CA2361410A1 - Optimized signal quantification - Google Patents
Optimized signal quantification Download PDFInfo
- Publication number
- CA2361410A1 CA2361410A1 CA002361410A CA2361410A CA2361410A1 CA 2361410 A1 CA2361410 A1 CA 2361410A1 CA 002361410 A CA002361410 A CA 002361410A CA 2361410 A CA2361410 A CA 2361410A CA 2361410 A1 CA2361410 A1 CA 2361410A1
- Authority
- CA
- Canada
- Prior art keywords
- signal
- quantification
- subband
- function
- applying
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Abandoned
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/12—Selection from among a plurality of transforms or standards, e.g. selection between discrete cosine transform [DCT] and sub-band transform or selection between H.263 and H.264
- H04N19/122—Selection of transform size, e.g. 8x8 or 2x4x8 DCT; Selection of sub-band transforms of varying structure or type
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/13—Adaptive entropy coding, e.g. adaptive variable length coding [AVLC] or context adaptive binary arithmetic coding [CABAC]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/124—Quantisation
- H04N19/126—Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/61—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/61—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding
- H04N19/619—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding the transform being operated outside the prediction loop
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/63—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding using sub-band based transform, e.g. wavelets
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/63—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding using sub-band based transform, e.g. wavelets
- H04N19/635—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding using sub-band based transform, e.g. wavelets characterised by filter definition or implementation details
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/80—Details of filtering operations specially adapted for video compression, e.g. for pixel interpolation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/85—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
Abstract
A method for optimizing signal quantification (140) including applying reversible filters (110) to the signal in order to pre-quantify the signal a s a continuous function of the frequency domain; pre-processing the signal (12 0) in order to place it in proper color space and frequency domain; applying subband transforms to the signal (130) in order to split the signal into frequency regions (140), and entropy coding the signal (150).
Description
Optimized Signal Quantification Related Application The subject matter of this application is related to the subject matter of the following commonly owned applications: Serial Number 09/112,668, attorney docket number 3486, titled "Apparatus And Method For Entropy Coding', filed on July 9, 1998, also by Kenbe Goertzen; Serial Number , attorney docket number 4754, titled "Scaleable Resolution Motion Image Recording And Storage System', filed concurrently, also by Kenbe Goertzen; Serial Number , attorney docket number 4753, titled "A System And Method For Improving Compressed Image Appearance Using Stochastic Resonance And Energy Replacement", filed concurrently, also by Kenbe Goertzen; Serial Number , attorney docket number 4754, titled "Scaleable Resolution Motion Image Recording And Storage System', filed concurrently, also by Kenbe Goertzen; Serial Number attorney docket number 4756, titled "Quality Priority Image Storage and Commurucatiori', filed concurrently, also by Kenbe Goertzen; the contents of which are incorporated by reference as if fully disclosed herein.
Technical Field This invention pertains to the field of digital signal compression and quantification. More specifically, the present invention related to a method for optimizing signal quantification, particularly the quantification of signals transmitting still and motion image components.
Shortcomings of Prior Art Multiband compression methods have generally divided a signal into frequency components and then used some method to quantify the values in each of the frequency bands in order to represent the desired signal quality. Some of the problems with this approach include:
1. Only a small number of frequency bands are used so the regional quantification method is a Less than optimal coarse approximation of the desired function.
Technical Field This invention pertains to the field of digital signal compression and quantification. More specifically, the present invention related to a method for optimizing signal quantification, particularly the quantification of signals transmitting still and motion image components.
Shortcomings of Prior Art Multiband compression methods have generally divided a signal into frequency components and then used some method to quantify the values in each of the frequency bands in order to represent the desired signal quality. Some of the problems with this approach include:
1. Only a small number of frequency bands are used so the regional quantification method is a Less than optimal coarse approximation of the desired function.
2. Unequal quantification of neighboring frequency bands generally increases the amount of aliasing in the reconstruction mechanism.
3. Many quantification methods can generate undesirable artifacts in various degenerate cases.
4. Quantification as a separate process adds time or hardware to the implementation.
5. Quantification as a separate process can add additional noise.
What is needed is a system and method for quantifying one or more signals in an image stream such that the method can be easily implemented while offering better coding efficiency and more degrees of freedom in designing subband filters.
