METHOD AND APPARATUS FOR SORTING

Abstract

A method of sorting is described and which includes a step of acquiring a multiplicity of synchronized image signals of a product stream which is to be sorted; generating a multiplicity of fused sensor signals; forming an image model previously acquired from the objects to be sorted; identifying objects in the product stream, and generating object presence and defect signals; determining a spatial orientation of the objects in the product stream; detecting the defects and removing the defects from the product stream.

Claims

1. A method of sorting comprising: acquiring a multiplicity of synchronized image signals of individual objects of interest and defects from a plurality of image generating devices; generating a multiplicity of fused sensor signals by combining the multiplicity of synchronized image signals of the image generating devices; forming an image model comprising image signals previously acquired from the objects of interest and defects; applying the image model to the multiplicity of fused sensor signals, and forming a resulting object presence and defect signals; identifying individual objects of interest with the respective object presence and defect image signals; determining a spatial orientation and location of the objects of interest in each of the image signals; detecting defects within the object presence and defect signals by comparing defect aspects relative to object aspects, to object images formed of the object presence and defect image signals; and removing the objects of interest having defects from the product stream.

2. A method as claimed in claim 1, and wherein the synchronized image signals represent individual objects of interest such as agricultural products traveling in a product stream, and which have characteristics, and aspects, which are deemed acceptable for further processing, and characteristics, and aspects, which are deemed unacceptable, for further processing.

3. A method as claimed in claim 1, and wherein the synchronized image signals are formed by a selective synchronization of the image generating devices, and by utilizing a known position, orientation, and an operational response of the respective image generating devices so as to allow the generation of an accurate spatial resolution of each of the objects of interest travelling in the product stream, and to further align the signal features of each of the image signals.

4. A method as claimed in claim 1, and wherein the image model is formed from a methodology which includes previously acquiring a multiplicity of image signals from known acceptable and unacceptable objects of interest such as agricultural products, and the like.

5. A method as claimed in claim 1, and wherein the step of identifying individual objects of interest further includes the step of identifying one or more pixel groups in each of the object presence, and defect image signals, and which identify the objects of interest or defects.

6. A method as claimed in claim 1, and wherein the step of determining the spatial orientation and location of the respective objects of interest further comprises developing a prior source of knowledge of object aspects which is applied to object images, and which are formed of the object presence and defect signals.

7. A method as claimed in claim 1, and wherein the step of detecting unacceptable agricultural products or defects further comprises developing a prior source of knowledge of defect aspects, relative to object aspects to a multiplicity of object images formed of the object presence, and defect image signals.

8. A method as claimed in claim 1, and before the step of acquiring the multiplicity of synchronized image signals, the method further comprises providing a product stream of individual objects of interest, such as agricultural products having both acceptable agricultural products, and unacceptable products which must be removed from the product stream; and passing the product stream having both the acceptable agricultural products, and the unacceptable agricultural products through an inspection station.

9. A method as claimed in claim 8, and wherein the method further comprises providing a first controller which predicts the presence of the objects of interest, and defects, in the fused sensor signals, and which further applies the image model to at least some of the multiplicity of fused sensor signals.

10. A method as claimed in claim 9, and wherein the method further comprises providing a second controller which identifies individual objects of interest and defects in the object presence and defect image signals; determines the spatial orientation of the identified objects of interest and defects travelling in the product stream; identifies the defect in the defect image signal; identifies the location of the defect in the defect image signal; and generates an unacceptable agricultural product image signal.

11. A method as claimed in claim 10, and further comprising: positioning a defect removal station downstream of the inspection station and along a path of travel of the product stream; providing an ejector, and positioning the ejector in the defect removal station, and which is effective, when made operational, to remove the unacceptable agricultural products from the product stream passing by the defect removal station; and providing a third controller which is controllably coupled with, and renders operational the ejector, and wherein the third controller is coupled in signal receiving relation relative to the unacceptable agricultural product image signal which is generated by the second controller, and which further renders the ejector operational to remove the unacceptable agricultural products from the product stream passing by the defect removal station.

