Image analysis by object addition and recovery

• US 7,940,959 B2
• Filed: 09/10/2007
• Issued: 05/10/2011
• Est. Priority Date: 09/08/2006
• Status: Active Grant

• Current Assignee: Image Insight, Inc.
• Original Assignee: Advanced Fuel Research, Inc.
• Inventors: Rubenstein, Eric P.
• Primary Examiner(s): AZARIAN, SEYED H
• Application Number:
  US11/852,614
• Publication Number:
  US 20080063237A1
• Time in Patent Office: 1,338 Days
• Field of Search: 382/106, 382/107, 382/155, 382/168, 382/181, 382/199, 382/203, 382/224, 382/232, 382/254, 382/274, 382/276, 382286-291, 382/305, 382/312, 382/103, 700/83, 700/94, 348/155, 348/143, 348/349, 715/716
• US Class Current: 382/103
• CPC Class Codes: G06V 20/13 Satellite images

Abstract

The invention described herein is generally directed to methods for analyzing an image. In particular, crowded field images may be analyzed for unidentified, unobserved objects based on an iterative analysis of modified images including artificial objects or removed real objects. The results can provide an estimate of the completeness of analysis of the image, an estimate of the number of objects that are unobserved in the image, and an assessment of the quality of other similar images.

Claims

1. A method for detecting an object in an image comprising:

obtaining one or more real images;

analyzing the one or more real images and deriving a set of parameters defining each of the one or more real images;

creating a modified image by;

inserting at least one artificial object into at least one of the one or more real images, wherein the at least one artificial object comprises at least a portion of an object identified in the one or more real images, at least a portion of a known object, at least portion of an archived object, or a combination thereof;

removing at least one object in at least one of the one or more real images;

or

a combination of inserting and removing;

analyzing the modified image to derive a set of parameters defining the modified image;

comparing the set of parameters derived from the modified image with the set of parameters derived from the one or more real images;

determining the likelihood of additional objects being present within the one or more real images;

repeating the steps of creating, analyzing, comparing and determining;

stopping the steps of creating, analyzing, comparing and determining when the probability of additional, unidentified objects being present in the one or more real images has been calculated to a predetermined level of precision; and

generating a report describing compiled results of repeating the steps of inserting, analyzing, comparing and determining.

2. The method of claim 1, wherein the one or more real images are selected from digital images, analog images and a combination thereof obtained from aerial still images, aerial video images, satellite still images, satellite video images, ground to ground still images, ground to ground video images, ground to air still images, ground to air video images, ground to water still images, ground to water video images, water to ground still images, water to ground video images, water to water still images, water to water video images, air to air still images, air to air video images or a combination thereof.

3. The method of claim 1, wherein the area encompassed by the one or more image is selected from a terrestrial area, an aquatic area, an aerial area, an extraterrestrial area or a combination thereof.

4. The method of claim 1, wherein the step of analyzing further comprises digitizing the one or more real images, digitizing the modified image and combination thereof.

5. The method of claim 1, wherein the step of analyzing further comprises creating a plurality of image chips, wherein each of the plurality of image chips comprise a portion of the one or more real images or modified images and the step of analyzing comprises parsing the plurality of image chips at least into one or more groups of image chips having objects and one or more groups of image chips not having objects.

6. The methods of claim 1, wherein the step of analyzing the one or more real images and the step of analyzing the modified image are carried out using the same process.

7. The method of claim 1, wherein the steps of obtaining, inserting, analyzing or a combination thereof further comprise enhancing at least a portion of the one or more real images or modified image by a method selected from adjusting contrast, adjusting color, image extraction, collaging, registering, coadding, averaging, median-filter combining, sigma-clip averaging, splicing, histogram matching, mosaicking, convolution filtering, deconvolution filtering, unsharp-masking, edge detection, Fourier transformation, reducing background noise, texture processing, creating and using multispectral data, creating or using hyperspectral data and a combination thereof.

8. The method of claim 1, wherein the step of analyzing the one or more real image, the modified image or a combination thereof and deriving a set of parameters further comprises determining a data value for parameters selected from a number of objects, brightness, contrast, color, shape, orientation, location and a combination thereof.

9. The method of claim 1, further comprising identifying at least a portion of an object.

10. The method of claim 9, wherein the step of identifying at least a portion of an object further comprises determining a total number of objects in an image or a portion of an image.

11. The method of claim 9, wherein the step of identifying objects further comprises comparing the identified object to known objects to identify a type of object.

12. The method of claim 9, wherein the step of identifying objects occurs concurrently with or following the step of analyzing.

13. The method of claim 1, wherein the at least a portion of the at least one artificial object comprises a set of parameters selected from brightness, contrast, color, shape, size, texture, reflectivity, reflection, orientation, location and a combination thereof;

wherein all or a subset of parameters are substantially the same as a derived set of parameters defining at least a portion of one or more real image.