Summary of Invention Wavelet compression of images generally consists of subband transforms of an image into frequency regions. These regions are then quantified to relative resolutions and entropy coded. The present invention uses a continuous and frequency specific function to quantify the regions rather than a regional approximation.
More specifically, the method uses reversible filters before and after signal quantification as a continuous function of the frequency domain. This allows the scaling function to be either the exact desired function, or more closely approximate the desired continuous quantification function. It also allows the characteristics of any quantification and aliasing artifacts to be tailored to the particular application. As a result, the present invention provides better interpolation of quantification errors resulting in less noticeable artifacts in the quantified stream. The present invention also provides lower aliasing energy between subbands.
Brief Description of the Drawings These and other more detailed and specific objects and features of the present invention are more fully disclosed in the following specification, reference being had to the accompanying drawings in which:
Fig.1 is a flowchart illustrating the preferred method of the present invention;
Fig. 2 is an example of a typical 2D quantification map for a 2 band pyramid transform;
Fig. 3 is an example of an optimum quantification map for a separable complete 2 band transform;
Fig. 4 is an example of an optimum quantification map for a non-separable complete 2 band transform;
Fig. 5 is an example slice of the quantification surfaces demonstrating the advantages provided by the present invention; and Fig. 6A and 6B are examples of a bandsplit filter that was made more accurate by applying the method of optimum quantification.
Detailed Description Wavelet compression of images generally consists of subband transforms of an image into frequency regions. These regions are then quantified to relative resolutions and entropy coded. Although the method will be described in terms of steps, the steps involving application of the quantification and inverse quantification functions to the signal can also be incorporated directly into the subband analysis and synthesis filters thus avoiding a separate quantification and dequantification step.
Referring now to figure 1, the method of the present invention is shown. In a first embodiment, the method begins by applying 110 reversible filters to the signal in order to pre-quantify the signal as a continuous function of the frequency domain.
'This allows the scaling function to be either the exact desired function, or more closely approximate the desired continuous quantification function. It also allows the characteristics of any quantification and aliasing artifacts to be tailored to the specific application or signal.
After filtering the signal, the signal is pre-processed 120. This might include performing any number of different processes on the signal, such as converting the colorspace. Other forms of preprocessing may be applied depending on the type of signal and the desired output. In step 130, subband transforms of the signal split the signal into frequency regions. This enables quantification of the signal by region.
While the region quantification step is described as a separate step from subband transforms, these two steps may be combined into a single step if desired. In this step 140, each region is separately quantified. In the preferred embodiment of the present invention, a continuous function is used to quantify the frequency regions.
This can provide a significant improvement in efficiency over a stepwise approximation, as well as a reduction in aliasing. Finally, the quantified signal is subjected to entropy coding resulting in maximum compression. In the preferred embodiment, the step of entropy coding is performed in accordance with the entropy coding method described in the related application entitled, "Apparatus and Method for Entropy Coding."
An important application of this method is still and motion image encoding.
Images are often sampled as discrete x,y or x,y,z grids of digital values. If these values are Iinearized with uncorrelated noise, then sampling theory indicates that as the frequency of interest is reduced from the Nyquest frequency, the signal resolution increases. It is desirable to quantify the transformed image to a curve matching the resolution function established by sampling theory. This function in one dimension is:
resolution = (1/frequency)~0.5 * Nyquest resolution where frequency is in the range of {0..1} where 1 represents the Nyquest frequency The separable n dimensional case is:
resolution = (product from 1 to n of (1/Fn)~0.5 ) * Nyquest resolution Subband image coders typically quantify the subbands using a stepwise approximation of this function to obtain acceptable images after compression.
If a continuous function is used for the quantification, it can provide a significant improvement in efficiency over a stepwise approximation, as well as a reduction in aliasing. Reversible quantification filter functions can also be developed to provide other types of continuous quantification functions other than those provided by sampling theory. An example would be a continuous quantification function matching human perception resolution.
Even in cases where an exact continuous filter and its inverse cannot be designed for the desired quantification function, benefit from the method can be obtained by treating the desired function as a lower bound and designing a filter and its inverse which are bounded by the desired continuous function but better than a stepwise approximation by region.
