Reconstruction of quantized DPCM or PCM signals

Claim 3

The method defined in claim 1 wherein said quantized samples each represent the intensity value of an element (pel) of a picture, and said assigning step includes storing reconstructed intensity values of pels in a neighborhood surrounding the picture element to which said present sample corresponds.

Claim 4

The method defined in claim 2 wherein said assigning step includes: (a) determining a normal reconstructed value for said present sample using said representative value, (b) determining the number (n and m) of said stored reconstructed values which are respectively less than and more than said normal reconstructed value, and (c) computing said reconstructed value as a function of n and m.

Claim 5

The method defined in claim 3 wherein said computing step includes: (d) forming a weighted average of an endpoint of said range and said normal reconstructed value using the values of n and m for weighting.

Claim 6

The method defined in claim 3 wherein said computing step includes selecting said reconstructed value I X ' such that ##EQU4## where (p+e r ) is said normal reconstructed value, and (p+e 1 ) and (p+e 2 ) are the endpoints of said range.

Claim 9

The method defined in claim 6 wherein said present quantized sample represents the intensity value of a particular element (pel) of a picture and said other ones of said samples represent intensity values of pels which surround said particular pel.

Claim 10

The method defined in claim 6 wherein said present quantized sample represents the prediction error value for a particular element (pel) of a picture formed by subtracting the predicted intensity value of said pel from the true value thereof, and said other ones of said samples represent pels which surround said particular pel.

Claim 12

The method defined in claim 9 wherein said weighting step includes: (a) weighting the lower endpoint of said range by n and weighting said normal reconstructed value by m, if n is greater than m, or (b) weighting the upper endpoint of said range by m and weighting said normal reconstructed value by n, if n is less than m.

Claim 15

The method defined in claim 11 wherein said assigning step includes selecting said previously processed samples based on their correlation with said present sample.

Claim 16

The method of claim 12 wherein said assigning step further includes: (a) determining the numbers m and n of said selected samples having values near the upper and lower ends of said range, respectively, and (b) weighting an average formed using at least one of the endpoints of said range with m and n.

Claim 19

Apparatus for reconstructing the value of samples of a correlated signal which have been quantized, said quantized samples each having a representative value associated with one of a predefined set of mutually exclusive qwuantization levels having upper and lower limits, characterized by means for determining a range for said reconstructed value of a present sample, based upon said upper and lower limits of said quantizer level associated with said representative value for said present sample, and means for assigning said reconstructed value within said range as a function of reconstructed values of samples spatially or temporally correlated with said present sample.

Claim 20

Apparatus for reconstructing the value of samples of a correlated signal which have been quantized, said quantized samples each having a representative value associated with one of a predefined set of mutually exclusive qwuantization levels having upper and lower limits, characterized by means for determining a range for said reconstructed value of a present sample, based upon said upper and lower limits of said quantizer level associated with said representative value for said present sample, and means for assigning said reconstructed value within said range as a function of reconstructed values of samples spatially or temporally correlated with said present sample.

Claim 21

The invention defined in claim 15 wherein said quantized samples each represent the intensity value of an element (pel) of a picture, and said assigning means includes means for storing reconstructed intensity values of pels in a neighborhood surrounding the picture element to which said present sample corresponds.

Claim 22

The invention defined in claim 16 wherein said assigning means includes: (a) means for determining a normal reconstructed value for said present sample using said representative value, (b) means for determining the number (n and m) of said stored reconstructed values which are respectively less than and more than said normal reconstructed value, and (c) means for computing said reconstructed value as a function of n and m.

Claim 23

The invention defined in claim 17 wherein said computing means includes: (d) means for forming a weighted average of an endpoint of said range and said normal reconstructed value using the values of n and m for weighting.

Claim 24

The invention defined in claim 17 wherein said computing means includes means for selecting said reconstructed value I X ' such that ##EQU5## where (p+e r ) is said normal reconstructed value, and (p+e 1 ) and (p+e 2 ) are the endpoints of said range.

Claim 25

Apparatus for increasing the accuracy with which the value of each of a series of quantized samples is reconstructed, each quantized sample having an original value within one of a predefined set of mutually exclusive quantization levels, each level having upper and lower limits and an associated representative value, characterized by: (a) means for determining a range of possible reconstructed values for a present sample, based upon said upper and lower limits of said quantizer level associated with said representative value for said present sample, and (b) means for assigning said reconstructed value within said range as a function of reconstructed values for other ones of said samples.