12. A method of sorting, comprising: acquiring a multiplicity of synchronized image and sensor signals, each having discreet signal features, from a plurality of image generating devices and sensors, and wherein the synchronized image and sensor signals represent individual objects of interest such as agricultural products which are traveling in a product stream, and which have characteristics, and aspects which are deemed acceptable for further processing, and characteristics and aspects which are deemed a defect, and unacceptable, for further processing; generating a multiplicity of fused sensor signals by combining the multiplicity of synchronized image and sensor signals by a selective synchronization of the image generating devices and sensors, and by utilizing a known position, orientation, and an operational response of the respective image generating devices and sensors so as to allow the generation of an accurate spatial resolution of each of the objects of interest products travelling in the product stream, and to further align the signal features of each of the image and sensor signals; predicting the presence of the objects of interest, and possible defects in the fused sensor signals by applying an image model previously formed from a multiplicity of image signals which were acquired from the objects of interest, and the defects, to the multiplicity of fused sensor signals so as to facilitate the formation of a resulting object presence image signal; and a defect image signal; identifying the individual objects of interest with the object presence, and defect image signals, by identifying one or more of a group of pixels in each of the object presence, and defect image signals; determining a spatial orientation of the objects of interest travelling in the product stream by applying a prior source of knowledge of the object aspects to a multiplicity of object images which are formed of the object presence, and defect image signals; detecting defects within unacceptable objects of interest by applying a prior source of knowledge of defect aspects relative to object aspects, to the object images formed of the object presence, and defect signals; identifying the location of the unacceptable objects of interest having defects in the object image signals; and removing the unacceptable objects of interest having defects from the product stream.

13. A method as claimed in claim 12, and wherein the discreet signal features of the multiplicity of synchronized image signals are selected from the group of image signals provided by a hyperspectral or multispectral imager and/or scanner.

14. A method as claimed in claim 12, and wherein the synchronized image signals are formed by a methodology which includes a step of spatially registering the respective image signals.

15. A method as claimed in claim 12, and wherein the aspects and characteristics of the objects of interest which are deemed acceptable for further processing are selected from individual products having known acceptable qualities.

16. A method as claimed in claim 12, and wherein the aspects and characteristics of the objects of interest which are deemed unacceptable for further processing are selected from individual products having known unacceptable qualities.

17. A method as claimed in claim 12, and wherein the aligning of the signal features of each of the synchronized image signals so as to form, at least in part, the multiplicity of fused sensor signals comprises a partial registration of the image signals with each other, and with an ejector controller.

18. A method as claimed in claim 12, and wherein the image model is formed by a methodology which includes a step of utilizing a standard classification algorithm.

19. A method as claimed in claim 12, and wherein the prior source of knowledge of the object aspects which is applied to the multiplicity of object images, and which are used to determine the spatial orientation of the identified objects of interest is formed by the methodology which comprises a step of conducting an object shape analysis; and conducting an object aspect measurement.

20. A method as claimed in claim 12, and wherein the prior source of knowledge of the defect aspects, and which is applied to the multiplicity of object images, is formed by the methodology which comprises a step of qualifying unacceptable pixel groups found in the object images, with object regions identified in the object aspects.

21. A method as claimed in claim 12, and wherein the step or removing the unacceptable objects of interest from the product stream further comprises a step of removing an unacceptable portion of an object of interest from a remaining acceptable portion of the same object of interest.

22. A method as claimed in claim 12, and before the step of acquiring the multiplicity of synchronized image signals, the method further comprises providing a product stream of individual objects of interest which have the character and aspects of both acceptable and unacceptable objects of interest; and passing the product stream having both the acceptable and unacceptable objects of interest through an inspection station.

23. A method as claimed in claim 12, and wherein the method further comprises providing a first controller which predicts the presence of the objects of interest, and defects in the fused sensor signals, and which further applies the image model to the multiplicity of fused sensor signals.

24. A method as claimed in claim 23, and wherein the method further comprises providing a second controller which identifies individual objects of interest and defects travelling in the product stream; determines the spatial orientation of the identified individual objects of interest travelling in the product stream; detects the objects of interest; identifies the location of the defects in the object presence and defect image signals; and generates a signal indicating the presence and location of the defect in the production stream.