The greater the dynamic range required of the quantification filter and its inverse, the more difficult they are to design. This problem can be solved by dividing the range into regions combined with the subband transform process. Once the range has been divided, a family of filters may be used to approximate the desired continuous function. Alternatively, a single filter can be used recursively.
If the subband mechanism has adequate resolution, the subband filters can be convolved with the quantification filters to combine the transform steps. In a preferred embodiment, the quantification filter design is combined with the subband filter design to increase the degrees of freedom for design of the subband filters. This would enable one to design a subband filter in a similar fashion to biorthagonal filter design.
Subband transforms may also be applied separately by dimension. If subband 1 S transforms are to be applied separately by dimension, then the type of transform is determined by the desire to generate a smooth and continuous quantification function matching the resolution specified by sampling theory. This requires that only the low frequency halfband from each bandsplit be further subdivided.
In the case of two or more dimensions, this requires a more complete transform than the typical pyramid subband transform. A two-dimensional two-band pyramid transform generates 7 regions, a complete transform of low frequencies by dimension generates 9 regions, and a full transform would generate 16. Non-separable multidimensional filters and subband transforms can then be designed which allow the generation of the smooth, continuous quantification function for a pyramid transform. While the above method was described as the preferred embodiment, the following embodiments represent alternative compression processing methods.
The following is an alternate set of steps for a Precompensation -based subband image compression process.
Step 1: preprocessing such as color space conversion Step 2: precompensation filter matching sampling theory resolution (may be combined with transform) Step 3: subband transform Step 4: entropy coding The following is an alternate combined subband image compression process Step 1: preprocessing such as color space conversion Step 2: subband transform matching sampling theory resolution Step 3: entropy coding Examples and Related Calculations The following three plots outline the difference for a quantification surface which will assure 0 dB of loss at the Nyquest frequency. Referring to Figure 2, an example of a typical 2D quantification map for a 2 band pyramid transform is shown.
The illustrated transform was generated by using the following plotting equation:
Plot3D[dB[rs2b[x,y]],{x,O,Pi},{y,O,Pi},PlotPoints->30,PlotRange->{0,18}];
Referring to Figure 3, an example of an optimum quantification map for a separable complete 2 band transform is shown. The illustrated transform was generated by using the following plotting equation:
Plot3D[dB[rc2b[x,y]],{x,O,Pi},{y,O,Pi},PlotPoints->30,P1otRange->{0,18}];
What is needed is a system and method for quantifying one or more signals in an image stream such that the method can be easily implemented while offering better coding efficiency and more degrees of freedom in designing subband filters.
Summary of Invention Wavelet compression of images generally consists of subband transforms of an image into frequency regions. These regions are then quantified to relative resolutions and entropy coded. The present invention uses a continuous and frequency specific function to quantify the regions rather than a regional approximation.
More specifically, the method uses reversible filters before and after signal quantification as a continuous function of the frequency domain. This allows the scaling function to be either the exact desired function, or more closely approximate the desired continuous quantification function. It also allows the characteristics of any quantification and aliasing artifacts to be tailored to the particular application. As a result, the present invention provides better interpolation of quantification errors resulting in less noticeable artifacts in the quantified stream. The present invention also provides lower aliasing energy between subbands.
Brief Description of the Drawings These and other more detailed and specific objects and features of the present invention are more fully disclosed in the following specification, reference being had to the accompanying drawings in which:
Fig.1 is a flowchart illustrating the preferred method of the present invention;
Fig. 2 is an example of a typical 2D quantification map for a 2 band pyramid transform;
Fig. 3 is an example of an optimum quantification map for a separable complete 2 band transform;
Fig. 4 is an example of an optimum quantification map for a non-separable complete 2 band transform;
Fig. 5 is an example slice of the quantification surfaces demonstrating the advantages provided by the present invention; and Fig. 6A and 6B are examples of a bandsplit filter that was made more accurate by applying the method of optimum quantification.
Detailed Description Wavelet compression of images generally consists of subband transforms of an image into frequency regions. These regions are then quantified to relative resolutions and entropy coded. Although the method will be described in terms of steps, the steps involving application of the quantification and inverse quantification functions to the signal can also be incorporated directly into the subband analysis and synthesis filters thus avoiding a separate quantification and dequantification step.
