the articles) with or without beam spreaders such that wide feed streams can be accommodated. A collector is provided for collecting energy reflected from the articles. A deflector is provided for deflecting selected articles toward an alternative destination. A control system is operably connected to the collector and the deflector for actuating the deflector in response to a sensed parameter (such as color) of the energy collected in the collector.
By providing a series of sources of electromagnetic radiation of narrow bandwidth (i.e., a bandwidth range of from about 5 nm to about 250 nm), the identification and separation of several classes of articles can be accomplished. It is well known that the amount of reflected radiation at specific frequencies varies for differing materials. In the visible range this variation determines the color of an object. In the near infrared range (i.e. from about 680 nm to 2000 nm) the amount of reflected radiation is determined by the molecular structure of the material, and therefore its composition.
Conventional separation systems for recyclable materials typically illuminate the articles with a steady state broadband radiation from a light source such as a halogen lamp. The reflected light is then measured at various frequencies utilizing a spectrometer type system (diffraction grating, etc.) or a system of detectors with individual frequency filter sets. This approach is costly due to the number of expensive optical and detector components required. An improved approach utilizing a multiplexer minimizes the number of detectors and filters required but introduces a mechanical system which limits reliability and throughput speed.
The improved system utilizes a series of narrow bandwidth sources which can be switched on and off very rapidly such that the amount of reflected radiation can be measured at a number of specific frequencies without the necessity of a broadband light source or a multiplexer. Further the shape of the narrow bandwidth source can be selected or shaped to optimize the resulting reflection intensity differences between differing materials and therefore the identification accuracy.
For instance, assuming several individual light sources, aligned in the direction of travel of the articles on the conveyor, each actuated sequentially, as an article travels along the conveyor, a pulse of radiation from each of the sources strikes each article sequentially and in substantially the same place. The reflected radiation is collected by multiple collectors. By analyzing the amount of radiation reflected by an article from each differing radiation frequency the article can be identified as for example, polyethylene terephthalate (PET) plastic, newspaper, brown carrier board, white paper, etc.
Referring to
For example in
Other methods beside ratiometric calculation can also be used to determine the type of material utilizing the amount of radiation reflected at differing frequencies. These methods include the use of neural net engines, spectrum comparison with predetermined spectrum stored in a look-up table, spectrum stored by training the system with feed materials, or other similar methods.
The number of different frequencies required depends upon the number of different type of materials in the feed stream and the accuracy of identification required. In a typical feedstream of recyclable materials, it is likely that employing eight different frequencies would provide acceptable accuracy. More frequencies may be utilized to obtain increased accuracy.
It can be seen in
Laser diodes may be required when the reflection “dip” is very narrow, such as the relatively narrow 940 nm dip for HDPE and the 1660 dip for PET plastic. Contrariwise, the LED radiation bandwidth matches very well with the wider reflection dips of HDPE between 1150 nm and 1250 nm and between 1375 nm and 1475. To obtain the greatest difference in the ratios of the reflected radiation intensity at different frequencies it is desirable to “match” the illuminated spectrum with the spectrum of the reflected radiation “dip”.
The power level of available LEDs and laser diodes is limited so matching the illuminator spectrum with the reflected radiation spectrum is advantageous to maximize the signal to noise ratio of the sensor system. Further, laser diodes with acceptable power output are available in fewer frequencies than that of LEDs. Therefore, it may be necessary to modify the output spectrum of an LED at a specific frequency. This can be accomplished, for instance, by placing the appropriate filter between the LED source and the feedstream articles. For example, the PET plastic dip at 1660 is more narrow than a typical LED output spectrum, but wider than that of a typical laser diode. Hence, a filter that limits the LED 20% bandwidth to about 1640 nm to 1690 nm, or 50 nm bandwidth, will result in a better spectrum match than either a laser diode or an LED without a filter.
In practice the identification process would include:
1) Sequentially illuminating the same region of the feedstream articles with each of the different frequency sources as the article passes the region of the sensor.