25. A method as claimed in claim 24, and further comprising: providing a defect removal station, and positioning the defect removal station downstream of the inspection station; providing an ejector, and positioning the ejector in the defect removal station, and which is effective, when made operational, to remove the unacceptable objects of interest from the product stream; and providing a third controller which is controllably coupled with, and renders operational the ejector, and wherein the third controller is coupled in defect signal receiving relation relative to the second controller, and which further renders the ejector operational to remove the unacceptable objects of interest from the product stream.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Preferred embodiments of the invention are described below with reference to the following accompanying drawings.

[0012] FIG. 1 is a greatly simplified, schematic view of an apparatus which may be utilized to practice the present methodology.

[0013] FIG. 2 is a greatly simplified, schematic view of the methodology of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] This disclosure of the invention is submitted in furtherance of the constitutional purposes of the U.S. Patent Laws to promote the progress of science and useful arts (Article 1, Section 8).

[0015] The present methodology of the invention is generally indicated by the numeral 10, and is best understood by a study of FIGS. 1 and 2, respectively. As best understood by a study of FIG. 1, the method employs, in one form of the invention, a conveying device which is fragmentarily shown in FIG. 1 as a conveying device 11. The conveying device, as illustrated in that drawing, shows a distal, discharge end 12 of a conveyor belt 11, and which further has an upper conveying surface 13 which supports a product stream 14 to be inspected. The product stream 14 is formed of individual objects of interest 15, which are released from the distal, discharge end 12, and allowed to move, under the influence of gravity, in a downwardly directed path of travel which is generally indicated by the numeral 16. A source of the objects of interest or products which is to be inspected, and sorted, is labeled by the numeral 17.

[0016] The product stream 14 has objects of interest or products 15 which, in one form of the invention, may include various agricultural products which have both acceptable features for further processing 20, or unacceptable features for processing which are generally indicated by the numeral 21. For example, in the processing of potatoes, unacceptable features 21 of a potato product would be the presence of sugar ends or regions of rot, which will be detected by the methodology as described, hereinafter. Positioned downstream of the distal discharge end 12, of the conveying device 11, are a multiplicity of image capturing devices 22 which are generally shown, and which further are positioned laterally outwardly relative to the downwardly directed path of travel 16. The multiplicity of imaging capturing devices 22 (which may include, for example hyperspectral or multispectral cameras of assorted designs) are utilized, in a first step of the present method, and which includes acquiring a multiplicity of synchronized image signals 23 each having discreet signal features, from a plurality of image generating devices 22. The multiplicity of image capturing devices 22 produce a multiplicity of synchronized image signals 23 which are then selectively supplied to a first controller, and which is further generally indicated by the numeral 24. The synchronized image signals 23 represent individual objects of interest 15 such as agricultural products which are traveling in the product stream 14, and which have characteristics and aspects which are deemed acceptable 20, for further processing, and characteristics and aspects which are deemed a defect, or unacceptable 21 for further processing. The downwardly directed path of travel 16 of the product stream 14 passes through a downstream inspection station which is generally indicated by the numeral 25, and a downstream defect removal station 26, and which is further located elevationally, below the inspection station 25. The multiplicity of image capturing devices 22 are positioned so as to acquire image signals 23 from the objects of interest 15 while they pass through the inspection station 25 in a manner well understood in the art. Assorted optical reflectors 27, and optical combiners 28, are provided, and which co-align multiple image capturing devices 22. Further, well known background elements are provided, and which additionally are positioned laterally, outwardly relative to the product stream 14, and which is passing through the inspection station 25.

