Back to results
US 6,748,118 B1Granted
Method of quantizing signal samples of an image during same
AssigneeIntel Corporation
InventorsTinku Acharya, Ping-Sing Tsai
Life of this patent
Estimating the term…
Abstract
Briefly, in accordance with one embodiment of the invention, a method of quantizing signal samples of an image during image compression includes the following. A process to transform the signal samples from a first domain to a second domain is applied. During the transform process, signal samples are filter, by first applying scaled filter coefficients to signal samples along the image in a first direction and then applying scaled filter coefficients to signal samples along the image in a second direction, so that at the completion of the transform process of the image, selected regions of the transformed signal samples are quantized by a common value. Many other embodiments in accordance with the invention are also described.
Claims23
A method of quantizing signal samples of an image during image compression comprising:
- applying a process to transform the signal samples from a first domain to a second domain;
- and
- during the transform process, filtering signal samples, by first applying scaled filter coefficients to signal samples along the image in a first direction and then applying scaled filter coefficients to signal samples along the image in a second direction, so that at the completion of the transform process of the image, selected regions of the transformed signal samples are quantized by a common value, the selected regions are the LL, LH, HL, and HH subbands of the transformed image, and the quantization is substantially in accordance with the following form:
- LL subband at level k :
- 1 Q ( LL k ) 1 Q ( LL k ) = 1 Q ( LL k ) when k = K , or 1 when k < K ;
- HL subband at level k :
- Q ( LL k ) Q ( HL k ) 1 Q ( LL k ) = 1 Q ( HL k ) ;
- LH subband at level k :
- 1 Q ( LL k ) Q ( LL k ) Q ( LH k ) = 1 Q ( LH k ) ;
- HH subband at level k :
- Q ( LL k ) Q ( HL k ) Q ( HL k ) Q ( HH k ) Q ( LL k ) = 1 Q ( HH k ) ;
- K being the total level of the DWT and k being a positive integer less than or equal to K.
9 more elements
A method of quantizing signal samples of an image during image compression comprising:
- applying a process to transform the signal samples from a first domain to a second domain;
- and
- during the transform process, filtering signal samples, by first applying scaled filter coefficients to signal samples along the image in a first direction and then applying scaled filter coefficients to signal samples along the image in a second direction, so that at the completion of the transform process of the image, selected regions of the transformed signal samples are quantized by a common value, the transform process comprises a discrete wavelet transform (DWT), the selected regions are the LL, LH, HL, and HH subbands of the transformed image, scaling in a first direction comprises:
- applying the scale factor 1 Q ( LL k )
- to each filter coefficient in the low pass filtering operation over the LLk−1 subband;
- and
- applying the scale factor Q ( LL k ) Q ( HL k )
- to each filter coefficient in the high pass filtering operation over the LLk−1 subband;
- Q (HLk) and Q (LLk) being the quantization thresholds of the subbands, HL and LL in the kth level, respectively, K being the total level of the DWT, k being a positive integer less than or equal to K, Q (LLk) being 1 when k is less than K, and LL1 being the input image.
6 more elements
A method of quantizing signal samples of an image during image compression comprising:
- applying a process to transform the signal samples from a first domain to a second domain;
- and
- during the transform process, filtering signal samples, by first applying scaled filter coefficients to signal samples along the image in a first direction and then applying scaled filter coefficients to signal samples along the image in a second direction, so that at the completion of the transform process of the image, selected regions of the transformed signal samples are quantized by a common value, the transform process comprises a discrete wavelet transform (DWT), the selected regions are the LL, LH, HL, and HH subbands of the transformed image, scaling in the second mutually orthogonal direction comprises:
- applying the scale factor 1 Q ( LL k )
- to each filter coefficient in the low pass filtering operation to generate subbands LLk and HLk;
- applying the scale factor Q ( LL k ) Q ( LH k )
- to each filter coefficient in the high pass filtering operation to generate subband LHk;
- and
- applying the scale factor Q ( HL k ) Q ( HH k ) Q ( LL k )
- to each filter coefficient in the high pass filtering operation to generate subband HHk;
- Q (HLk), Q (HHk), Q (LHk), and Q (LLk) being the quantization thresholds of the subbands, HL, HH, LH, and LL in the kth level, respectively, and Q (LLk) being equal to 1, when level k is less than K.
8 more elements
A device comprising:
- an integrated circuit;
- said integrated circuit having input ports to receive signal samples associated with at least one image;
- said integrated circuit including digital circuitry;
- said digital circuitry having a configuration to apply a process to transform the signal samples from a first domain to a second domain and during the transform process, filtering signal samples, by first applying scaled filter coefficients to signal samples along the image in a first direction and then applying scaled filter coefficients to signal samples along the image in a second direction, so that at the completion of the transform process of the image, selected regions of the transformed signal samples are quantized by a value, the selected regions are the LL, LH, HL, and HH subbands of the transformed image, scaling in the second mutually orthogonal direction comprises:
- applying the scale factor 1 Q ( LL k )
- to each filter coefficient in the low pass filtering operation to generate subbands LLk and HLk;
- applying the scale factor Q ( LL k ) Q ( LH k )
- to each filter coefficient in the high pass filtering operation to generate subband LHk;
- and
- applying the scale factor Q ( HL k ) Q ( HH k ) Q ( LL k )
- to each filter coefficient in the high pass filtering operation to generate subband HHk;
- Q (HLk), Q (HHk), Q (LHk), and Q (LLk) being the quantization thresholds of the subbands, HL, HH, LH, and LL in the kth level, respectively, and Q (LLk) being equal to 1, when level k is less than K.
9 more elements
An article comprising:
- a storage medium, said storage medium having stored thereon, instructions, that when executed by a system to execute said instructions, results in:
- applying a process to transform signal samples associated with at least one image from a first domain to a second domain;
- and
- during the transform process, filtering signal samples, by first applying scaled filter coefficients to signal samples along the image in a first direction and then applying scaled filter coefficients to signal samples along the image in a second direction, so that at the completion of the transform process of the image, selected regions of the transformed signal samples are quantized by a value, scaling in the second mutually orthogonal direction comprises:
- applying the scale factor 1 Q ( LL k )
- to each filter coefficient in the low pass filtering operation to generate subbands LLk and HLk;
- applying the scale factor Q ( LL k ) Q ( LH k )
- to each filter coefficient in the high pass filtering operation to generate subband LHk;
- and
- applying the scale factor Q ( HL k ) Q ( HH k ) Q ( LL k )
- to each filter coefficient in the high pass filtering operation to generate subband HHk;
- Q (HLk), Q (HHk), Q (LHk), and Q (LLk) being the quantization thresholds of the subbands, HL, HH, LH, and LL in the kth level, respectively, and Q (LLk) being equal to 1, when level k is less than K.
9 more elements
Dependencies are inferred from the claim text.