2) Measuring and storing the reflected radiation levels from each of the sources at a number of positions across the width of the feedstream. The articles are measured in at least 5 places across the feedstream, and are measured often enough that for a given feedstream speed (of say 500 to 1,000 feet per minute), the article is measured in at least five places along the length of the article, or at least such that on average each article is measured in at least 20 to 30 places.
3) Taking ratios of, or comparing the spectrum to, the measured reflected radiation levels at the various frequencies to determine the type of material for each measured area of the article.
4) Determining which type of material the article is substantially composed of by examining the measurements for a majority type of material, or type of material in selected regions of the article, or type of material with the highest contiguous counts.
In another embodiment of the invention, the system is capable of detecting the presence of carrier board (which does not contain lignin) in an article stream having newsprint (which contains lignin) and carrier board by determining the presence of lignin in articles in the stream by measuring the fluorescence of the articles when exposed to electromagnetic radiation at a frequency of about 532 nanometers (nm) (“green” light) and measuring the fluorescence at a frequency between about 600 and 700 nm; articles in which lignin is not detected are deflected to thereby separate carrier board from lignin-containing articles.
In another embodiment of the invention the apparatus includes a plurality of narrow bandwidth sources of electromagnetic energy including at least two sources of differing frequencies within the near-infrared range of from about 680 nm wavelength to about 2000 nm wavelength, for illuminating the articles.
In another embodiment of the invention the apparatus includes a plurality of narrow bandwidth sources of electromagnetic energy including red, green and blue narrow bandwidth sources, and at least four sources of differing frequencies within the near-infrared range of from about 680 nm wavelength to about 2000 nm wavelength, for illuminating the articles.
And in another embodiment, the invention includes a conveyor having a width of at least seven feet.
The present invention further includes methods of using the sorting system and its various components.
It is therefore a general object of the present invention to provide improved apparatus and methods for sorting objects by material and/or color, and particularly for sorting lignin-containing articles from those not containing lignin.
Still another object of the present invention is the provision of a system for sorting objects wherein the objects are analyzed as they travel along a conveyor.
Yet another object of the present invention is the provision of a system for detecting multiple classes of articles flowing along a conveyor without the need for a multiplexer or other complex mechanical systems.
Other and further objects, features and advantages of the present invention will be readily apparent to those skilled in the art upon a reading of the following disclosure when taken in conjunction with the accompanying drawings.
In a preferred embodiment, illustrated in
Still referring to
In addition to the use of entrainment air, it is also contemplated that other systems can be employed to maintain the sheeted material spread consistently on conveyor 20 and flowing in the proper direction. Exemplary of such a system is that disclosed by Grubbs, Kenny and Gaddis in U.S. Pat. No. 6,250,472, the disclosure of which is incorporated herein by reference.
Sorting system 10 can further comprise a plurality of receiving bins 40 into which material 1000 traveling along conveyor 20 can be sorted. Receiving bins 40 comprise a “default” receiving bin 42 into which material 1000 will flow if not directed into any of the preceding receiving bins, as well as at least one “selection” bin 44, and in the embodiment shown in
Selection bins 44A and 44B can also have associated therewith a source of directional gas 50A and 50B. Directional gas sources 50A and 50B comprise conduits for gas (e.g. air) flow in a direction across the top opening of each of selection bins 44A and 44B (and indicated by arrows) to ensure that sheeted material 1000 flowing along with the entrainment gas does not inadvertently enter receiving bins 44A and 44B. In other words, because the openings of receiving bins 40 would ordinarily cause eddying and other current variations of entrainment gas, it is possible that, without the use of directional gas flow, individual ones of material 1000 may enter one of selection bins 44A and 44B when not intended. Directional gas sources 50A and 50B provide a directional gas flow to maintain the flow of material 1000 along the flow of entrainment gas. Typically, directional gas sources 50A and 50B are powered by fans or blowers (not shown).
