Method and apparatus for image acquisition with adaptive compensation for image exposure variation

• US 5,751,844 A
• Filed: 01/29/1996
• Issued: 05/12/1998
• Est. Priority Date: 04/20/1992
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Bolin, Mark Robert, Kishi, Gregory Tad, McIntosh, Michael Philip
• Primary Examiner(s): Boudreau, Leo
• Assistant Examiner(s): KELLEY, CHRISTOPHER S
• Application Number:
  US08/593,538
• Time in Patent Office: 834 Days
• Field of Search: 382/156, 382/254, 382/274, 348/221, 348/229, 348/362, 354/410, 354/412, 354/485, 395/21, 395/22, 395/23, 396/49
• US Class Current: 382/156
• CPC Class Codes: G06T 2207/10144 Varying exposure
• CPC Class Codes: G06T 2207/20084 Artificial neural networks [ANN]
• CPC Class Codes: G06T 5/40 using histogram techniques
• CPC Class Codes: G06T 5/60 using machine learning, e.g. neural networks
• CPC Class Codes: G06T 5/90 Dynamic range modification of images or parts thereof
• CPC Class Codes: H04N 23/64 Computer-aided capture of images, e.g. transfer from script file into camera, check of taken image quality, advice or proposal for image composition or decision on when to take image
• CPC Class Codes: H04N 23/70 Circuitry for compensating brightness variation in the scene
• CPC Class Codes: H04N 23/71 Circuitry for evaluating the brightness variation

Abstract

In an image acquisition system, which produces at least one image, each of which is scorable with reference to at least one image quality criterion, a control system is provided for optimizing the image quality criterion. Typically, the image quality criterion is at least in-part established by an exposure parameter. When considered as an apparatus, the present invention includes a number of components which cooperate together to automatically and continually adjust the value of the exposure parameter to optimize the image quality criterion. An image acquisition means is provided to obtain one or more acquired images under selected exposure characteristics. A transform system is provided for receiving the one or more acquired images and developing an energy distribution map, or histogram, of at least a portion of the one or more acquired images. A neural network means is provided for maintaining a learned relationship between the image quality criterion and the exposure parameter, and for receiving the energy distribution map of the one or more acquired images, and for automatically providing a corrected exposure parameter, so that subsequent acquired images will be obtained under optimal settings of the exposure parameter. A controller member is provided for supplying corrected exposure parameters, which are the output of the neural network means, to the image acquisition means.

Claims

1. In an image acquisition device which produces at least one acquired image, each of which is characterizeable with reference to at least one image quality criterion which is at least in-part established by an exposure parameter value, an apparatus for optimizing said image quality criterion, comprising:

an image acquisition means for obtaining at least one image under selected exposure characteristics as determined by at least one exposure parameter value;

a neural network means for (a) maintaining a learned relationship between said at least one image quality criterion and said at least one exposure parameter value which is developed from training data during a training data acquisition mode of operation from image acquisition over a range of values for said exposure parameter values, and (b) receiving said at least one image quality criterion of said at least one image from said image acquisition means and providing a correction for said at least one exposure parameter value without modification of said learned relationship; and

a feed forward controller member for supplying said correction for said at least one exposure parameter value to said image acquisition means for use in digitizing at least one subsequent image to compensate for changes in illumination of areas that are viewed repetitively.

2. An apparatus for optimizing said at least one image quality criterion according to claim 1, wherein said image acquisition means comprises at least one video camera.

3. An apparatus for optimizing said at least one image quality criterion according to claim 1, wherein said image acquisition means comprises at least one video camera, and wherein said neural network means maintains a learned relationship between said at least one image quality criterion and at least one video camera input variable which affects receipt of video images by said at least one video camera.

4. An apparatus for optimizing said at least one image quality criterion according to claim 1, wherein said image acquisition means comprises at least one video camera and wherein said neural network means maintains a learned relationship between said at least one image quality criterion and a plurality of video camera input variables, including at least one control parameter which affects digitization of analog video signals from said at least one video camera.

