Method and system for characterizing noise
AssigneePicture Code
InventorJames D. Christian
Life of this patent
Estimating the term…
Abstract
Embodiments of the present invention provide a method and system for characterizing noise in a set of data based on an analysis of subsets of the data. For example, noise for an image can be characterized based on analyzing selected regions of the image. The characterization of the noise can be stored in a data structure that models the noise as jointly dependent on values of multiple parameters (pixel values for multiple channels). The noise can be further characterized for one or more wavelet subbands of the set of data.
Claims46
A method for characterizing noise in an image comprising:
- loading a digital image stored in a computer readable medium;
- selecting plurality of regions of the image to analyze;
- generating a set of subbands of the image for each color channel, wherein generating a set of subbands further comprises applying a wavelet transform to the image data for the one or more regions to generate one or more sets of subband coefficients;
- analyzing image data for the regions, wherein analyzing image data for the one or more regions further comprises:
- determining based on the one or more sets of subband coefficients a set of dispersion values that comprise a plurality of dispersion values for each subband of each color channel, wherein each dispersion value corresponds to a selected region, subband and color channel;
- determining based on the set of dispersion values a set of decay pattern values indicating how noise decays from higher to lower level subbands, wherein determining the set of decay pattern values comprises dividing a first dispersion value corresponding to a first region, first color channel and first wavelet level by a second dispersion value corresponding to the first region, the first color channel and a second wavelet level to determine a first normalized dispersion value;
- generating a noise profile for the image using the dispersion values, wherein the noise profile correlates noise in a color channel to color, luminance and frequency band;
- storing the noise profile in a computer readable memory.
A computer program product comprising a non-transitory computer readable medium storing a set of computer instructions executable by a computer processor, the set of computer instructions comprising instructions executable to:
- access a digital image stored in a memory;
- generate a set of subbands for each color channel by applying a wavelet transform to the image data for the one or more regions to generate one or more sets of subband coefficients;
- select image data to analyze for one or more regions of the image;
- analyze the image data for the one or more regions to determine based on the one or more sets of subband coefficients a set of dispersion values that comprise a plurality of dispersion values for each subband of each color channel, wherein each dispersion value corresponds to a selected region, subband and color channel;
- determine based on the set of dispersion values a set of decay pattern values indicating how noise decays from higher to lower level subbands, wherein determining the set of decay pattern values comprises dividing a first dispersion value corresponding to a first region, first color channel and first wavelet level by a second dispersion value corresponding to the first region, the first color channel and a second wavelet level to determine a first normalized dispersion value;
- generate a noise profile for the image using the dispersion values, wherein the noise profile, jointly correlates noise in a color channel to color and luminance and frequency band;
- and
- store the noise profile in a computer readable memory.
A system comprising a processor, a set of computer readable medium accessible by the processor and a set of computer instructions stored on the computer readable medium and executable by the computer processor, the set of computer instructions comprising instructions executable to:
- access a set of data stored in a memory containing values for a plurality of parameters for which noise is to be characterized;
- generate subbands of the set of data for each of the plurality of parameters by applying a wavelet transform to the subset data for the one or more subsets to generate one or more sets of subband coefficients;
- select subsets of data to analyze for each of the subbands;
- analyze the subset data for the one or more selected subsets to determine a set of dispersion values based on the one or more sets of subband coefficients comprising a plurality of dispersion values for each subband, wherein each dispersion value corresponds to a selected subset, subband and parameter for which noise is to be characterized;
- determine a set of decay pattern values based on the set of dispersion values for the one or more subsets that indicate how noise decays from higher to lower level subbands, wherein determining the set of decay pattern values comprises dividing a first dispersion value corresponding to a first subset, first parameter and first wavelet level by a second dispersion value corresponding to the first subset, the first parameter and a second wavelet level to determine a first normalized dispersion value;
- generate a noise profile using the dispersion values and decay pattern values, wherein the noise profile correlates noise to the one or more parameters;
- and
- store the noise profile in a computer readable memory.
Dependencies are inferred from the claim text.