As illustrated in
In addition, the possibility exists on any surface after the termination of conveyor 20 that the flow of material 1000 may be interrupted due to friction. In order to reduce this possibility, in another preferred embodiment, a fluidizing flow of gas can also be created along such surface such as by providing a source of fluidizing gas 60 which creates a fluidizing flow of gas along the surface (indicated by arrows) to keep material 1000 from being hung up. For instance, the gas flow from directional gas source 50B can be partially diverted to be outletted at a proximate end of the surface 45 between the openings of selection bin 44A and 44B, as illustrated in
Each of selection bins 44A and 44B also has a deflector or sorter 70 associated therewith to direct selected individual ones of material 1000 into the respective selection bin 44A or 44B. Sorter 70 preferably comprises an air jet or other like device which, when actuated, will cause the selected material 1000 to pass through any directional gas flow across the opening of the specific selection bin 44A or 44B and thereinto.
More preferably, sorter 70 can comprise a plurality of air jets 72 extending generally across the width of sorting system 10. In this manner, when individual ones of the material 1000 is arrayed cross the width of conveyor 20 and the path of travel of material 1000, individual ones across the width of the path of travel of material 1000 can be selected to be directed into one of the selection bins 44A or 44B by actuating only those air jets 72 as would direct the selected material 1000 into the respective receiving bin 40.
Upstream from the first selection bin 44A, sorting system 10 comprises a detector system 100 capable of detecting one or more characteristics of material 1000 flowing along conveyor 20. Characteristics detected by detector system 100 can comprise reflectance (indicative of whiteness), color, presence of printing, presence of lignin or other characteristics of material 1000. Signals from detector system 100 are provided to a microprocessor 200 which then can provide a control system to sorters 70 to direct sorters 70 to direct individual ones of material 1000 into selection bins 44A or 44B provided certain measured criteria are met, or, microprocessor 200 can permit material 1000 to flow past selection bins 44A and 44B, by not actuating any of sorters 70, and thus be directed into default bin 42 if selection criteria are not met, or vice versa.
Detector system 100 comprises a plurality of sources of narrow bandwidth electromagnetic energy or radiation 110, such as lasers 110a, 110b, 110c, 110d, 110e, etc. As noted above, LEDs can also be employed, and/or LEDs having a filter limiting their bandwidth. The number of sources 110 employed and the center frequency of those sources will depend on material 1000 is to be sorted. For instance, if any type of plastic resin is to be sorted from a paper stream fewer frequencies 110 will be required than if the type of plastic resin also has to be sorted as well. For instance, frequencies of interest for plastics identification are 920 nm, 1210 nm, 1425 nm, 1660, 1725 to 2000 nm and 2125 nm. The primary aseptic packaging frequency of interest is 1455 to 1485 nm and 2000 nm and 2125 nm.
Sources 110 are positioned above conveyor 20 and sequentially illuminate a section across conveyor 20. In order to avoid overlap between adjoining illuminated sections, sources 110 preferably illuminate conveyor 20 in a relatively narrow line across the width of conveyor 20. The width (i.e., thickness of the beam along the direction of travel of material 1000) of the line across conveyor 20 illuminated by sources 110 will depend on factors such as how far apart sources 110 are disposed and the rate of travel of material 1000 on conveyor 20. In a typical example, the lines illuminated across the width of conveyor 20 by sources 110 should be no more than about 1.5 centimeters (cm) in thickness, most preferably no more than about 1.0 cm in thickness.
Electromagnetic energy from sources 110 illuminates material 1000 and is then reflected into a reflectance collector or detector array 120 to measure the reflected light intensity from material 1000 illuminated by sources 110. Data from the detector array 120 is processed by a control cabinet 130, which then actuates sorters 70. Detector array 120 is comprised of an array of devices which function to collect the light reflected from material 1000 when illuminated by electromagnetic energy from sources 110, such as photodiodes or a lens array. When lignin detection via fluorescence is desired, the relevant detector array 120 would have two associated photodiodes to enable lignin detection via fluorescence.
The use of narrow bandwidth sources 110 is especially important to enable both the lignin and the plastics detection and identification. Color identification could be accomplished with a broadband source, but the lignin identification will require a source with a narrow enough bandwidth in the green range so as not to overlap the red fluorescence. Plastics and other material identification in the near infrared range will also require narrow source bandwidths to identify their characteristic sharp absorption dips and/or reflective peaks.
