B-adaptive adpcm image data compressor

Claim 3

The method, as recited in claim 1, further comprising the step of: repeating each of the steps of claim 1 for each block of data in the stream of digital data.

Claim 4

The method, as recited in claim 1, wherein the step of providing a distortion level comprises the step of calculating a RMS distortion level over the block of data.

Claim 5

The method, as recited in claim 1, wherein the step of providing a distortion level comprises the step of calculating a peak distortion level of a data sample within the block of data.

Claim 6

The method, as recited in claim 1, further comprising the step of: storing quantized linear prediction coefficients, quantized input mean level, quantized gain and quantized prediction error sequence, and a word representing the number of bits used to represent the prediction error in the quantized prediction error; and wherein said step of comparing said distortion level to a preselected distortion level further comprises the steps of: transmitting said stored parameters when the computed distortion level is less than the preselected distortion level maximum.

Claim 7

The method, as recited in claim 5, wherein during said comparing step the distortion level exceeds the preselected distortion level maximum an wherein said method further comprises the step of: storing recomputated quantized linear predictive coefficients, quantized input signal mean, quantized gain, and quantized prediction error sequence for said block of data and a new value of the number of bits used to represent the prediction error in said quantized prediction error sequence.

Claim 8

The method, as recited in claim 1, wherein said input data stream represents image sonar data.

Claim 9

The method, as recited in claim 7, wherein a quantized linear prediction coefficient, input mean level and gain, and prediction error sequence which provides a reconstructed input data that does not exceed a predetermined distortion level, is provided, and wherein said method further comprises the step of: transmitting said parameters to a user device and wherein said user device reconstructs the input data from said transmitted parameters.

Claim 10

The method, as recited in claim 1, wherein during said comparing step when the distortion level exceeds the distortion level maximum and a value of the number of bits used to represent the prediction error in the quantized prediction error sequence equals the maximum of bits, further comprising the step of: recomputing quantized linear predictive coefficients, quantized input mean, quantized gain, and the unquantized prediction error sequence for said block of data and a quantized prediction error sequence using a higher number of bits in the quantized prediction error sequence to represent the prediction error in each sample of the unquantized prediction error sequence and further using a higher number of bits to represent the linear predictive coefficients, input mean, and gain in the quantized linear predictive coefficients, quantized input mean, and quantized gain.

Claim 11

Apparatus for processing a stream of digital data indicative of an image to provide to a user device, the apparatus comprising: means for partitioning the stream of data into a plurality of blocks of data samples; means for computing quantized linear prediction coefficients, quantized gain and quantized input signal mean level, and a unquantized prediction error sequence associated with a first one of said blocks of data, said computing means further comprising: means for quantizing said unquantized prediction error sequence with each prediction error sample in said unquantized prediction error sequence being quantized to a first selected number of bits in said quantized prediction error sequence; means for reconstructing the input data in each block of data from the quantized linear prediction coefficients, quantized gain and quantized input signal mean level, and quantized prediction error sequence; means for calculating a distortion level representative of a difference between the reconstructed data and the input data; means for comparing said distortion level to a preselected distortion level maximum to produce an indication when the distortion level exceeds the predetermined maximum; and means for recomputing the unquantized prediction error sequence for the block of input data when the distortion level exceeds the preselected distortion level maximum and means for quantizing said recomputed unquantized prediction error sequence using a higher number of bits to represent each prediction error sample; and means for providing said quantized linear prediction coefficients, quantized gain and quantized input signal mean level, and quantized prediction error sequence associated with the first one of said blocks of data to the user device.

Claim 12

Apparatus for processing a stream of digital data indicative of an image to provide to a user device, the apparatus comprising: means for partitioning the stream of data into a plurality of blocks of data samples; means for computing quantized linear prediction coefficients, quantized gain and quantized input signal mean level, and a unquantized prediction error sequence associated with a first one of said blocks of data, said computing means further comprising: means for quantizing said unquantized prediction error sequence with each prediction error sample in said unquantized prediction error sequence being quantized to a first selected number of bits in said quantized prediction error sequence; means for reconstructing the input data in each block of data from the quantized linear prediction coefficients, quantized gain and quantized input signal mean level, and quantized prediction error sequence; means for calculating a distortion level representative of a difference between the reconstructed data and the input data; means for comparing said distortion level to a preselected distortion level maximum to produce an indication when the distortion level exceeds the predetermined maximum; and means for recomputing the unquantized prediction error sequence for the block of input data when the distortion level exceeds the preselected distortion level maximum and means for quantizing said recomputed unquantized prediction error sequence using a higher number of bits to represent each prediction error sample; and means for providing said quantized linear prediction coefficients, quantized gain and quantized input signal mean level, and quantized prediction error sequence associated with the first one of said blocks of data to the user device.

Claim 13

Apparatus, as recited in claim 10, further comprising means for providing an RMS distortion level over the block of data.

Claim 14

Apparatus, as recited in claim 11, further comprising means for computing a peak distortion level of a data sample within the block of data samples.

Claim 15

Apparatus, as recited in claim 11, further comprising means for storing quantized linear prediction coefficients, quantized gain and quantized input signal mean level, and quantized prediction error sequence and a word representing the number of bits used to represent the prediction error in the quantized prediction error sequence and means for transmitting said stored parameters when a computed distortion level is less than the preselected distortion level maximum.

Claim 16

Apparatus, as recited in claim 13, further comprising means for increasing the number of bits used to represent the quantized linear predictive coefficients, quantized gain, and quantized input mean, and means for computing new values of said parameters using a higher number of quantizing bits and for recomputing the quantized error prediction sequence using a higher number of bits when the value of bits used to represent the prediction error equals a maximum number of selected bits.