Concept for determining a confidence/uncertainty measure for disparity measurement

An apparatus for determining a confidence measure measuring a confidence in a correctness of, or an uncertainty measure measuring a liability of incorrectness of, a measurement of a disparity of a predetermined position of a first picture relative to a second picture, wherein the apparatus comprises an electronic circuit or microprocessor configured to, or a computer programmed to,

• accumulate, for each of a set of disparity candidates for the predetermined position, a contribution value which depends on the respective disparity candidate and a dissimilarity to the second picture which is associated with the respective disparity candidate according to a function which
• comprises a first monotonicity with the dissimilarity associated with the respective disparity candidate, and
• comprises a second monotonicity, opposite to the first monotonicity, with an absolute difference between the respective disparity candidate and a predetermined disparity having a minimum dissimilarity associated therewith among dissimilarities associated with the set of disparity candidates,
• wherein the first monotonicity is a monotonic increase and the second monotonicity is a monotonic decrease and the apparatus is configured to subject a result of the accumulation to a monotonically decreasing function to acquire the confidence measure, or use the result of the accumulation as the uncertainty measure, or
• wherein the first monotonicity is a monotonic decrease and the second monotonicity is a monotonic increase and the apparatus is configured to subject a result of the accumulation to a monotonically decreasing function to acquire the uncertainty measure, or use the result of the accumulation as the confidence measure.

An apparatus for determining a confidence measure measuring a confidence in a correctness of, or an uncertainty measure measuring a liability of incorrectness of, a measurement of a disparity of a predetermined position of a first picture relative to a second picture, wherein the apparatus comprises an electronic circuit or microprocessor configured to, or a computer programmed to, calculate the confidence measure

• by accumulating, for each of a set of disparity candidates for the predetermined position, a contribution value which depends on the respective disparity candidate and a dissimilarity to the second picture which is associated with the respective disparity candidate, so that the confidence measure C(x,y) is
• C
• ⁡
• (
• x
• ,
• y
• )
• =
• 1
• ∑
• d
• =
• d
• min
• d
• max
• ⁢
• max
• ⁡
• (
• min
• ⁡
• (
• Δ
• ⁢
• ⁢
• d
• ⁡
• (
• x
• ,
• y
• ,
• d
• )
• -
• 1
• ,
• D
• 3
• )
• ,
• 0
• )
• 2
• max
• ⁡
• (
• Δ
• ⁢
• ⁢
• c
• ⁡
• (
• x
• ,
• y
• ,
• d
• )
• -
• c
• max
• ⁡
• (
• x
• ,
• y
• )
• 5
• ,
• 1
• )
• ,
• where
• (x,y) is the predetermined position, d denotes a disparity candidate,
• max
• ⁡
• (
• min
• ⁡
• (
• Δ
• ⁢
• ⁢
• d
• ⁡
• (
• x
• ,
• y
• ,
• d
• )
• -
• 1
• ,
• D
• 3
• )
• ,
• 0
• )
• 2
• max
• ⁡
• (
• Δ
• ⁢
• ⁢
• c
• ⁡
• (
• x
• ,
• y
• ,
• d
• )
• -
• c
• max
• ⁡
• (
• x
• ,
• y
• )
• 5
• ,
• 1
• )
• is the contribution value for the disparity candidate d,
• dmin and dmax are a lower and upper bounds of an interval within which the set of disparity candidates are spread,
• D=dmax−dmin,
• cmax is a maximum dissimilarity among the dissimilarities associated with the set of disparity candidates,
• Δc(x,y,d)=c(d(x,y))−c1(x,y) with c(d) being the dissimilarity associated with the disparity candidate d, and c1 being the minimum dissimilarity,
• Δd=|d−d1| with d1 being the predetermined disparity having the minimum dissimilarity associated therewith among dissimilarities associated with the set of disparity candidates.

A system for generating a depth/disparity map for first picture relative to a second picture, comprising

• a determinator for determining a disparity of a predetermined position of the first picture relative to the second picture, and
• an apparatus for determining a confidence measure measuring a confidence in the correctness of, or an uncertainty measure measuring a liability of incorrectness of, a measurement of the disparity of the predetermined position of the first picture relative to the second picture by accumulating, for each of a set of disparity candidates for the predetermined position, a contribution value which depends on the respective disparity candidate and a dissimilarity to the second picture which is associated with the respective disparity candidate according to a function which
• comprises a first monotonicity with the dissimilarity associated with the respective disparity candidate, and
• comprises a second monotonicity, opposite to the first monotonicity, with an absolute difference between the respective disparity candidate and a predetermined disparity having a minimum dissimilarity associated therewith among dissimilarities associated with the set of disparity candidates,
• wherein the determinator is configured to determine the predetermined disparity having a minimum dissimilarity associated therewith among the dissimilarities associated with the set of disparity candidates as the disparity and
• wherein at least one of the determinator and the apparatus is implemented by an electronic circuit or microprocessor, or programmed computer.