Lignin content would be detected using illumination of material 1000 with a source 110 comprising a narrow band green laser at 532 nm and then measuring the resulting red fluorescence via a filter and high gain detector. The intensity of the red fluorescence is dependent on the distance between lignin-containing material 1000 and detector array 120.
A potential problem with this approach lies in the fact that not all material 1000 lies flat on conveyor 20. When material 1000 is raised up from the surface of conveyor 20, such as when material 1000 is “crumpled”, it is thus closer to detector array 120 and can skew the measurement of red fluorescence, since the reflection from material 1000 would be coming from a location closer to detector array 120 than if material 1000 was lying flat on conveyor 20. A solution to this problem is to factor out the intensity variation by determining lignin content through the ratio of the red fluorescence to the reflected green light, or a ratio with an average of the reflected intensity of the blue, red, and green sources.
An additional problem associated with lignin detection is the difference in the lignin fluorescence intensity due to the color of the object. Fluorescence from red and green colors tend to have a higher intensity than other colored material containing the same percentage of lignin. One possible solution to this problem is to compensate the calculated lignin content depending on the color of the material being analyzed. This can be accomplished by developing a look-up table which could be determined experimentally for the various colors and shades of colors.
There are several possible implementations of the lignin portion of the sensing. They all require two photodiode detectors per detection channel 125, with one diode allowed to receive only the red fluorescence, and potentially longer, wavelengths. In one embodiment, illustrated in
More specifically, the embodiment shown in
In the embodiment of
Further, in order to reduce noise each pulse from a source 110 would be split into a plurality of short pulses to achieve the effect of a “chopper” system. For instance, source 110a would be actuated, for example, for 35 to 40 μsec, the reflected light measured, and then the detector signal measured with no illumination for 10 to 15 μsec. A “train” of such on-off pulses would require about 250 μsec to complete. During this time, if material 1000 were travelling at 1,000 feet per minute, it would have moved about 0.125 cm. Source 110b would then be pulsed in the same fashion as above but with the beam offset in the direction of motion by about 0.125 cm. The illumination from each subsequent source 110 would be offset by about 0.125 cm, so that each different frequency source 110 sequentially illuminates the same line across the material 1000. Sources 110 would be aligned vertically to minimize effects from variation in height of the object. The light collected by detector array 120 would be maximized, as the field of view is approximately 2.5 cm in diameter while material 1000 is illuminated during travel through the center 1.25 cm of the field of view. The beam width from each source 110 would be on the order of about 0.3 cm to 0.63 cm further “averaging” the measurements. If a slower speed for conveyor 20 is used, the on pulse would be lengthened such that the same line across material 1000 is still illuminated by each source 110. This approach does require that the measurements from each source 110 laser illumination are stored for each detector array 120 until a full set of 8 to 12 measurements are made. Once the full set of measurements is made for each array 120 the appropriate ratios can be calculated and identification made.
In operation, material 1000 is fed onto conveyor 20 using, e.g., the system disclosed by Grubbs, Kenny and Gaddis in U.S. Pat. No. 6,250,472. Entrainment airflow is also directed in the direction of the flow of travel of material 1000 defined by conveyor 20, along the direction indicated by the arrows in
Near the end of conveyor belt 304 there is located a sensor system 312 which is shown in more detail in
Downstream of the conveyor belt 302 there is a primary discharge conveyor 314, an upper discharge chute 316, and a lower discharge chute 318.
An array of downwardly directed air jets 320 can blow a first category of selected articles out of the primary discharge stream into the lower discharge chute 318, which can be referred to as a first alternative destination.
An array of upwardly directed air jets 322 can deflect a second category of selected articles from the primary stream into the upper discharge chute 316, which can be referred to as a second alternative destination.
Thus the apparatus 300 can separate the stream of articles on conveyor belt 302 into three separate discharge streams.
Referring now to
Also mounted within housing 24 adjacent the light source panel 326 is an array of light collectors 344.