[0017] As best seen by reference to FIG. 1, the methodology of the present invention includes a step of generating a source of synchronized electromagnetic radiation 31 which is directed towards, and reflected at least in part from, the product stream 14, and which further is formed of the individual objects of interest 15 passing through the inspection station 25. As seen in FIG. 1, an electromagnetic radiation emitter 30 is generally illustrated, and which is positioned laterally, outwardly relative to the inspection station 25. When selectively energized, the electromagnetic radiation emitter 30 emits a source of electromagnetic radiation 31 which is directed towards, and reflected at least in part by the individual objects of interest 15 passing through the inspection station 25. This emitter of electromagnetic radiation may produce various wavelengths of electromagnetic radiation which enhances the ability of the respective image capturing devices 22 to capture individually unique images of the objects of interest 15 which allow for the identification of assorted characteristics and aspects of each object of interest, and in particular, various defects in the product stream 14. For example, the defects could include sugar ends or regions of rot which are present in a potato products undergoing the inspection and sorting. As also seen in FIG. 1, an electromagnetic radiation emitter control signal 32 is provided to each electromagnetic radiation emitter 30. This control signal 32 causes the selective energizing of each of the emitters 30. The selective energizing of the individual electromagnetic radiation emitters 30 to achieve the aforementioned benefits of this invention, is explained in significant detail in U.S. Pat. No. 9,266,148, the teachings of which are incorporated by reference herein. The method 10 of the present invention includes still another step of generating a multiplicity of fused image and sensor signals 34 by combining the multiplicity of synchronized image and sensor signals 23 by a selective synchronization of the image and sensor generating devices 22 and the respective electromagnetic radiation emitters 30, and by utilizing a known position, orientation and an operational response of the respective image and sensor generating devices 22, so as to allow the generation of an accurate spatial resolution of each of the objects of interest 15 which are traveling in the product stream 14, and to further align the signal features of each of the image and sensor signals. As seen in FIG. 1, the control signal 32 is operably controlled by the selective synchronization step 34. The selective synchronization 34 of the image and sensor signals and the energizing of the electromagnetic radiation emitters 30, is performed in a manner so as to inhibit the destructive interference that may occur when the electromagnetic radiation emitters 30, and sensors 22 are simultaneously rendered operable. In the methodology 10 of the present invention, the synchronized image and sensor signals 23 are formed by a methodology which includes the step of selecting image signals 23 from multiple, different sensors 22. Still further in the methodology 10 of the present invention the aligning of the signal features of each of the synchronized image and sensor signals so as to form, at least in part, the multiplicity of fused sensor signals 35 comprises the step of conducting a spatial registration of the sensor signals 23 with each other, and with an ejector controller 100, and which will be discussed in greater detail, below. The high aspect spatially fused sensor signals 35 are then supplied to the first controller 24 as seen in FIG. 2.

[0018] The high aspect spatially fused sensor and image signals 35 are provided to the controller 24, and to individual modules within the controller 24, (FIG. 2) and which perform many of the methodology steps of the present invention. In this regard, the first controller 24 has a module 40 which contains prior knowledge of object defects which have been acquired from the previous inspection of similar objects of interest. This prior knowledge of the object defects 40 is provided to another module 41, and which conducts supervised training of the first controller 24, so as to allow the present methodology to teach itself how to improve the sorting reliability of the present invention 10. The supervised training module 41 is supplied with a portion of the high aspect, spatially fused sensor and image signals 35, and which has been acquired from the multiple image capturing devices 22. Still further the first controller 24 includes a prediction module 42 which predicts the presence of the objects of interest 15, and possible defects in the fused sensor signals 35 by applying an image model 43, which was previously formed from a multiplicity of image and sensor signals which were acquired from the objects of interest 15, and the defects, to the multiplicity of fused sensor signals 35, so as to facilitate the formation of a resulting object presence image signal 44, and defect image signal 45, respectively. The present methodology 10 includes a second controller 50.

[0019] The second controller 50 is operably coupled with the first controller 24. Still further the second controller has a module 60 which implements a step in the methodology 10 which includes identifying the individual objects of interest 15 with the object presence signals 44, and defective image signals 45, by identifying one, or more of a group of pixels in each of the object presence and defect image signals. Still further, the second controller 50, and more specifically the module for identifying objects of interest 60, is operable to supply a signal 61 to the module for supervised training 41, so as to allow the module for supervised training 41 to continue to learn as the inspection process proceeds so as to increase the accuracy and sorting efficiency of the presently disclosed methodology 10. The object presence and defect signals 44 and 45 are supplied to other modules in the second controller 50. More specifically, the second controller 50 has a module for implementing a step which includes storing and supplying a source of knowledge of the object aspects for use in the sorting process. This module 70 supplies the stored information to another module 71, and which implements a step in the methodology 10 of determining a spatial orientation of the objects of interest 15 traveling in the product stream 14 by applying the prior source of knowledge of the object aspects 70, to a multiplicity of object images which are formed of the object presence and defect image signals 44 and 45, respectively. Still further the second controller 50 includes a module which provides a prior source of knowledge of defect aspects 80, relative to object aspects. In this regard this prior knowledge 80 is provided to a module 81 for detecting defects within the unacceptable objects of interest 15 by applying the prior source of knowledge 80 to object aspects 70, and to the object images formed of the object presence and defect signals 44 and 45, respectively. The module for detecting defects 81, and the object presence, and defect signals 44 and 45, generates a defect signal which is generally indicated by the numeral 82, and which further is itself supplied to an object removal control 90. The object removal control 90 generates a signal 91 which is provided to an ejector controller, which further is generally indicated by the numeral 100. Therefore, the methodology of the present invention 10 after identifying the location of the unacceptable objects of interest having defects 81, in the object image signals, the methodology 10 includes a step of removing the unacceptable objects of interest having defects from the product stream by means of the ejector controller 100. The ejector controller operably controls an air manifold 101, (FIG. 1) of conventional design, and which is further configured to release a pressurized blast of ambient air 102, which is operable to remove defective objects of interest 21 from the downwardly descending product stream 16, so as to form a product stream having only acceptable objects of interest 15.