Referring now to figure 1, the method of the present invention is shown. In a first embodiment, the method begins by applying 110 reversible filters to the signal in order to pre-quantify the signal as a continuous function of the frequency domain.
'This allows the scaling function to be either the exact desired function, or more closely approximate the desired continuous quantification function. It also allows the characteristics of any quantification and aliasing artifacts to be tailored to the specific application or signal.
After filtering the signal, the signal is pre-processed 120. This might include performing any number of different processes on the signal, such as converting the colorspace. Other forms of preprocessing may be applied depending on the type of signal and the desired output. In step 130, subband transforms of the signal split the signal into frequency regions. This enables quantification of the signal by region.
While the region quantification step is described as a separate step from subband transforms, these two steps may be combined into a single step if desired. In this step 140, each region is separately quantified. In the preferred embodiment of the present invention, a continuous function is used to quantify the frequency regions.
This can provide a significant improvement in efficiency over a stepwise approximation, as well as a reduction in aliasing. Finally, the quantified signal is subjected to entropy coding resulting in maximum compression. In the preferred embodiment, the step of entropy coding is performed in accordance with the entropy coding method described in the related application entitled, "Apparatus and Method for Entropy Coding."
An important application of this method is still and motion image encoding.
Images are often sampled as discrete x,y or x,y,z grids of digital values. If these values are Iinearized with uncorrelated noise, then sampling theory indicates that as the frequency of interest is reduced from the Nyquest frequency, the signal resolution increases. It is desirable to quantify the transformed image to a curve matching the resolution function established by sampling theory. This function in one dimension is:
resolution = (1/frequency)~0.5 * Nyquest resolution where frequency is in the range of {0..1} where 1 represents the Nyquest frequency The separable n dimensional case is:
resolution = (product from 1 to n of (1/Fn)~0.5 ) * Nyquest resolution Subband image coders typically quantify the subbands using a stepwise approximation of this function to obtain acceptable images after compression.
If a continuous function is used for the quantification, it can provide a significant improvement in efficiency over a stepwise approximation, as well as a reduction in aliasing. Reversible quantification filter functions can also be developed to provide other types of continuous quantification functions other than those provided by sampling theory. An example would be a continuous quantification function matching human perception resolution.
Even in cases where an exact continuous filter and its inverse cannot be designed for the desired quantification function, benefit from the method can be obtained by treating the desired function as a lower bound and designing a filter and its inverse which are bounded by the desired continuous function but better than a stepwise approximation by region.
The greater the dynamic range required of the quantification filter and its inverse, the more difficult they are to design. This problem can be solved by dividing the range into regions combined with the subband transform process. Once the range has been divided, a family of filters may be used to approximate the desired continuous function. Alternatively, a single filter can be used recursively.
If the subband mechanism has adequate resolution, the subband filters can be convolved with the quantification filters to combine the transform steps. In a preferred embodiment, the quantification filter design is combined with the subband filter design to increase the degrees of freedom for design of the subband filters. This would enable one to design a subband filter in a similar fashion to biorthagonal filter design.
Subband transforms may also be applied separately by dimension. If subband 1 S transforms are to be applied separately by dimension, then the type of transform is determined by the desire to generate a smooth and continuous quantification function matching the resolution specified by sampling theory. This requires that only the low frequency halfband from each bandsplit be further subdivided.
In the case of two or more dimensions, this requires a more complete transform than the typical pyramid subband transform. A two-dimensional two-band pyramid transform generates 7 regions, a complete transform of low frequencies by dimension generates 9 regions, and a full transform would generate 16. Non-separable multidimensional filters and subband transforms can then be designed which allow the generation of the smooth, continuous quantification function for a pyramid transform. While the above method was described as the preferred embodiment, the following embodiments represent alternative compression processing methods.
The following is an alternate set of steps for a Precompensation -based subband image compression process.