The LEDs 328 through 342 illuminate the conveyor belt 302 and the paper articles 1000 carried by conveyor belt 302, and light reflected or fluoresced from the articles 1000 is collected by the collectors 344. The collectors 344 may also be described as telescopes 344. The telescopes 344 comprise a lens 346 at one end of a barrel 348 with antireflective grooves and a photodiode 350 at the other end. Light is received at the photodiode 350 from an area on the target material below. The size of the target area is controlled by the focal length and location of the lens. In this case the area is about 1 inch diameter. And the lenses are on one inch spacing across the array. The sensor utilizes a silicon photodiode for the visible part of the spectrum, but these are not responsive in the infrared area of the spectrum. So for the infrared (>1000 nm wavelength) wavelengths, InGaAs (Indium-Gallium-Arsenide) type diodes are used. Additionally, the lignin sensor system must operate with a silicon detector which is covered with a red filter so it only responds to the red (fluorescence) component of the reflected light from a bright green LED flash. Rather than use three telescope arrays to implement the three above mentioned receiving tasks, one telescope array with three sensor elements is used. The three photodiodes (silicon, InGaAs, and silicon/red filter) are side by side in a row under each telescope lens 346. The row of three sensors is oriented along the material travel direction. The InGaAs diode is on the centerline of the telescope, so it receives light from directly under the telescope on the target material. The other diodes, being located off center, receive light from either upstream on the target material or downstream. In this way the telescope array reads data from a row of pixels (spots) on the target material for infrared, and a separate row of pixels on the target material for the visible located upstream and parallel to the infrared row, and a 3rd row of pixels on the target material for the lignin fluorescence located downstream and parallel to the infrared reading. The data which is “non-coherent” is corrected in the software by delaying the upstream readings data so that it is returned to its correct location to produce a coherent image. In other words, the various data is not read from the target material at the same place and at the same time, but this is corrected in the software.
As also seen in
As seen in
For each different frequency or wavelength of narrow-band width source utilized, the array of light sources includes a row such as row 342 across the width 360 of the conveyor 302. Only four such rows are illustrated in
In combination with the high speed paper handling system of the present invention which can operate at speeds of as much as 1200 feet per minute (fpm), a paper sorting system utilizing the sensor apparatus of the present invention can provide the extremely high capacities that are necessary to make automated paper sorting economical. As will be appreciated, the width of the belt and the speed of the belt basically determine the volume of paper that can be handled, assuming that the sensor system is capable of identifying and sorting the material at such a speed. The system of the present invention as noted operates at high speeds which can be generally described as operating at speeds of at least about 600 fpm, more preferably at least about 1000 fpm and most preferably 1200 fpm or greater.
Turning now to
One preferred sensor system 312 uses a light source array having red, green and blue sources in the visible spectrum and eight different infrared lengths as shown in the following Table I:
As previously noted a green light source located at 532 nm wavelength (not shown on
The eight different near infrared wavelengths from Table I are illustrated along the lower portion of
The first curve 362 represents the bandwidth of the 935 nm source, which may for example be a model HEMT-3301 LED available from Agilent Technologies.
Curve 364 is representative of the 1050 nm wavelength source which may for example be a model LED 1050-03 available from Epitex.
Curve 366 is representative of the source centered at around 1200-1210 nm which may for example be a model LED 1200-03 available from Epitex, in combination with a custom 1200 nm×40 nm bandwidth interference filter available from Intor, Inc.
Curve 368 is representative of the 1300 nm source which may for example be a model LED 1300-03 available from Epitex.
Curve 370 is representative of two of the sources, namely the 1420 nm source and the 1480 nm source. The 1420 nm source is for example provided with a model LED 1450-03 available from Epitex and having a center frequency of 1450 nm wavelength in combination with a custom 1420 nm×40 nm bandwidth interference filter available from Intor, Inc. to provide a source centered at approximately 1420 nm wavelength.
The 1480 nm source is in turn provided by an identical 1450 nm LED which may be a model LED 1450-03 available from Epitex, combined with a custom 1480 nm×40 nm bandwidth filter available from Intor, Inc. to provide a source centered at approximately 1480 nm wavelength.