5. An apparatus for optimizing said at least one image quality criterion according to claim 4, wherein said at least one control parameter includes an offset parameter which defines digital video image zero.

6. An apparatus for optimizing said at least one image quality criterion according to claim 4, wherein said at least one control parameter includes a gain parameter which defines a maximum analog signal voltage span which is accepted for digitization without loss of information contained therein.

7. An apparatus for optimizing said at least one image quality criterion according to claim 1, further comprising:

transform means for developing an energy distribution mapping of at least a portion of said at least one image which provides said at least one image quality criterion by relating image quality of said portion of said at least one image to said image quality criterion.

8. An apparatus for optimizing said at least one image quality criterion according to claim 7, wherein said transform means develops at least one grayscale histogram of said at least one image.

9. In an image acquisition device which produces at least one acquired image, each of which is characterizeable with reference to at least one image quality criterion which is at least in-part established by an exposure parameter value, an apparatus for optimizing said image quality criterion, comprising:

an image acquisition means for obtaining a single image under selected exposure characteristics as determined by at least one exposure parameter value;

a neural network exposure parameter correction model which maintains an established relationship between said at least one image quality criterion and said at least one exposure parameter value which is developed from training data during a training data acquisition mode of operation from image acquisition over a range of values for said at least one exposure parameter value, for receiving said at least one image quality criterion of said single image from said image acquisition means and providing a correction for said at least one exposure parameter value during an image acquisition mode of operation without modification of said established relationships; and

a feed forward controller member for supplying said correction for said at least one exposure parameter value to said image acquisition means for use in digitizing at least one subsequent image to compensate for changes in illumination of areas that are viewed repetitively.

10. An apparatus for optimizing said at least one image quality criterion according to claim 9, wherein said single image comprises one region of interest of a selected image, and wherein said at least one subsequent image comprises other regions of interest of said selected image.

11. An apparatus for optimizing said at least one image quality criterion according to claim 9, wherein said neural network exposure parameter correction model comprises a software model which relates image quality as determined by grayscale value distribution of a single image to correction values for at least one exposure parameter value.

12. An apparatus for optimizing said at least one image quality criterion according to claim 11, wherein said software model comprises a feed forward neural network.

13. An apparatus for optimizing said at least one image quality criterion according to claim 12, wherein neural nodes in said feed forward neural network are connected by linkages which are weighted according to weight values developed during a training mode of operation.

14. A method of obtaining improved quality images in an image acquisition system, comprising the method steps of:

developing, during a training data acquisition mode of operation, training data from image acquisition over a range of image exposure parameter values;

developing a neural network software model from said training data of a relationship between said image quality criteria and said image exposure parameter values;

acquiring, during normal operations, a single image, with known image exposure parameter values;

determining an image quality criterion from said single image;

supplying said image quality criterion and said known image exposure parameter value as inputs to said neural network software model of said relationship between said image quality criterion and said image exposure parameter values;

developing, with said neural network software model, improved image exposure parameter values as an output; and

supplying said improved image exposure parameter values to said image acquisition system to compensate for changes in illumination of areas that are viewed repetitively when subsequent images are digitized.

15. In an image acquisition device for use with a magnetic tape library which produces at least one acquired image of machine readable labels affixed to magnetic tape cartridges in said magnetic tape library, each of which is characterizeable with reference to at least one image quality criterion which is at least in-part established by an exposure parameter value, an apparatus for optimizing said image quality criterion, comprising:

an image acquisition means for obtaining at least one image of selected ones of said machine readable labels under selected exposure characteristics as determined by at least one exposure parameter value;

a neural network means for (a) maintaining a learned relationship between said at least one image quality criterion and said at least one exposure parameter value which is developed from training data during a training data acquisition mode of operation from image acquisition over a range of values for said at least one exposure parameter values, and (b) receiving said at least one image quality criterion of said at least one image from said image acquisition means and providing a correction for said at least one exposure parameter value during an image acquisition mode of operation without modification of said learned relationship; and

a feed forward controller member for supplying said correction for said at least one exposure parameter value to said image acquisition means for use in digitizing at least one subsequent image of said selected ones of said machine readable labels to compensate for changes in illumination of areas that are viewed repetitively.

