Multiscale Coding Mode Selection for Video Compression

Solution Overview

Problem

Existing image compression methods, such as MPEG-2 and MPEG-4, make coding decisions for macroblocks in a causal manner without considering the impact on future macroblocks, leading to suboptimal choices and increased calculation costs, especially in heterogeneous images.

Innovation Solution

A multiscale approach that iteratively chooses a mode of coding for a set of blocks, refines the choice for subsets, and estimates coding costs for neighboring blocks, optimizing the energy function to minimize distortion and coding cost across the entire image.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If a causal coding decision method is used to choose the mode of coding for each macroblock in scanning order, then the calculation complexity for each macroblock is reduced, but the image quality deteriorates and coding cost increases due to suboptimal decisions affecting future macroblocks

Engineering Contradiction:

Improvecoding speedVSAvoidimage quality

Core Design Contradiction:

ProductivityVSManufacturing precision

Solution Approach 1:

The invention performs preliminary calculations of the energy function for all possible coding modes of future macroblocks in advance, storing these calculation elements in a database. This allows the current macroblock's optimal coding mode to be determined by considering future impacts without performing full energy function calculations at each step, thus maintaining high coding speed while improving image quality through more optimal decisions.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The invention extends the optimization from a one-dimensional sequential process to a multi-dimensional approach by considering not only the current macroblock but also future macroblocks in the scanning order. The energy function is calculated across multiple dimensions (current and future macroblocks) to find the globally optimal coding mode that minimizes overall distortion and coding cost.

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

2Manufacturing precision

If multiple iterations are performed to reconsider decisions for future macroblocks, then the image quality improves, but the calculation amount becomes extremely high and incompatible with video coding

Engineering Contradiction:

Improveimage qualityVSAvoidcalculation amount

Core Design Contradiction:

Manufacturing precisionVSPower

Solution Approach 1:

The invention performs preliminary calculations of the energy function for all possible coding modes of future macroblocks in advance, storing these calculation elements in a database. This allows the current macroblock's optimal coding mode to be determined by considering future impacts without performing full energy function calculations at each step, thus maintaining high coding speed while improving image quality through more optimal decisions.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The invention changes the parameter representation by pre-calculating and storing energy function values for different coding modes of future macroblocks. Instead of recalculating these values during iterations, the system uses stored parameters (calculation elements) to quickly evaluate the impact of different coding decisions, dramatically reducing the calculation amount while maintaining optimization quality.

Inventive Principle:

Principle #35Parameter changes

3Manufacturing precision

If the energy function is calculated over the entire set of blocks to find the global minimum, then the coding cost is optimized, but the calculation time becomes prohibitively long for practical application

Engineering Contradiction:

Improvecoding cost optimizationVSAvoidcalculation time

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The invention segments the calculation process into two parts: (1) pre-calculation of energy function values for future macroblocks' coding modes, and (2) selection of the optimal mode for the current macroblock using stored calculation elements. This segmentation allows the system to approach the global minimum without calculating the entire energy function from scratch for each macroblock, significantly reducing calculation time while maintaining optimization quality.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The invention performs preliminary calculations of the energy function for all possible coding modes of future macroblocks in advance, storing these calculation elements in a database. This allows the current macroblock's optimal coding mode to be determined by considering future impacts without performing full energy function calculations at each step, thus maintaining high coding speed while improving image quality through more optimal decisions.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS8059722B2Method and device for choosing a mode of coding

Publication Date: 2011.11.15 INTERDIGITAL MADISON PATENT HLDG

AI summary

The present invention relates to a method and a device for choosing a mode of coding from among a plurality of modes of coding for a subset of blocks included in a set of blocks. An energy function over the set of blocks for each of the modes of coding is calculated and calculation elements of the energy function are stored for at least the subset of blocks. A first mode of coding, minimizing the energy function over the set of blocks is chosen. The energy function over the subset of blocks for each of the modes of coding is thereafter calculated by using the calculation elements stored for the subset of blocks and by said subset, of a cost of coding for the modes of coding distinct from the first mode of coding. The choice of a mode of coding minimizing the energy function over the subset is then effected.