EP0685814A2
Expired / Lapsed
EP
⏱️ 0.0 years remaining
Produce recognition system
Assignee: International Business Machines Corp
•
Filing: 1995-02-24
•
Granted: —
•
Expires: 2015-02-24
Abstract
Abstract The present system and apparatus uses image processing to recognize objects within a scene. The system includes an illumination source for illuminating the scene. By controlling the illumination source, an image processing system can take a first digitize image of the scene with the object illuminated a higher level and a second digitized image with the object illuminated at a lower level. Using an algorithm, the object(s) image is segmented from a background image of the scene by a comparison of the two digitized images taken. A processed image (that can be used to characterize features) of the object(s) is then compared to stored reference images. The object is recognized when a match occurs. The system can recognize objects independent of size and number and can be trained to recognize objects that it was not originally programmed to recognize.
Inventors
- 👤 Rudolf Maarten Bolle
- 👤 Jonathan Hudson Connell
- 👤 Norman Haas
- 👤 Rakesh Mohan
- 👤 Gabriel Taubin
Bibliographic Details
Application Number
—
Jurisdiction
EP
Data Source
google_patents
Last Synchronized
2026-11-09 22:22
CPC Classifications
G; G06; G06V; G06V10/00; G06V10/70; G06V10/764; G07; G07G; G07G1/00; G07G1/0036; G07G1/0045; G07G1/0054; G06V10/20; G06V10/32; G06V10/40; G06V10/50; G06V10/507; G06V10/56; G06V20/00; G06V20/60; G06V20/68
Claims Analysis
Total Claims: 24
Claim 1
Independent Claim
162 words
A system for recognizing objects comprising: a) a light source for illuminating the object, the light source having a non-monochromatic light frequency distribution that is constant over a period of time; b) a computer system having a visual input device for creating an scene image including a target object image and a background image, a memory storage, and an algorithm for segmenting an object image from a background image in the scene image; c) a plurality of reference normalized characterizations, each of a feature associated with a segmented reference object, the reference characterizations being stored in the computer memory storage; and d) one or more target normalized characterizations, each of a feature of a segmented target object image, the segmented target object image segmented from the scene by the algorithm, whereby a target normalized characterization is compared with one or more reference normalized characterizations and the target object is recognized as the associated reference object if the target and reference characterizations match.
Claim 2
Dependent Claim
11 words
A system as in claim 1, where the feature is hue.
Claim 3
Dependent Claim
19 words
A system as in claim 1, where the characterizations are histogram, or where the characterization is a hue histogram.
Claim 4
Independent Claim
228 words
A system for recognizing objects comprising: a) a light source for illuminating a scene, the light source having a non-monochromatic light frequency distribution that is constant over a period of time, the light source controlled to illuminate the scene at a first illumination level and at a second illumination level lower than the first illumination level, the scene comprising the object and a background; b) a computer system having a memory storage, a visual input device for creating a scene image including a target object image and a background image, and an algorithm that segments a target object image from a background image in the scene image, the algorithm segmenting the object image by comparing a scene image taken at the first illumination level with a scene image taken at a second illumination level; c) a plurality of reference normalized histograms, each of a feature associated with a segmented reference object, the reference histograms being stored in the computer memory storage; and d) one or more target normalized histograms, each of a feature of a segmented target object image, the target normalized histogram normalized the same way as the reference normalized histograms, whereby a target normalized histogram is compared with one or more reference normalized histograms and the target object is recognized as the associated reference object if the target and reference normalized histograms of a feature match.
Claim 5
Dependent Claim
27 words
A system, as in claim 4, where the normalization is area normalization or length normalization, or with respect to a measure extracted from the segmented object image.
Claim 6
Dependent Claim
31 words
A system, as in claim 4, where the feature is hue and the hue feature is area normalized, or where the feature is saturation and the saturation histogram is area normalized.
Claim 7
Dependent Claim
36 words
A system, as in claim 6, where a second feature is saturation and a target normalized saturation histogram of the segmented image also has to match a reference normalized saturation histogram before the object is recognized.
Claim 8
Dependent Claim
32 words
A system, as in claim 4, where the feature is texture and whereby a target texture histogram must match a reference texture histogram in order to the target object to be recognized.
Claim 9
Dependent Claim
17 words
A system, as in claim 8, where texture is determined using region calculations, or using edge calculations.