Claim 26

Apparatus for increasing the accuracy with which the value of each of a series of quantized samples is reconstructed, each quantized sample having an original value within one of a predefined set of mutually exclusive quantization levels, each level having upper and lower limits and an associated representative value, characterized by: (a) means for determining a range of possible reconstructed values for a present sample, based upon said upper and lower limits of said quantizer level associated with said representative value for said present sample, and (b) means for assigning said reconstructed value within said range as a function of reconstructed values for other ones of said samples.

Claim 27

The invention defined in claim 20 wherein said present quantized sample represents the intensity value of a particular element (pel) of a picture and said other ones of said samples represent intensity values of pels which surround said particular pel.

Claim 28

The invention defined in claim 20 wherein said present quantized sample represents the prediction error value for a particular element (pel) of a picture formed by subtracting the predicted intensity value of said pel from the true value thereof, and said other ones of said samples represent pels which surround said particular pel.

Claim 29

The invention defined in claims 21 or 22 wherein said assigning means includes: (1) means for determining the numbers (n and m) of said surrounding pels having reconstructed values which are respectively more than and less than the normal reconstructed value of said present quantized sample calculated using said representative value, and (2) means for weighting the average of an endpoint of said range and said normal reconstructed value as a function of n and m.

Claim 30

The invention defined in claim 23 wherein said weighting means is arranged to weight the lower endpoint of said range by n and weight said normal reconstructed value by m, if n is greater than m, or weight the upper endpoint of said range by m and weight said normal reconstructed value by n, if n is less than m.

Claim 31

Apparatus for reconstructing the value of a present sample of a signal, said present sample having an associated prediction error value e formed from the difference between the original value I X of said sample and a prediction p thereof, said error value having a quantized representative value e r associated with one of a predefined set of mutually exclusive quantization levels having upper and lower limits e rU and e rL , respectively, including: means for defining a range (e rL +p) to (e rU +p) for said reconstructed value of said present sample, based upon said upper and lower limits of said quantizer level associated with said normal representative value for said present sample, and means for assigning said reconstructed value within said defined range as a function of reconstructed values for previously processed samples.

Claim 32

Apparatus for reconstructing the value of a present sample of a signal, said present sample having an associated prediction error value e formed from the difference between the original value I X of said sample and a prediction p thereof, said error value having a quantized representative value e r associated with one of a predefined set of mutually exclusive quantization levels having upper and lower limits e rU and e rL , respectively, including: means for defining a range (e rL +p) to (e rU +p) for said reconstructed value of said present sample, based upon said upper and lower limits of said quantizer level associated with said normal representative value for said present sample, and means for assigning said reconstructed value within said defined range as a function of reconstructed values for previously processed samples.

Claim 33

The invention defined in claim 25 wherein said assigning means is arranged to select said previously processed samples based on their correlation with said present sample.

Claim 34

The invention of claim 26 wherein said assigning means includes: (a) means for determining the numbers m and n of said selected samples having values near the upper and lower ends of said range, respectively, and (b) means for weighting an average formed using at least one of the endpoints of said range with m and n.

Claim 35

Apparatus for reconstructing the value of quantized DPCM samples each representing the difference between the intensity value of an element (pel) of a picture and a predicted version thereof, said quantized samples each having a representative value associated with one of a predetermined set of mutually exclusive quantization levels, said representative value being intermediate upper and lower limits of the associated quantization level, including: (a) means for storing reconstructed intensity values for pels neighboring the present pel for which an intensity value is being reconstructed, (b) means for determining the numbers n and m of said stored values which are respectively less than and more than the reconstructed value of said present pel calculated using said representative value, and (c) means for calculating said reconstructed value as a joint function of m and n and said upper and lower limits of the quantization level with which said representative value is associated.

Claim 36

Apparatus for reconstructing the value of quantized DPCM samples each representing the difference between the intensity value of an element (pel) of a picture and a predicted version thereof, said quantized samples each having a representative value associated with one of a predetermined set of mutually exclusive quantization levels, said representative value being intermediate upper and lower limits of the associated quantization level, including: (a) means for storing reconstructed intensity values for pels neighboring the present pel for which an intensity value is being reconstructed, (b) means for determining the numbers n and m of said stored values which are respectively less than and more than the reconstructed value of said present pel calculated using said representative value, and (c) means for calculating said reconstructed value as a joint function of m and n and said upper and lower limits of the quantization level with which said representative value is associated.