Image analysis – Pattern recognition – Feature extraction
Reexamination Certificate
1999-06-16
2004-04-20
Patel, Jayanti K. (Department: 2625)
Image analysis
Pattern recognition
Feature extraction
C382S282000, C382S224000, C382S218000, C345S597000, C345S215000, C348S263000
Reexamination Certificate
active
06724937
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a method and system for identifying image frames for printing, storing or flagging from a video shot comprising a series of video image frames.
BACKGROUND OF THE INVENTION
Current systems for identifying image frames suitable for printing are based on extraction of frames on the basis of scene changes, colour changes, and/or camera motion. These systems focus on extracting frames which represent different aspects of camera utilisation, or alternatively different scene attributes. U.S. Pat. No. 5,576,950, selects image frames on the basis of scene quality and matching value. A disadvantage of this approach is the high computational load imposed by the algorithms used for assessing image frame quality and matching value.
SUMMARY OF THE INVENTION
It is an object of the present invention to substantially overcome, or at least ameliorate, one or more disadvantages of the prior art.
Therefore the invention discloses a method for identifying one or more image frames from a series of video image frames for printing, storing or flagging, the method including the steps of:
a) calculating a gradient value for each image frame in the series, thereby to generate a raw data series;
b) smoothing data within the raw data series to generate a smoothed data series;
c) identifying one or more local maxima in the smoothed data series;
d) identifying one or more gradient value peaks in a corresponding range of the raw data series based on positions of respective local maxima; and
e) selecting the image frames corresponding to each of the identified gradient value peaks.
The invention further discloses a method of identifying one or more image frames from a series of image frames, the method including the steps of:
n) calculating a gradient value for each image frame in the series, thereby to generate a raw data series;
o) defining a gradient value threshold;
p) identifying sets of adjacent gradient values in the raw data series which exceed the gradient value threshold and are of similar value; and
q) selecting a representative image frame from each set of adjacent gradient values.
The invention yet further discloses a method for identifying one or more image frames from a series of video image frames suitable for printing, storing or flagging, the method including the steps of:
a) calculating a gradient value for each image frame in the series;
b) obtaining smoothed data based on the gradient value; and
c) selecting the image frames on the basis of the smoothed data.
The invention yet further discloses an apparatus for identifying one or more image frames from a series of video image frames for printing, storing or flagging, the apparatus comprising:
gradient calculation means arranged to receive a series of image frames and to calculate and output a raw data series of gradient values, one gradient value for each frame;
data smoothing means responsive to the raw data series and arranged to generate and output a smoothed data series;
identification means responsive to the smoothed data series and arranged to identify and output one or more local maxima in the smoothed data series;
gradient peak selection means responsive to the one or more local maxima and to the raw data series and arranged to identify one or more gradient value peaks in a corresponding range of the raw data series based upon positions of respective local maxima; and
image frame selection means responsive to the series of image frames and to the identified gradient value peaks, and arranged to select image frames corresponding to each of the identified gradient value peaks.
The invention yet further discloses an apparatus for identifying one or more image frames from a series of image frames, the apparatus comprising:
means for calculating a gradient value for each image frame in the series, thereby to generate a raw data series;
means for defining a gradient value threshold;
means for identifying sets of adjacent gradient values in the raw data series which exceed the gradient value threshold and are of similar value; and
means for selecting a representative image frame from each set of adjacent gradient values.
The invention yet further discloses an apparatus for identifying one or more image frames from a series of video images for printing, storing or flagging, the apparatus comprising:
gradient calculation means arranged to receive a series of image frames and to calculate and output a series of gradient values;
data smoothing means responsive to the series of gradient values for generating a smoothed data series; and
image frame selection means responsive to the smoothed data series for selecting image frames.
The invention yet further discloses a computer program product including a computer readable medium having recorded thereon a computer program for identifying one or more image frames from a series of video image frames for printing, storing or flagging, the computer program comprising:
gradient calculation process steps arranged to receive a series of image frames and to calculate and output a raw data series of gradient values, one gradient value for each frame;
data smoothing process steps responsive to the raw data series and arranged to generate and output a smoothed data series;
identification process steps responsive to the smoothed data series and arranged to identify and output one or more local maxima in the smoothed data series;
gradient peak selection process steps responsive to the one or more local maxima and to the raw data series and arranged to identify one or more gradient value peaks in a corresponding range of the raw data series based upon positions of respective local maxima; and
image frame selection process steps responsive to the series of image frames and to the identified gradient value peaks, and arranged to select image frames corresponding to each of the identified gradient value peaks.
The invention yet further discloses a computer program product including a computer readable medium having recorded thereon a computer program for identifying one or more image frames from a series of video image frames for printing, storing or flagging, the computer program comprising:
process steps for calculating a gradient value for each image frame in the series, thereby to generate a raw data series;
process steps for defining a gradient value threshold;
process steps for identifying sets of adjacent gradient values in the raw data series which exceed the gradient value threshold and are of similar value; and
process steps for selecting a representative image frame from each set of adjacent gradient values.
REFERENCES:
patent: 4379632 (1983-04-01), Dedden et al.
patent: 5576950 (1996-11-01), Tonomura et al.
patent: 5875040 (1999-02-01), Matraszek et al.
patent: 5903660 (1999-05-01), Huang et al.
Windle John Richard
Wu Jing
Kassa Yosef
Patel Jayanti K.
LandOfFree
Method of identifying image frames suitable for printing does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of identifying image frames suitable for printing, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of identifying image frames suitable for printing will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3255860