14. The method of claim 13, wherein the step of inserting at least one artificial object further comprises modifying the at least one artificial object using a method selected from adding shadows, adding one or more obscuration, adding a layover, performing a multipath, adding indirect illuminations, adding reduced/partial transparency effect, simulating camouflage, simulating netting, simulating vegetation, simulating ground covering, simulating water covering, simulating cloud covering, simulating weather, convolving with a nearby object and a combination thereof.

15. The method of claim 1, wherein the step of inserting at least one artificial object further comprises rotating, scaling, shearing, smearing, convoluting, convolving, degrading, or a combination thereof of the at least one artificial object such that the at least one artificial object is consistent within the context of the image.

16. The method of claim 1, further comprising recovering an optical characteristic from the at least one modified image wherein the optical characteristic comprise a property of an object selected from brightness, color, location, orientation, reflectivity, a probability of recovering an optical characteristic, an uncertainty for the optical characteristic recovered and a combination thereof.

17. The method of claim 16, further comprising applying the recovered optical characteristic to one or more objects of the one or more real images, applying the recovered optical characteristic to one or more archived object in a library, applying the recovered optical characteristic to one or more real or modified images being analyzed, applying the recovered optical characteristic to one or more real or modified images that are subsequently analyzed, or a combination thereof.

18. The method of claim 1, further comprising determining a confidence level for the one or more real images wherein the confidence level is the probability of a number of objects in the one or more original images being identified.

19. The method of claim 1, wherein an identified object, artificial object or combination thereof is selected from a human, an animal, a building, a machine, a geological formation, a type of plant, an aquatic feature, an aerial feature, an airplane or airplanes, an extraterrestrial feature, a vehicle, a military implement, an artillery installation, a tank and a combination thereof.

20. The method of claim 1, wherein the method is automated.

21. A method for detecting an object in an image comprising:

obtaining one or more real images;

analyzing the one or more real, images and deriving a set of parameters defining each of the one or more real images;

inserting at, least one artificial object into at least one of the one or more real images to create a modified image, wherein the at least one artificial object comprises at least a portion of an object identified in the one or more real images, at least a portion of a known object, at least portion of an archived object, or a combination thereof;

analyzing the modified image to derive a set of parameters defining the modified image;

comparing the set of parameters, derived from the modified image with the set of parameters derived from the one or more real images;

determining the likelihood of additional objects being present within the one or more real images;

repeating the steps of inserting, analyzing, comparing and determining;

stopping the steps of inserting, analyzing, comparing and determining when the probability of additional, unidentified objects being present in the one or more real images has been calculated to a predetermined level of precision; and

generating a report describing compiled results of repeating, the steps of inserting, analyzing, comparing and determining.

22. A method for detecting an object in an image comprising:

obtaining one or more real images;

analyzing the one or more real images and deriving a set of parameters defining each of the one or more real images;

removing at least one object in at least one of the one or more real images to create a modified image;

analyzing the modified image to derive a set of parameters defining the modified image;

comparing the set of parameters derived from the modified image with the set of parameters derived from the one or more real images;

determining the likelihood of additional objects being present within the one or more real images;

repeating the steps of, removing, analyzing, comparing and determining;

stopping the steps of removing, analyzing, comparing and determining when the probability of additional, unidentified objects being present in the one or more real images has been calculated to a predetermined level of precision; and

generating a report describing compiled results of repeating, the steps of removing, analyzing, comparing and determining.

23. A system comprising:

at least one detector for acquiring at least one real image;

a processor configured to analyze images and images and generate a set parameters in communication with the at least one detector;

a processor configured to create at least one modified image by inserting at least one artificial object into the at least one real image, wherein the at least one artificial object comprises at least a portion of an object identified in the one or more real images, at least a portion of a known object, at least portion of an archived object, or a combination thereof;

removing at least one object from at least one of real image or a combination thereof in communication with the processor for analyzing images;

a processor configured to compare one or more sets of parameters generated for the at least one real image and one or more sets of parameters generated for at least on modified image in communication with the processor for analyzing images;

a processor configured determine whether enough modified images have been created in communication with the processor for comparing sets of parameters and the processor for creating modified images; and

an output device.

24. A computer readable medium storing instructions algorithm in computer readable form for;

analyzing the one or more real images and generating a set of parameters defining the one or more real images;

creating a modified image by inserting at least one artificial object into at least one real image, wherein the at least one artificial object comprises at least a portion of an object identified in the one or more real images, at least a portion of a known object, at least portion of an archived object, or a combination thereof, removing at least one object from at least one of real image or a combination thereof;

analyzing the modified image and generating a set of parameters defining the one or more modified images;

comparing the parameters defining the modified image with the parameters defining the one or more real images;

determining the probability of at least one additional object being present in the image;

determining the uncertainty in at least one parameter;

repeating the steps of creating, analyzing, comparing and determining; and

stopping the steps of creating, analyzing mid comparing when it is determined that the probability of at least one additional object being present at one or more locations on the one or more real images, and/or the uncertainty in the at least one parameter has reached a threshold confidence level.