16. A method of obtaining improved quality images of machine readable labels affixed to magnetic tape cartridges in a magnetic tape library with an image acquisition system, comprising the method steps of:

developing, during a training data acquisition mode of operation, training data from image acquisition over a range of image exposure parameter values;

developing a neural network software model from said training data of a relationship between said image quality criteria and said image exposure parameter values;

acquiring, during normal operations, a single image of at least one machine readable label, with known image exposure parameter values;

determining an image quality criterion from said single image;

supplying said image quality criterion and said known image exposure parameter value as inputs to said neural network software model of said relationship between said image quality criterion and said image exposure parameter values;

developing, with said neural network software model, improved image exposure parameter values as an output; and

supplying said improved image exposure parameter values to said image acquisition system to compensate for changes in illumination of areas that are viewed repetitively when subsequent images of at least one machine-readable label are digitized.

17. A method of obtaining improved quality images of target objects in an image acquisition system, comprising the method steps of:

illuminating said target objects with an illumination source which fluctuates in intensity;

developing, during a training data acquisition mode of operation, training data from image acquisition over a predefined range of illumination intensities and corresponding image exposure parameter values;

developing a neural network software model from said training data of a relationship between said image quality criteria and said image exposure parameter values;

acquiring, during normal operations, a single image, with known image exposure parameter values;

determining an image quality criterion from said single image;

supplying said image quality criterion and said known image exposure parameter value as inputs to said neural network software model of said relationship between said image quality criterion and said image exposure parameter values;

developing, with said neural network software model, improved image exposure parameter values as an output; and

supplying said improved image exposure parameter values to said image acquisition system to compensate for changes in illumination of areas that are viewed repetitively when subsequent images are digitized.

18. An improved vision system for use with a magnetic tape library which produces at least one acquired image of machine readable labels affixed to magnetic tape cartridges in said magnetic tape library, each of which is characterizeable with reference to at least one image quality criterion which is at least in-part established by an exposure parameter value, comprising:

an illumination source providing illumination on said machine readable labels affixed to said magnetic tape cartridges which fluctuates in intensity;

an image acquisition means for obtaining at least one image of selected ones of said machine readable labels under selected illumination intensities and corresponding exposure characteristics as determined by at least one exposure parameter value;

a neural network means for (a) maintaining a learned relationship between said at least one image quality criterion and said at least one exposure parameter value which is developed from training data during a training data acquisition mode of operation from image acquisition over a range of values for said at least one exposure parameter values, and (b) receiving said at least one image quality criterion of said at least one image from said image acquisition means and providing a correction for said at least one exposure parameter value during an image acquisition mode of operation without modification of said learned relationship; and

a feed forward controller member for supplying said correction for said at least one exposure parameter value to said image acquisition means for use in digitizing at least one subsequent image of said selected ones of said machine readable labels to compensate for changes in illumination of areas that are viewed repetitively when subsequent images are digitized.

19. A method of obtaining improved quality images of machine readable labels affixed to magnetic tape cartridges in a magnetic tape library with an image acquisition system, comprising the method steps of:

developing, during a training data acquisition mode of operation, training data from image acquisition over a range of illumination intensities and corresponding image exposure parameter values;

developing a neural network software model from said training data of a relationship between said image quality criteria and said image exposure parameter values;

acquiring, during normal operations, a single image of at least one machine readable label, at a particular illumination intensity with known image exposure parameter values;

determining an image quality criterion from said single image;

supplying said image quality criterion and said known image exposure parameter value as inputs to said neural network software model of said relationship between said image quality criterion and said image exposure parameter values;

developing, with said neural network software model, improved image exposure parameter values as an output; and

supplying said improved image exposure parameter values to said image acquisition system to compensate for changes in illumination of areas that are viewed repetitively when subsequent images of at least one machine-readable label are digitized.