Method of transform data compression using save-maps

• US 5,038,390 A
• Filed: 06/05/1990
• Issued: 08/06/1991
• Est. Priority Date: 06/05/1990
• Status: Expired due to Term

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Ravi Chandran, Vellore C.
• Primary Examiner(s): NOT, DEFINED
• Assistant Examiner(s): Santos, Daniel
• Application Number:
  US07/533,348
• Time in Patent Office: 427 Days
• Field of Search: 358/433, 358/426, 382/56, 382/50, 364/260.6
• US Class Current: 382/248
• CPC Class Codes: H04N 19/124 Quantisation
• CPC Class Codes: H04N 19/132 Sampling, masking or truncation of coding units, e.g. adaptive resampling, frame skipping, frame interpolation or high-frequency transform coefficient masking
• CPC Class Codes: H04N 19/154 Measured or subjectively estimated visual quality after decoding, e.g. measurement of distortion use of rate-distortion criteria H04N19/147
• CPC Class Codes: H04N 19/60 using transform coding

Abstract

A method for compressing the data required to represent an image using a set of save-maps, with each save-map being identified by an index and each save-map identifying a subset of transform coefficients. The method begins by spatially partitioning the image into a plurality of blocks, includes the generation of transform coefficients representative of the data spatially located in each block; the generation of an index for each block as a function of the transform coefficients for that block; and selecting one of the save-maps as a function of the index. By employing the index and the value of the magnitude components of the subset of transform coefficients identified by the selected save-map for each block, compressed data may be obtained representative of that portion of the image spatially contained in each of the corresponding blocks. Thus, each save-map permits retrieving the original image in accordance with the save-map and data representative of the magnitude components of the subsets of transform coefficients identified by the save-maps.

Claims

1. A method for compressing the data required to represent an original image by using a predefined set of save-maps, with each save-map being identified by an index and each save-map identifying a subset of transform coefficients, comprising the steps of:

spatially partitioning said image into a plurality of blocks;

generating, for each block of at least a subset of said plurality of blocks, a set of transform coefficients representative of the data spatially located in that block, with each of said transform coefficients having a magnitude component and a frequency component;

partitioning each block of transform coefficients into a plurality of zones, the dimensions of each said zone being a function of the different visual components of the image;

generating an index independently for each said block as a function of the energy level in more than one zone of the block;

selecting one of said save-maps as a function of said index; and

employing said index and the magnitude components of the subset of transform coefficients for each said block identified by said selected save-map for that block to represent that portion of said image spatially contained in that block.

2. The method of claim 1 wherein the step of generating an index for each said block comprises the steps of:

organizing the set of transform coefficients for that block into a matrix based on said frequency components of said transform coefficients;

partitioning at least a portion of said matrix into a plurality of zones;

summing the absolute values of the magnitude components of the transform coefficients in each zone to obtain a plurality of zone sums;

summing the absolute values of the magnitude components of all the transform coefficients to obtain a total sum;

dividing each zone sum by said total sum to obtain a normalized zonal activity level for each zone; and

comparing said total sum and said normalized zonal activity levels with corresponding predefined empirical values to generate a code which constitutes the index for each said block.

3. The method of claim 2 wherein the substep of comparing comprises:

comparing said total sum for each said block with a series of thresholds to generate a block code for the total sum of that block;

comparing the normalized zonal activity level of each zone in that block with a threshold to generate a zone code for each zone sum; and

combining said block code and the plurality of zone codes to obtain the index for that block.

4. The method of claim 1 further comprising the step of generating said set of save-maps by:

spatially partitioning at least one sample image similar in character to said original image into a plurality of blocks;

generating a set of transform coefficients for each block of at least a subset of said plurality of said blocks, said transform coefficients for each block having a magnitude component and a frequency component, and each said set of transform coefficients being representative of the data located in the corresponding block;

partitioning each block of transform coefficients into a plurality of zones, the dimensions of each said zone being a function of the different visual components of the image;

generating an index for each said block as a function of the energy level in more than one zone of the block, said index being capable of being computed independently for each said block;

selecting from each set of transform coefficients for each block which have the same index, a predefined subset of coefficients to represent that portion of said sample image spatially contained in the corresponding blocks; and

generating for each index one save-map as a function of said subsets of transform coefficients corresponding to each said index.

5. The method of claim 1 further comprising the step of repeating said steps of generating, selecting and employing for each of said plurality of said blocks.

6. The method of claim 5 further comprising the step of reconstructing said original image in response to said index and said magnitude components of said subset of transform coefficients for each of said blocks.

7. A method for compressing the data required to represent an original image by using a set of save-maps, with each save-map being identified by an index and each save-map identifying a subset of transform coefficients, comprising the steps of:

spatially partitioning said image into a plurality of blocks;

generating, for each block of at least a subset of said plurality of blocks, a set of transform coefficients representative of the data spatially located in that block, with each of said transform coefficients having a magnitude component and a frequency component;

generating an index for each said block as a function of said transform coefficients for that block, comprising the steps of;

organizing the set of transform coefficients for that block into a matrix based on said frequency components of said transform coefficients;

partitioning at least a portion of said matrix into a plurality of zones;

summing the absolute values of the magnitude components of the transform coefficients in each zone to obtain a plurality of zone sums;

summing the absolute values of the magnitude components of all the transform coefficients to obtain a total sum;

dividing each zone sum by said total sum to obtain a normalized zonal activity level for each zone; and

comparing said total sum and said normalized zonal activity level with corresponding predefined empirical values to generate a code which constitutes the index for each said block;

selecting one of said save-maps as a function of said index; and

employing said index and the magnitude components of the subset of transform coefficients for each said block identified by said selected save-map for that block to represent that portion of said image spatially contained in that block.

8. The method of claim 7 wherein the substep of comparing comprises:

comparing said total sum for each said block with a series of thresholds to generate a block code for the total sum of that block;

comparing the normalized zonal activity level of each zone in that block with a threshold to generate a zone code for each zone sum; and

combining said block code and the plurality of zone codes to obtain the index for that block.