A method for determining a confidence measure measuring a confidence in a correctness of, or an uncertainty measure measuring a liability of incorrectness of, a measurement of a disparity of a predetermined position of a first picture relative to a second picture, comprising

• accumulating, for each of a set of disparity candidates for the predetermined position, a contribution value which depends on the respective disparity candidate and a dissimilarity to the second picture which is associated with the respective disparity candidate according to a function which
• comprises a first monotonicity with the dissimilarity associated with the respective disparity candidate, and
• comprises a second monotonicity, opposite to the first monotonicity, with an absolute difference between the respective disparity candidate and a predetermined disparity having a minimum dissimilarity associated therewith among dissimilarities associated with the set of disparity candidates, and
• computing the contribution value as a product comprising a first factor and a second factor,
• the first factor comprises the first monotonicity with the dissimilarity associated with the respective disparity candidate, and
• the second factor comprises the second monotonicity with the absolute difference between the respective disparity candidate and the predetermined disparity.

A method for determining a confidence measure measuring a confidence in a correctness of, or an uncertainty measure measuring a liability of incorrectness of, a measurement of a disparity of a predetermined position of a first picture relative to a second picture, comprising

• accumulating, for each of a set of disparity candidates for the predetermined position, a contribution value which depends on the respective disparity candidate and a dissimilarity to the second picture which is associated with the respective disparity candidate according to a function which
• comprises a first monotonicity with the dissimilarity associated with the respective disparity candidate, and
• comprises a second monotonicity, opposite to the first monotonicity, with an absolute difference between the respective disparity candidate and a predetermined disparity having a minimum dissimilarity associated therewith among dissimilarities associated with the set of disparity candidates,
• wherein the first monotonicity is a monotonic increase and the second monotonicity is a monotonic decrease and a result of the accumulation is subject to a monotonically decreasing function to acquire the confidence measure, or the result of the accumulation is used as the uncertainty measure, or
• wherein the first monotonicity is a monotonic decrease and the second monotonicity is a monotonic increase and a result of the accumulation is subject to a monotonically decreasing function to acquire the uncertainty measure, or the result of the accumulation is used as the confidence measure.

A method for determining a confidence measure measuring a confidence in a correctness of, or an uncertainty measure measuring a liability of incorrectness of, a measurement of a disparity of a predetermined position of a first picture relative to a second picture, comprising

• calculating the confidence measure by accumulating, for each of a set of disparity candidates for the predetermined position, a contribution value which depends on the respective disparity candidate and a dissimilarity to the second picture which is associated with the respective disparity candidate, and so that the confidence measure C(x,y) is
• C
• ⁡
• (
• x
• ,
• y
• )
• =
• 1
• ⁢
• ∑
• d
• =
• d
• min
• d
• max
• ⁢
• max
• ⁡
• (
• min
• ⁡
• (
• Δ
• ⁢
• d
• ⁡
• (
• x
• ,
• y
• ,
• d
• )
• -
• 1
• ,
• D
• 3
• )
• ,
• 0
• )
• 2
• max
• ⁡
• (
• Δ
• ⁢
• ⁢
• c
• ⁡
• (
• x
• ,
• y
• ,
• d
• )
• -
• ⁢
• c
• m
• ⁢
• ⁢
• ax
• ⁡
• (
• x
• ,
• y
• )
• 5
• ,
• 1
• )
• ,
• where
• (x,y) is the predetermined position, d denotes a disparity candidate,
• max
• ⁡
• (
• min
• ⁡
• (
• Δ
• ⁢
• d
• ⁡
• (
• x
• ,
• y
• ,
• d
• )
• -
• 1
• ,
• D
• 3
• )
• ,
• 0
• )
• 2
• max
• ⁡
• (
• Δ
• ⁢
• ⁢
• c
• ⁡
• (
• x
• ,
• y
• ,
• d
• )
• -
• c
• m
• ⁢
• ⁢
• ax
• ⁡
• (
• x
• ,
• y
• )
• 5
• ,
• 1
• )
• is the contribution value for the disparity candidate d,
• dmin and dmax are a lower and upper bounds of an interval within which the set of disparity candidates are spread,
• D=dmax−dmin,
• cmax is a maximum dissimilarity among the dissimilarities associated with the set of disparity candidates,
• □c(x,y,d)=c(d(x,y))−c1(x,y) with c(d) being the dissimilarity associated with the disparity candidate d, and c1 being the minimum dissimilarity,
• □d=|d−d1| with d1 being a predetermined disparity having a minimum dissimilarity associated therewith among dissimilarities associated with the set of disparity candidates.