Operation

[0020] The operation of the described embodiment of the present invention is believed to be readily apparent, and is briefly summarized at this point.

[0021] In its broadest aspect the present invention relates to a method of sorting 10 which comprises a first step of acquiring a multiplicity of synchronized image signals 23 of individual objects of interests 15, and defects 21, from a plurality of image generating devices 22. The method includes another step of generating a multiplicity of fused sensor signals 34 by combining the multiplicity of synchronized image signals 23 of the image generating devices 22. Still further the method of sorting 10 of the present invention includes yet another step of forming an image model 43 comprising image signals 23 which were previously acquired from the objects of interest 15, and the defects 20. The method includes yet another step of applying the image model 43, to the multiplicity of fused sensor signals 34, and forming a resulting object presence 44, and defect signals 45, respectively. The method of the present invention 10 includes another step of identifying individual objects of interest 60, with the respective object presence 44, and defect image signals 45. The method of the present invention includes yet another step of determining a spatial orientation and location of the objects of interest 71 in each of the image signals 23. The method includes yet another step of detecting defects 81 within the object presence and defect signals 44 and 45, respectively, by comparing defect aspects 80 relative to object aspects 70, to object images formed of the object presence and defect image signals 44 and 45, respectively. Finally, the present invention in its broadest aspect includes a last step of removing 100 the objects of interest 15 having defects 21 from the product stream 14.

[0022] The method 10 of the present invention includes another step, and wherein the synchronized image signals 23 represent individual objects of interest 15 such as agricultural products traveling in a product stream 14, and which have characteristics, and aspects, which are deemed acceptable for further processing 20, and characteristics and aspects which are deemed unacceptable for further processing 21. This is best seen in FIG. 1. The method includes another step, and wherein the synchronized image signals 23 are formed by a selective synchronization 23 of the image generating devices 22, and by utilizing a known position, orientation and an operational response of the respective image generating devices 22 so as to allow the generation of an accurate, spatial resolution of each of the objects of interest 15 traveling in the product stream 14, and to further align the signal features of each of the image signals 23. In the present invention the method includes a further step, and wherein the image models 43 are formed from a methodology which includes previously acquiring 40 a multiplicity of image signals 23 from known acceptable 20, and unacceptable 21 objects of interest 15, such as agricultural products, and the like. In the present methodology 10 the step of identifying individual objects of interest 60 further includes the step of identifying one or more pixel groups in each of the object presence 44, and defect image signals 45 which identify the acceptable objects of interest 20 or defects 21.

