Multilevel thresholding for target tracking apparatus

Claim 1

A method for segmenting target information accompanying background information in a video signal, said method comprising the steps of: sampling pixels of possible target or background information; sampling pixels known to be, at least in predominant part, background information; establishing an average of the video signal intensities associated with said pixels known to be at least in predominant part background information; comparing each of said pixels known to be at least in predominant part background information, to the established average of their video signal intensities, for separating those of said pixels having video signal intensities greater than average from those of said pixels having video signal intensities less than average; accumulating statistics concerning said pixels known to be at least in predominant part background information having intensities greater than average and establishing from these statistics, without recourse to statistics concerning the intensities of target information, a range of normally expected intensities associated with high-intensity background information pixels; accumulating statistics concerning said pixels known to be at least in predominant part background information having intensities less than average and establishing from these statistics, without recourse to statistics concerning the intensities of target information, a range of normally expected intensities associated with low-intensity background information pixels; and determining whether each of said pixels of possible target or background information falls outside both said ranges of normally expected intensities associated with background information pixels to establish a likelihood of its sampling target information rather than background information.

Claim 2

A method for segmenting target information accompanying background information in a video signal, said method comprising the steps of: sampling pixels of possible target or background information; sampling pixels known to be, at least in predominant part, background information; establishing an average of the video signal intensities associated with said pixels known to be at least in predominant part background information; comparing each of said pixels known to be at least in predominant part background information, to the established average of their video signal intensities, for separating those of said pixels having video signal intensities greater than average from those of said pixels having video signal intensities less than average; accumulating statistics concerning said pixels known to be at least in predominant part background information having intensities greater than average and establishing from these statistics, without recourse to statistics concerning the intensities of target information, a range of normally expected intensities associated with high-intensity background information pixels; accumulating statistics concerning said pixels known to be at least in predominant part background information having intensities less than average and establishing from these statistics, without recourse to statistics concerning the intensities of target information, a range of normally expected intensities associated with low-intensity background information pixels; and determining whether each of said pixels of possible target or background information falls outside both said ranges of normally expected intensities associated with background information pixels to establish a likelihood of its sampling target information rather than background information.

Claim 4

A method for improving image segmentation in video signals containing target information, as used in target tracking apparatus, said method comprising the steps of in each of successive field scans: averaging the values of video signal falling within a target window to establish a first average; averaging the values of video signal falling within a background surround enclosing said target window to establish a second average; determining a third average from video signal values within said target window which are greater than said first average; determining a fourth average from video signal values within said target window which are less than said first average; determining a fifth average from video signal values within said background surround which are greater than said second average; determining a sixth average from video signal values within said background surround which are less than said second average; establishing ranges of expected deviation from each of said third, fourth, fifth and sixth averages of the signals used in their respective determination; setting a first threshold level at a predetermined fraction of the distance between the limits of the ranges around said third and fifth averages in the direction of increased video signal amplitude; setting a second threshold level at a predetermined fraction of the distance between the limits of the ranges around said third and fifth averages in the direction of decreased video signal amplitude; setting a third threshold level at a predetermined fraction of the distance between the limits of the ranges around said fourth and sixth averages in the direction of increased video signal amplitude; setting a fourth threshold level at a predetermined fraction of the distance between the limits of the ranges around said fourth and sixth averages in the direction of decreased video signal amplitude; identifying portions of video signal which are within said target window as target if those video signal portions are between said first and second threshold levels or between said third and fourth threshold levels, if the range around the fifth average falls within the range around the third average, and if the range around the sixth average falls within the range around the fourth average; and defining the target window for the next field scan and its background surround by prediction from the positioning of those portions of the video signal identified as target.

Claim 5

A method for improving image segmentation in video signals containing target information, as used in target tracking apparatus, said method comprising the steps of in each of successive field scans: averaging the values of video signal falling within a target window to establish a first average; averaging the values of video signal falling within a background surround enclosing said target window to establish a second average; determining a third average from video signal values within said target window which are greater than said first average; determining a fourth average from video signal values within said target window which are less than said first average; determining a fifth average from video signal values within said background surround which are greater than said second average; determining a sixth average from video signal values within said background surround which are less than said second average; establishing ranges of expected deviation from each of said third, fourth, fifth and sixth averages of the signals used in their respective determination; setting a first threshold level at a predetermined fraction of the distance between the limits of the ranges around said third and fifth averages in the direction of increased video signal amplitude; setting a second threshold level at a predetermined fraction of the distance between the limits of the ranges around said third and fifth averages in the direction of decreased video signal amplitude; setting a third threshold level at a predetermined fraction of the distance between the limits of the ranges around said fourth and sixth averages in the direction of increased video signal amplitude; setting a fourth threshold level at a predetermined fraction of the distance between the limits of the ranges around said fourth and sixth averages in the direction of decreased video signal amplitude; identifying portions of video signal which are within said target window as target if those video signal portions are between said first and second threshold levels or between said third and fourth threshold levels, if the range around the fifth average falls within the range around the third average, and if the range around the sixth average falls within the range around the fourth average; and defining the target window for the next field scan and its background surround by prediction from the positioning of those portions of the video signal identified as target.

Claim 6

A method for segmenting target information accompanying background information in a video signal, said method comprising the steps of: sampling pixels of possible target or background information; sampling pixels known to be, at least in predominant part, background information; accumulating statistics concerning the density of distribution of the amplitudes of said background information; establishing boundaries of ranges containing amplitudes of said background information in localized peak portions of that distribution, in a plural thresholding technique not depending upon statistics concerning the intensities of target information pixels; and identifying, as having a likelihood of being target information, each of said pixels of possible target or background information which fall outside all these ranges.

Claim 7

A method for segmenting target information accompanying background information in a video signal, said method comprising the steps of: sampling pixels of possible target or background information; sampling pixels known to be, at least in predominant part, background information; accumulating statistics concerning the density of distribution of the amplitudes of said background information; establishing boundaries of ranges containing amplitudes of said background information in localized peak portions of that distribution, in a plural thresholding technique not depending upon statistics concerning the intensities of target information pixels; and identifying, as having a likelihood of being target information, each of said pixels of possible target or background information which fall outside all these ranges.