Claim 10
Dependent Claim
32 words
A system, as in claim 5, where the feature is shape and whereby a target shape histogram must match a reference shape histogram in order for the target object to be recognized.
Claim 11
Dependent Claim
67 words
A system, as in claim 4, further comprising: a) a scale that weighs the target object; b) a reference segmented object average projected density of the object, a representation of the reference projected density being stored in computer memory storage; and c) a target segmented object projected density; whereby the target object projected density must match the reference projected density in order to identify the target object.
Claim 12
Dependent Claim
22 words
A system, as in claim 11, where the average projected density is determined by dividing the object weight by the object area.
Claim 13
Dependent Claim
20 words
A system, as in claim 4, where both the target and reference object images are obtained through a polarizing filter.
Claim 14
Dependent Claim
61 words
A system, as in claim 4, where the target object has two or more target region features each representing a distinct region on the target object, where each target region feature histogram and the relative positions of the region feature match a respective reference region feature histogram stored in computer memory storage in order for the target object to be recognized.
Claim 15
Dependent Claim
37 words
A system, as in claim 14, where the target object region features are in a relative position and the relative position has to match a stored relative position in order for the target object to be recognized.
Claim 16
Dependent Claim
60 words
A system, as in claim 4, where the area of the target object is determined by removing the background from the object image by taking a first image without the object and a second image with the object and subtracting the first image from the second image at pixel locations where the first image is equal to the second image.
Claim 17
Dependent Claim
52 words
A system, as in claim 4, where the first image is taken when the light source is off and the second image is taken when the light source is on and the target object image is segmented by selecting the pixels that are dark in the first brighter in the second image.
Claim 18
Dependent Claim
38 words
A system, as in claim 5, where the objects are bulk items and the video input device is a color video camera, or where the objects are produce and the video input device is a color video camera.
Claim 19
Independent Claim
263 words
A system for recognizing objects comprising: a) a light source for illuminating a scene, the light source having a non-monochromatic light frequency distribution that is constant over a period of time, the light source controlled to illuminate the scene at a first illumination level and at a second illumination level lower than the first illumination level, the scene comprising the object and a background; b) a computer system having a memory storage, a visual input device for creating a scene image including a target object image and a background image, and an algorithm that segments the target object image from the background image in the scene image, the algorithm segmenting the object image by comparing a scene image taken at the first illumination level with a scene image taken at a second illumination level; c) a plurality of reference normalized histograms, each of a feature associated with a segmented reference object, the reference histograms being stored in the computer memory storage; and d) one or more target normalized histograms, each of a feature of a segmented target object image, the target normalized histogram normalized the same way as the reference normalized histograms; e) a means for determining if an unrecognized target object image meets a set of storage criteria; whereby a target normalized histogram is compared with one or more reference normalized histograms and the target object is not recognized as the associated reference object because the target and reference histograms of a feature do not match and the target normalized histogram is stored in memory storage if it meets the storage criteria.
Claim 20
Dependent Claim
17 words
A system, as in claim 19, that identifies the object to a user through a user interface.
Claim 21
Dependent Claim
33 words
A system, as in claim 20, where the user interface gives the user a selection of two or more possible identities of the object, or where the user interface is a touch screen.
Claim 22
Dependent Claim
26 words
A system, as in claim 20, where the interface is a voice recognition system, or where the interface enables the user to browse through object selections.
Claim 23
Dependent Claim
32 words
A system, as in claim 19, further comprising: a scale that weighs the object, whereby the price of the bulk item is determined based on the weight and recognition of the object.
Claim 24
Independent Claim
153 words
A method for recognizing objects comprising the steps of: a) illuminating the object with a light source, the light source having a non-monochromatic light frequency distribution that is constant over a period of time; b) creating a scene image with a computer system having a visual input device, the scene image comprising a target object image and a background image, the computer system having a memory storage an algorithm for segmenting the object image from the background image of the scene; c) comparing one or more target normalized characterizations, each of a feature of a segmented target object image determined by the algorithm to one or more reference normalized characterization in memory storage, each reference characterization of a feature associated with a segmented reference object; and d) recognizing the target object as a reference object when one or more of the target normalized characterizations matches one or more of the reference normalized characterizations.
References Cited (Backward Citations)
No citation records indexed.
USPTO Annuity & Maintenance Windows
Maintenance schedule is calculated for granted US utility patents.