Step 1: preprocessing such as color space conversion Step 2: precompensation filter matching sampling theory resolution (may be combined with transform) Step 3: subband transform Step 4: entropy coding The following is an alternate combined subband image compression process Step 1: preprocessing such as color space conversion Step 2: subband transform matching sampling theory resolution Step 3: entropy coding Examples and Related Calculations The following three plots outline the difference for a quantification surface which will assure 0 dB of loss at the Nyquest frequency. Referring to Figure 2, an example of a typical 2D quantification map for a 2 band pyramid transform is shown.
The illustrated transform was generated by using the following plotting equation:
Plot3D[dB[rs2b[x,y]],{x,O,Pi},{y,O,Pi},PlotPoints->30,PlotRange->{0,18}];
Referring to Figure 3, an example of an optimum quantification map for a separable complete 2 band transform is shown. The illustrated transform was generated by using the following plotting equation:
Plot3D[dB[rc2b[x,y]],{x,O,Pi},{y,O,Pi},PlotPoints->30,P1otRange->{0,18}];
Referring now to figure 4, an example of an optimum quantification map for a non-separable complete 2 band transform is provided. The illustrated transform was generated by using the following plotting equation:
Plot3D[dB[rr2b[x,y]]+3,{x,O,Pi},{y,O,Pi},PlotPoints->30,P1otRange->{0,18}]
Referring now to figure 5, an example slice of the quantification surfaces demonstrate the advantages provided by the present invention. The plot was generated using the following plotting equation:
Plot[{dB [rs2b [x,x]],dB [rc2b [x,x]],dB [rr2b[x,x]]+3}, {x,O,Pi},Plo tRange->
{0,20}
The following equations provide a method for calculating the non-separable and separable advantage over a stepwise approximation for a given channel resolution.
N[NIntegrate[dB[rs2b[x,y]]-(dB[rr2b[x,y]]+3),{x,O,Pi},{y,O,Pi}]/Pi~2] 3.77673 N[NIntegrate[dB[rs2b[x,y]]-dB[rc2b[x,y]],{x,O,Pi},{y,O,Pi}]/Pi~2] 3.25787 Referring now to figure 6A and 6B, an example of a bandsplit filter which was made more accurate by applying the method of optimum quantification is shown.
The variables used in the plotting equation are provided below.
tal={1168, 590, -106, -78, 34};
tsl={1168, 590, -106, -78, 34};
tah={990, -454+1, -166, 83-1, 42};
tsh={1390+6, -769, -18-3, 62, 30};
ralt[w_]:=Sum[Cos[n*w]*tal[[Abs[n]+1]],{n,-4,4}]/ (2048) raht[w_]:=Sum[Cos[n*w]*tah[[Abs[n]+1]],{n,-4,4}]/ (2048) rslt[w_]:=Sum[Cos[n*w]*tsl[[Abs[n]+1]],{n,-4,4}]/ (2048) rsht[w_]:=Sum[Cos[n*w]*tsh[[Abs[n]+1]],{n,-4,4}]/(2048) Figure 6A was plotted using the following equation:
Plot[{(raft[x]*rslt[x]+raht[x]*rsht[x]-1)*1000},{x,O,Pi}].
Figure 6B was plotted using the following equation:
Plot[{ralt[x],raht[x],rslt[x],rsht[x] }, {x,O,Pi}].
Although the description above contains many detailed descriptions, these descriptions should not be construed as limiting the scope of the invention but merely as providing illustrations of some of the presently preferred implementations of this invention. For example, although this method was described with reference to standard motion and still images, this method can be used to optimize quantification of any signal stream. Thus the scope of the invention should be determined by the appended claims and their legal equivalents, rather than by examples given.
Plot3D[dB[rr2b[x,y]]+3,{x,O,Pi},{y,O,Pi},PlotPoints->30,P1otRange->{0,18}]
Referring now to figure 5, an example slice of the quantification surfaces demonstrate the advantages provided by the present invention. The plot was generated using the following plotting equation:
Plot[{dB [rs2b [x,x]],dB [rc2b [x,x]],dB [rr2b[x,x]]+3}, {x,O,Pi},Plo tRange->
{0,20}
The following equations provide a method for calculating the non-separable and separable advantage over a stepwise approximation for a given channel resolution.