Curve 372 is representative of the 1550 nm source which may for example be a model LED 1550-03 available from Epitex.
Finally, curve 374 is representative of the 1650 nm source which may for example be a model L8245 LED available from Hamamatsu.
Referring now to the positions of the various sources 362 through 374 in relationship to the characteristic absorption spectra of the four materials located in the upper portion of
Similarly, sources 366 and 368 can be used to detect the characteristic dip 378 in the polyethylene absorption spectra centered at about 1210 nm in wavelength. Again for PVC there is also a dip at about 1210 nm which is less pronounced.
The two sources at 1420 nm and 1480 nm wavelength represented in curve 370 can be utilized to distinguish between fiber, polyethylene and PET. The PET material will show a small increase in reflectance between 1420 and 1480 nm. The polyethylene material will show a much sharper increase in reflectance between 1420 and 1480 nm. The PVC will show an increase somewhat between that of PET and polyethylene. The fiber material, in contrast, will show a decrease in reflection between 1420 and 1480 nm.
Through a combination of such observations focused on various characteristic portions of the absorption spectra of the materials of concern, an array of sources can be selected which will allow the desired selection to occur.
As previously described, a sequential LED flash and Read sequence is designed to reduce motion induced chromatic aberration. Since the wavelength readings must be conducted sequentially, a problem results because the material may move during the read sequence. If the total reflectivity of the pixels area changes because of the motion, for example if a black edge is moving into view in a white area, a type of false spectral signature is created. In this case the later readings would be lower reflectivity as the black edge moves in. This would result in a non-flat spectrum reading. However in this case the correct spectrum should be flat because there is no color present. (black and white only) This situation is partially corrected by using a symmetrical redundant read sequence shown in the following Table II.
The sequence is symmetrical about the center and each color or wavelength is read twice. To produce the output result the two readings for each wavelength are averaged together. The benefit of sequence is as follows. If a black edge, for example, is moving across the pixel, then the read spectrum will be “tilted”. If we read the spectrum again in the reverse order, the “tilt” will be in the opposite direction. If we average the two spectra together, the result is correct and flat.
Thus the present system utilizing the preferred array of sensors in Table I provides the ability to distinguish between paper materials and plastic materials. Additionally the ability is provided to distinguish and identify various types of paper materials including carrier board, white paper and newspaper articles. Additionally, the ability is provided to distinguish and identify various types of plastic materials including polyethylene, polyester (such as PET), polystyrene, and PVC materials.
It is seen in
Identification software developed for the sensor system 312 uses a spectrum shape analysis technique with selective weighting. The intent of the analysis is to identify a spectrum by its shape and to ignore features other than the shape. The spectrum data for each pixel is processed to normalize the size of the features to a standard level. Undesirable attributes which would detract from the shape analysis are removed. The shape may then be compared to several standard references to determine the best match. The mathematical technique used for this is to convert an example spectrum to a collection of slope segments which are thought of as a vector or point location in a space of order n=7. (This is the number of wavelengths−1) This set of 7 numbers is stored as a reference. Unknown spectra (data) are compared to the reference by computing the distance in space from the reference vector to the unknown vector. This calculation is made for all of the stored reference spectral shapes or vectors, and the smallest result represents the reference that most closely matches the unknown. For this reason the software is referred to as “vector” software. It is believed that this technique provides a maximally optimum identification. In other words, all valid information from the input data is utilized. Additionally, the vector match analysis may be weighted to favor shape features which are deemed to be most important for a particular separation. The weighting mask used can be generated by an analysis tool (software) to produce an optimum separation. The way this works is as follows. After reading data from sample material and generating a number of vector references, a subset of references may be chosen for weighted separation. This subset would include two types of materials for which an improved separation is desired. An example would be PE and PVC. It is known that these two plastics produce relatively similar spectra. The question that is essentially answered by the analysis tool is “Which part of the spectrum includes the shape difference that is most important for separating these items?” The answer is produced in the form of a weighting mask which is simply a set of n−1 (7) weighting factors. Having this, the vector or shape analyzer software can use the weighting mask to favor spectral features that are important to the separation, and ignore features that do not matter. This will produce a more reliable identification. The analysis tool generates the weighting mask by statistical analysis of two selected reference sets. (several PE and several PVC vectors, for example) to determine where the difference lies.