[0023] In the present invention the methodology 10 includes another step of determining the spatial orientation 71, and location of the respective objects of interest 15 and further comprises another step of developing a prior source of knowledge of object aspects 70, and which is applied to object images which are formed of the object presence and defect signals 44 and 45, respectively. In the present methodology 10, the step of detecting unacceptable agricultural products, or defects in the objects of interest 81, further comprises another step of developing a prior source of knowledge of defect aspects 80, relative to object aspects 70, to a multiplicity of object images formed of the object presence and defect image signals 44 and 45, respectively. In the present methodology 10, and before the step of acquiring the multiplicity of synchronized image signals 23, the method further comprises still another step of providing a product stream 14 of individual objects of interest 15, such as agricultural products having both acceptable agricultural products 20, and unacceptable products 21, and which must be removed from the product stream 14. The method includes another step of passing the product stream 14 having both the acceptable agricultural products 20 and unacceptable agricultural products or objects of interest 21 through an inspection station 25 (FIG. 1). In the present invention, the methodology 10 includes another step of acquiring the multiplicity of synchronized image signals 23, and after the step of providing the product stream 14, generating a source of synchronized electromagnetic radiation 31 which is directed towards and reflected, at least in part, from the product stream 14 of the agricultural products or objects of interest passing through the inspection station 25. In the present invention the method 10 further comprises a step of providing a first controller 24 which predicts the presence of the objects of interest 15 and defects 21 in the fused sensor signals 35, and which further applies the image model 43 to at least some of the multiplicity of fused sensor signals 35. The method further comprises another step of providing a second controller 50, and which identifies individual objects of interest 15, and defects 21, in the object presence and defect image signals 44 and 45; and further determines the spatial orientation 71 of the identified objects of interest 15 and defects 21 traveling in the product stream 14. The method 10 also includes another step of identifying the defects 81, and the defect image signal 82 identifies the location of the defects in the defect image signals. An unacceptable agricultural product image signal is provided to the third controller or object removal control 90. In the present methodology 10, the third controller 90 is controlled, and coupled with, and renders operational an ejector 101. The third controller 90 is coupled in signal receiving relation relative to the unacceptable agricultural product image or defect signal 82 which is generated by the second controller 50, and further renders the ejector 101 operational by means of the ejector controller 100 to remove the unacceptable agricultural product or any defects 21 from the product stream 14, and which is passing by, or through, the defect removal station 26.

[0024] More specifically the methodology 10 of the present invention further includes a step of acquiring a multiplicity of synchronized image and sensor signals 23, each having discreet signal features, from a plurality of image generating and sensor devices 22. The synchronized image and sensor signals 23 represent individual objects of interest 15, such as agricultural products, which are traveling in a product stream 14, and which have characteristics and aspects which are deemed acceptable for further processing 20, and characteristics and aspects which are deemed a defect and unacceptable 21 for further processing. The present method includes another step of generating a multiplicity of fused image and sensor signals 34 by combining the multiplicity of synchronized image and sensor signals 23 by a selective synchronization of the image and sensor generating devices 22, and by utilizing a known position, orientation, and an operational response of the respective image and sensor generating devices 22 so as to allow the generation of an accurate spatial resolution 35 of each of the objects of interest 15 traveling in the product stream 14, and to further align the signal features of each of the image and sensor signals 23. The method includes still another step of predicting the presence 42 of the objects of interest 15, and possible defects 21, in the fused image and sensor signals 35 by applying an image model 43 which is previously formed from a multiplicity of image signals 35, and which are further acquired from the objects of interest 15 and defects 21, to the multiplicity of fused sensor signals 35 so as to facilitate the formation of a resulting object presence image signal 44, and a defect image signal 45. The method 10 includes another step of identifying the individual objects of interest 15, with the object presence and defect image signals 44 and 45, by identifying one or more of a group of pixels in each of the object presence and defect image signals 44 and 45, respectively. The method 10 of the present invention further includes another step of determining a spatial orientation 71 of the objects of interest 15 traveling in the product stream 14 by applying a prior source of knowledge 70 of the object aspects to a multiplicity of the object images which are formed of the object presence and defect image signals 44 and 45, respectively. The method includes yet another step of detecting defects 81 within the unacceptable objects of interest 15 by applying a prior source of knowledge 80 of defect aspects relative to object aspects, to the object images formed of the object presence and object defect signals 44 and 45, respectively. The method includes still another step of identifying the location 81 of the unacceptable objects of interest having defects in the object image signals; and yet another step 100 of removing the unacceptable objects of interest 15 having defects from the product stream 14 so as to provide a resulting uniform product stream.

[0025] In the methodology of the present invention 10 the discreet signal features of the multiplicity of synchronized image signals 23 are selected from the group comprising signals generated by any one or more of individual hyperspectral or multispectral imagers or scanners 22 which are employed in the apparatus and which are schematically represented in FIG. 1.