N[NIntegrate[dB[rs2b[x,y]]-(dB[rr2b[x,y]]+3),{x,O,Pi},{y,O,Pi}]/Pi~2] 3.77673 N[NIntegrate[dB[rs2b[x,y]]-dB[rc2b[x,y]],{x,O,Pi},{y,O,Pi}]/Pi~2] 3.25787 Referring now to figure 6A and 6B, an example of a bandsplit filter which was made more accurate by applying the method of optimum quantification is shown.
The variables used in the plotting equation are provided below.
tal={1168, 590, -106, -78, 34};
tsl={1168, 590, -106, -78, 34};
tah={990, -454+1, -166, 83-1, 42};
tsh={1390+6, -769, -18-3, 62, 30};
ralt[w_]:=Sum[Cos[n*w]*tal[[Abs[n]+1]],{n,-4,4}]/ (2048) raht[w_]:=Sum[Cos[n*w]*tah[[Abs[n]+1]],{n,-4,4}]/ (2048) rslt[w_]:=Sum[Cos[n*w]*tsl[[Abs[n]+1]],{n,-4,4}]/ (2048) rsht[w_]:=Sum[Cos[n*w]*tsh[[Abs[n]+1]],{n,-4,4}]/(2048) Figure 6A was plotted using the following equation:
Plot[{(raft[x]*rslt[x]+raht[x]*rsht[x]-1)*1000},{x,O,Pi}].
Figure 6B was plotted using the following equation:
Plot[{ralt[x],raht[x],rslt[x],rsht[x] }, {x,O,Pi}].
Although the description above contains many detailed descriptions, these descriptions should not be construed as limiting the scope of the invention but merely as providing illustrations of some of the presently preferred implementations of this invention. For example, although this method was described with reference to standard motion and still images, this method can be used to optimize quantification of any signal stream. Thus the scope of the invention should be determined by the appended claims and their legal equivalents, rather than by examples given.
Claims (14)
1. A method for optimizing signal quantification comprising:
applying reversible filters to the signal in order to pre-quantify the signal as a continuous function of the frequency domain;
pre-processing the signal;
applying subband transforms to the signal wherein the signal is split into frequency regions; and entropy coding the signal.
applying reversible filters to the signal in order to pre-quantify the signal as a continuous function of the frequency domain;
pre-processing the signal;
applying subband transforms to the signal wherein the signal is split into frequency regions; and entropy coding the signal.
2. The method of claim 1, wherein the step of applying subband transforms to the signal comprises using a continuous function to quantify the frequency regions.
3. The method of claim 1, wherein the step of pre-processing the signal includes converting the colorspace of the signal.
4. The method of claim 1, wherein the step of applying reversible filters to the signal comprises using a quantification filter.
5. The method of claim 4, wherein the step of applying subband transforms includes using a subband filter.
6. The method of claim 5, wherein the step of applying reversible filters to the signal further comprises using a subband filter in conjunction with the quantification filter design to increase the degrees of freedom for design of the subband filters.
7. The method of claim 1, wherein the step of applying subband transforms includes applying the subband transforms separately by dimension.
8. The method of claim 1, further comprising the step of quantifying the transformed image to a curve matching the resolution function established by sampling theory.
9. The method of claim 1, wherein the method comprises the further step of calculating a quantification function that approximates the desired signal.
10. The method of claim 9, the method comprising the further step of:
designing an exactly continuous filter and its inverse for the calculated quantification function.
designing an exactly continuous filter and its inverse for the calculated quantification function.
11. The method of claim 10 comprising the further step of: responsive to the an inability to design an exact continuous filter and its inverse for the quantification function, designing a filter and its inverse which are bounded by the desired continuous function wherein the desired function is used as a lower bound.
12. The method of claim 1, further comprising the step of dividing signal frequency range into regions created by the step of applying a subband transform to the signal.
13. The method of claim 12, further comprising the step of: using a family of filters to approximate the desired continuous function.
