Arbitrary Shape Transform Zero Coefficient Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video compression techniques face inefficiencies in coding boundary blocks of arbitrarily shaped video objects, particularly due to the scattering of zero coefficients, which reduces coding efficiency and energy compaction, and require additional computations and information signaling.
Innovation Solution
A method that generates zero coefficients at known positions in the DCT block using a binary mask to determine the number of masked pixels and apply a padding scheme, allowing for the skipping of all zero coefficients during encoding and decoding, thereby improving coding efficiency and energy compaction without extra computational burden.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional DCT coding is applied to boundary blocks with masked pixels, then the transform can be performed, but zero coefficients are scattered throughout the DCT block reducing coding efficiency and energy compaction
Solution Approach 1:
The patent applies preliminary action by pre-determining the positions of zero coefficients based on the binary mask before performing the DCT transform. The padding scheme is applied in advance to arrange masked pixels in a way that guarantees zero coefficients will be generated at specific positions, allowing for efficient skipping during encoding without requiring complex runtime calculations.
Solution Approach 2:
The patent applies local quality by treating different regions of the boundary block differently based on whether pixels are masked or unmasked. The padding scheme selectively assigns values to masked pixels based on their specific positions and the corresponding binary mask values, creating localized patterns that produce zero coefficients at predictable locations rather than uniformly treating all pixels the same.
2Measurement precision
If additional computations are performed to handle boundary blocks with masked pixels, then coding accuracy improves, but computational complexity and processing time increase
Solution Approach 1:
The patent reduces computational complexity by performing all necessary preparations before the DCT transform. The binary mask is used to pre-determine padding patterns and zero coefficient positions, eliminating the need for complex runtime calculations during encoding. The decoder can then directly reconstruct using these pre-computed arrangements without additional computational burden.
Solution Approach 2:
The patent applies self-service by using the binary mask information to automatically determine both the padding scheme and the positions of zero coefficients. The system leverages the existing mask data to serve dual purposes: guiding the padding arrangement and predicting zero coefficient locations, thereby eliminating the need for separate computation steps and reducing overall processing complexity.
3Measurement precision
If information signaling is added to indicate masked pixel positions, then decoding accuracy improves, but the bitstream size and transmission requirements increase
Solution Approach 1:
The patent applies universality by making the binary mask serve multiple functions simultaneously. The same mask data that identifies masked pixels is also used to determine the padding scheme and predict zero coefficient positions. This multi-functional use of the mask eliminates the need for separate signaling mechanisms, as the mask itself carries all necessary information for both encoding and decoding operations.
Solution Approach 2:
The binary mask performs self-service by providing all necessary information for the encoding process without requiring additional signaling. The mask values directly determine which pixels receive padding, what padding pattern to apply, and where zero coefficients will be generated. This self-contained approach eliminates the need for extra bitstream data while maintaining full decoding accuracy.
Data Source
AI summary
To encode an arbitrary shape, a processor is configured to generate a rectangular block of transform coefficients by applying a discrete separable unitary transform to a rectangular boundary block of a video object. The boundary block includes a masked pixel and an unmasked pixel, the transform coefficients include a number of zero-valued transform coefficients greater than or equal to a number of masked pixels, and the transform coefficients include at least one non-zero-valued transform coefficient. The processor is also configured to generate an encoded output by compressing only the non-zero-valued transform coefficients.