[0026] Still further the synchronized image signals 23 are formed by a methodology which includes a step of conducting a spatial registration of the respective image signals. In addition to the foregoing, the aspects and characteristics of the objects of interest 15, and which are deemed acceptable for further processing are selected from individual products 15 having a known, and acceptable qualities. Still further, the characteristics of the objects of interest 15 which are deemed unacceptable for further processing are selected from the group comprising individual products 15 having known unacceptable qualities. Moreover, the aligning of the signal features of each of the synchronized image signals 23 so as to form, at least in part, the multiplicity of fused image or sensor signals 35 comprises another step of conducting a spatial registration of the respective sensors 22 with each other, and with the ejector controller 100.

[0027] The methodology 10 of the present invention further includes yet other steps which are directed to the formation of the image model 43. In this regard, the image model 43 is formed by a methodology which includes a step of utilizing a standard classification algorithm such as a partial least square algorithm (PLS). In addition to the foregoing, the prior source of knowledge of the object aspects 70, which is supplied to the multiplicity of object images, and which are further used to determine the spatial orientation of the identified objects of interest 71 is formed by a methodology which includes the steps of conducting an object shape analysis; and conducting an object aspect measurement. Moreover, the prior source of knowledge of the defect aspects 80, and which are applied to the multiplicity of object images formed by the present methodology comprises the step of qualifying unacceptable pixel groups found in the image signals with object regions identified in object aspects. Additionally, the step of removing the unacceptable objects of interest 90 from the product stream 14 further comprises the step of removing an unacceptable portion of an object of interest 15, from an acceptable portion of the same object of interest 15.

[0028] As should be understood, and in the present methodology 10, and before the step of acquiring the multiplicity of synchronized image signals 23, the method 10 further includes a step of providing a product stream 14 of individual objects of interest 15 which have characteristics and aspects of both acceptable 20, and unacceptable objects of interest 21, and passing the product stream 14 having both the acceptable and unacceptable objects of interest 20 and 21 through an inspection station 25. In addition to the foregoing the methodology further includes, before the step of acquiring the multiplicity of synchronized image signals 23, and after the step of providing the product stream 14, generating a source of synchronized electromagnetic radiation 31 which is directed towards, and reflected at least in part from, the product stream 14 which is formed of the objects of interest 15 passing through the inspection station 25. In addition to the foregoing, the method of the present invention 10 includes yet another step of providing a first controller 24 which predicts the presence of the objects of interest 15, and defects 21 in the fused sensor signals 35, and which further applies the image model 43 to the multiplicity of fused sensor and image signals 35. The method 10 further includes still other steps of providing a second controller 50 which identifies individual objects of interest 15, and defects 21, in the product stream; determines the spatial orientation of the identified individual objects of interest 71 traveling in the product stream 14; detects the objects of interest 15; identifies the location of the defects 21 in the object presence and defect image signals 44 and 45, and further generates a signal 82 which indicates the presence and location of the defect(s) 21 in the product stream 14. The method 10 of the present invention includes yet another step of providing a defect removal station 26, and positioning the defect removal station downstream of the inspection station 25 (FIG. 1). The method includes another step of providing an ejector 101, and positioning the ejector 101 in the defect removal station 26, and which is effective, when made operational, to remove the unacceptable objects of interest 21 from the product stream 14. The method 10 includes yet another step of providing an ejector controller 100 which is controllably coupled with, and renders operational the ejector 101. A third controller 90 is coupled in defect signal 82 receiving relation relative to the second controller 50, and which further renders the ejector controller 100 operational to cause the ejector 101 to remove the unacceptable objects of interest 21 from the product stream 14 (FIG. 1). Therefore it will be seen that the present invention provides a convenient means for detecting unacceptable objects of interest in a product stream, and which may include agricultural products such as potatoes having agricultural defects such as sugar ends, in a particularly efficient fashion which has not been available in automated sorting devices utilized, heretofore. The present methodology is efficient, operates reliably, and further has an inventive feature which allows it to self-learn so as to increase the efficiency and reliability of the sorting operations of a device employing same.

[0029] In compliance with the statute, the invention has been described in language more or less specific as to structural, and methodical features. It is to be understood, however, that the invention is not limited to the specific features shown and described since the means herein disclosed comprise preferred forms of putting the invention into effect. The invention is, therefore, claimed in any of its forms or modifications within the proper scope of the appended claims appropriately interpreted in accordance with the Doctrine of Equivalence.