All cited patents and publication referred in this application are incorporated by reference.
The invention thus being described, it will be apparent that it may be varied in many ways. Such variations are not to be regarded as a departure from the spirit and scope of the present invention and all such modifications as would be apparent to one skilled in the art are intended to be included within the scope of the following claims.
This application is a continuation-in-part of our pending U.S. patent application Ser. No. 10/921,000 filed Aug. 18, 2004, entitled “Sorting System Using Narrow-Band Electromagnetic Radiation”, the details of which are incorporated herein by reference.
Number | Name | Date | Kind |
---|---|---|---|
4131540 | Husome et al. | Dec 1978 | A |
4541530 | Kenny et al. | Sep 1985 | A |
4657144 | Martin et al. | Apr 1987 | A |
4718558 | Castaneda | Jan 1988 | A |
4909930 | Cole | Mar 1990 | A |
4919534 | Reed | Apr 1990 | A |
5085325 | Jones et al. | Feb 1992 | A |
5134291 | Ruhl, Jr. et al. | Jul 1992 | A |
5141110 | Trischan et al. | Aug 1992 | A |
5220172 | Berthold et al. | Jun 1993 | A |
5242059 | Low et al. | Sep 1993 | A |
5260576 | Sommer, Jr. et al. | Nov 1993 | A |
5297667 | Hoffman et al. | Mar 1994 | A |
5314072 | Frankel et al. | May 1994 | A |
5318172 | Kenny et al. | Jun 1994 | A |
5318173 | Datari | Jun 1994 | A |
5333739 | Stelte | Aug 1994 | A |
5423431 | Westin | Jun 1995 | A |
5443164 | Walsh et al. | Aug 1995 | A |
5520290 | Kumar et al. | May 1996 | A |
5615778 | Kaiser et al. | Apr 1997 | A |
5675416 | Campbell et al. | Oct 1997 | A |
5676256 | Kumar et al. | Oct 1997 | A |
5770864 | Dlugos | Jun 1998 | A |
5813543 | Gesing et al. | Sep 1998 | A |
5848706 | Harris | Dec 1998 | A |
5955741 | Kayani | Sep 1999 | A |
5966217 | Roe et al. | Oct 1999 | A |
6060677 | Ulrichsen et al. | May 2000 | A |
6068106 | Brizzi et al. | May 2000 | A |
6080950 | Jalink | Jun 2000 | A |
6250472 | Grubbs et al. | Jun 2001 | B1 |
6335501 | Khalfan | Jan 2002 | B1 |
6353197 | Ulrichsen et al. | Mar 2002 | B1 |
6369882 | Bruner et al. | Apr 2002 | B1 |
6374998 | Grubbs et al. | Apr 2002 | B1 |
6380503 | Mills et al. | Apr 2002 | B1 |
6497324 | Doak et al. | Dec 2002 | B1 |
6506991 | Eixelberger et al. | Jan 2003 | B1 |
6509537 | Krieg et al. | Jan 2003 | B1 |
6778276 | Bruner et al. | Aug 2004 | B2 |
6832783 | Lawandy | Dec 2004 | B2 |
6914678 | Ulrichsen et al. | Jul 2005 | B1 |
7019822 | Doak et al. | Mar 2006 | B1 |
7113272 | Bourely et al. | Sep 2006 | B2 |
7262380 | Ulrichsen et al. | Aug 2007 | B1 |
20030132142 | Kumar | Jul 2003 | A1 |
20060081510 | Kenny et al. | Apr 2006 | A1 |
Number | Date | Country |
---|---|---|
0484221 | May 1992 | EP |
Number | Date | Country | |
---|---|---|---|
20070158245 A1 | Jul 2007 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 10921000 | Aug 2004 | US |
Child | 11615052 | US |