Image data compression having minimum perceptual error

Claim 3

The method of transforming an image according to claim 1 further comprising the steps: (a) transmitting entropy coded image to a user of said digital image; (b) decoding said entropy coded image; (c) decoding said quantization matrix to derive said DCT transformation coefficients; and (d) applying an inverse Discrete Cosine Transform to derive said block of pixels.

Claim 4

The method of transforming an image according to claim 2, wherein block mean luminance and component contrast are included in said display and perceptual parameters and are determined by the following expression: t.sub.ijk =t.sub.ij (C.sub.00 /C.sub.00).sup.a.sbsp.t where a t is a luminance-masking exponent having a value of about 0.65, t ij is the un-adjusted threshold of luminance, C oo is an average of the DC terms of the DCT coefficients C ijk of the image, and cook, is the DC coefficient of block k.

Claim 5

The method of transforming an image according to claim 2, wherein said DCT mask is determined by the following expression: m.sub.ijk =Max [t.sub.ijk,|C.sub.ijk |.sup.W.sbsp.if t.sub.ijk .sup.1-w.sbsp.if ] where C ijk is the DOT coefficient, t ijk is a corresponding block mean luminance adjusted threshold, and w ij is an exponent that lies between 0 and 1 and is about 0.7.

Claim 6

The method of transforming an image according to claim 2, wherein said pooling of perceptual errors is determined by the following expression: ##EQU2##

Claim 7

The computer involved in converting a block having an index k of pixels from an electronic image into a digital representation of said image, said computer comprising: (a) means for applying a Discrete Cosine Transform (DCT) to transform said block of pixels into digital signals represented by DCT transformation coefficient (c ijk ); (b) means for selecting a DCT mask (m ijk ) for each block of pixels based on parameters comprising said DCT transformation coefficients c iik and display and perceptual parameters; (c) means for selecting a quantization matrix (q ij ) comprising: (i) means for selecting an initial value of q ij ; (ii) means for quantizing the DCT coefficient c ijk in each block k to form quantized coefficient u ijk ; (iii) means for de-quantizing the quantized coefficients U ijk by multiplying them by q ij to form q ij u ijk ; (iv) means for subtracting the de-quantized coefficient q ij u ijk from c ijk to compute a quantization error e ijk ; (v) means for dividing e ijk by the DCT mask m ijk to obtain perceptual errors; (vi) means for pooling the perceptual errors of one frequency ij over all blocks k to obtain an entry in a perceptual error matrix p ij ; (vii) means for adjusting the values q ij up or down until each entry in the perceptual error matrix p ij is within a target range: and (d) means for entropy coding said quantization matrix and the quantized DCT coefficients of said image.

Claim 8

The computer involved in converting a block having an index k of pixels from an electronic image into a digital representation of said image, said computer comprising: (a) means for applying a Discrete Cosine Transform (DCT) to transform said block of pixels into digital signals represented by DCT transformation coefficient (c ijk ); (b) means for selecting a DCT mask (m ijk ) for each block of pixels based on parameters comprising said DCT transformation coefficients c iik and display and perceptual parameters; (c) means for selecting a quantization matrix (q ij ) comprising: (i) means for selecting an initial value of q ij ; (ii) means for quantizing the DCT coefficient c ijk in each block k to form quantized coefficient u ijk ; (iii) means for de-quantizing the quantized coefficients U ijk by multiplying them by q ij to form q ij u ijk ; (iv) means for subtracting the de-quantized coefficient q ij u ijk from c ijk to compute a quantization error e ijk ; (v) means for dividing e ijk by the DCT mask m ijk to obtain perceptual errors; (vi) means for pooling the perceptual errors of one frequency ij over all blocks k to obtain an entry in a perceptual error matrix p ij ; (vii) means for adjusting the values q ij up or down until each entry in the perceptual error matrix p ij is within a target range: and (d) means for entropy coding said quantization matrix and the quantized DCT coefficients of said image.

Claim 9

The computer involved in converting an image according to claim 6 further comprising: (a) means for transmitting entropy coded image to a user of said digital image; (b) means for decoding said entropy coded image; (c) means for decoding said quantization matrix to derive said DCT transform coefficients; and (d) means for applying an inverse Discrete Cosine Transform to derive said block of pixels.

Claim 10

The computer involved in converting an image according to claim 6, wherein block mean luminance and component contrast are included in said display and perceptual parameters and are determined by the following expression embodied in the operation of said means for selecting a DCT mask: t.sub.ijk =t.sub.ij (C.sub.ook /C.sub.oo).sup.a.sbsp.t where a t is a luminance-masking exponent having a value of about 0.65, t ij is the un-adjusted threshold of luminance, C oo is an average of the DC terms of the DCT coefficients C ijk of the image, and C ook is the DC coefficient of block k.

Claim 11

The computer network involved in converting an image according to claim 6, wherein said DCT mask is determined by the following expression embodied in said means for selecting a DCT mask: m.sub.ijk =Max[t.sub.ijk,|C.sub.ijk |.sup.W.sbsp.if t.sub.ijk.sup.1-W.sbsp.if ] m ijk , is the contrast-adjusted threshold, c ijk is the DCT coefficient, t ijk is the corresponding threshold of expression 2, and w ij is the exponent that lies between 0 and 1 and typically has a value of 0.7.

Claim 12

The computer involved in converting an image according to claim 6, wherein said pooling of perceptual errors is determined by the following expression embodied in said means for pooling: ##EQU3## Where d ijk is an error in a particular frequency i, j, and block k, β s is a pooling exponent having a typical value of 4.