14. The method of claim 12, further comprising the step of: using a single filter recursively to approximate the desired continuous function.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US11855599P | 1999-02-04 | 1999-02-04 | |
| US60/118,555 | 1999-02-04 | ||
| PCT/US2000/003051 WO2000046738A1 (en) | 1999-02-04 | 2000-02-04 | Optimized signal quantification |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CA2361410A1 true CA2361410A1 (en) | 2000-08-10 |
Family
ID=22379338
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA002361410A Abandoned CA2361410A1 (en) | 1999-02-04 | 2000-02-04 | Optimized signal quantification |
Country Status (6)
| Country | Link |
|---|---|
| EP (1) | EP1157350A4 (en) |
| JP (1) | JP2002536896A (en) |
| KR (1) | KR20010101973A (en) |
| AU (1) | AU771802B2 (en) |
| CA (1) | CA2361410A1 (en) |
| WO (1) | WO2000046738A1 (en) |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5014134A (en) * | 1989-09-11 | 1991-05-07 | Aware, Inc. | Image compression method and apparatus |
| GB2281465B (en) * | 1993-08-27 | 1997-06-04 | Sony Uk Ltd | Image data compression |
| US5748786A (en) * | 1994-09-21 | 1998-05-05 | Ricoh Company, Ltd. | Apparatus for compression using reversible embedded wavelets |
-
2000
- 2000-02-04 EP EP00905993A patent/EP1157350A4/en not_active Withdrawn
- 2000-02-04 AU AU27571/00A patent/AU771802B2/en not_active Ceased
- 2000-02-04 WO PCT/US2000/003051 patent/WO2000046738A1/en not_active Application Discontinuation
- 2000-02-04 JP JP2000597747A patent/JP2002536896A/en not_active Withdrawn
- 2000-02-04 CA CA002361410A patent/CA2361410A1/en not_active Abandoned
- 2000-02-04 KR KR1020017009816A patent/KR20010101973A/en not_active Application Discontinuation
Also Published As
| Publication number | Publication date |
|---|---|
| JP2002536896A (en) | 2002-10-29 |
| AU2757100A (en) | 2000-08-25 |
| WO2000046738A1 (en) | 2000-08-10 |
| EP1157350A1 (en) | 2001-11-28 |
| EP1157350A4 (en) | 2005-09-14 |
| AU771802B2 (en) | 2004-04-01 |
| KR20010101973A (en) | 2001-11-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zettler et al. | Application of compactly supported wavelets to image compression | |
| Egger et al. | High-performance compression of visual information-a tutorial review. I. Still pictures | |
| US8027547B2 (en) | Method and computer program product for compressing and decompressing imagery data | |
| Egger et al. | Subband coding of images using asymmetrical filter banks | |
| DE19819405B4 (en) | Implementation of a reversible embedded wavelet system | |
| Karray et al. | Image coding with an L/sup/spl infin//norm and confidence interval criteria | |
| US6718065B1 (en) | Optimized signal quantification | |
| KR19990063831A (en) | Super Spatial Variable Apodization (Super SVA) | |
| Baligar et al. | Low complexity, and high fidelity image compression using fixed threshold method | |
| CA2361410A1 (en) | Optimized signal quantification | |
| Frost et al. | JPEG dequantization array for regularized decompression | |
| Hong et al. | Subband adaptive regularization method for removing blocking effect | |
| Kumar et al. | Quantization based wavelet transformation technique for digital image compression with removal of multiple artifacts and noises | |
| Chen et al. | Medical image compression with structure-preserving adaptive quantization | |
| Eddins et al. | A three-source multirate model for image compression | |
| Odegard et al. | Design of linear phase cosine modulated filter banks for subband image compression | |
| Cai et al. | Minimization of boundary artifacts on scalable image compression using symmetric-extended wavelet transform | |
| KR100200621B1 (en) | Method for compressing video signal through wavelet transform using human visual system | |
| Li et al. | Efficient quantization noise reduction device for subband image coding schemes | |
| Shen et al. | Minimization of aliasing artifacts during partial subband reconstruction with Wiener filters | |
| Sugavaneswaran | A novel enhanced compression scheme for digital images and its application for medical imaging | |
| Van Dyck et al. | Subband/VQ coding of color images using a separable diamond decomposition | |
| Li | Dynamic region-based wavelet coding for telemedicine applications | |
| Jónsson et al. | Efficient motion-oriented filter banks for video coding | |
| Jones et al. | Wavelet quad-tree compression of medical images using JPEG quantization and encoding strategies |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| EEER | Examination request | ||
| FZDE | Dead |