Error-resilient image sequence coder using independent wavelet transforms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image sequence coding technologies, such as SPIHT and variable length coders, are sensitive to channel errors, leading to error propagation and degradation in image reconstruction due to their sensitivity to bit errors in significance maps.
Innovation Solution
The method involves performing a wavelet transform on each frame independently, constructing spatial trees, and creating a progressive significance map with a sum and complementary part to encode frames as multiple bitstreams, ensuring that errors only affect the specific coding unit and not the entire sequence, thereby enhancing error resilience.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If a single significance map is used for the entire image sequence, then compression efficiency is improved, but error propagation affects the entire sequence when a bit error occurs
Solution Approach 1:
The patent divides the image sequence into multiple independent coding units (frames or groups of frames), each with its own significance map. This segmentation prevents error propagation from affecting the entire sequence, as errors are confined to individual coding units. The segmentation is achieved by processing each frame independently through wavelet transform and encoding with separate significance maps, thereby resolving the contradiction between compression efficiency and error resilience.
2Reliability
If error resilient techniques are applied to each frame independently, then error propagation is inhibited, but the complexity of encoding increases
Solution Approach 1:
By segmenting the image sequence into independent coding units, the patent enables parallel processing of each unit. This reduces encoding complexity compared to processing the entire sequence sequentially with a single significance map, while maintaining error resilience through independent error confinement in each segment.
Solution Approach 2:
The patent performs wavelet transform and significance map generation for each coding unit independently before encoding. This preliminary action prepares error-resilient data structures in advance, simplifying the subsequent encoding process and reducing overall complexity while ensuring error resilience.
3Reliability
If progressive significance map is used with sum and complementary parts, then error propagation is reduced, but the bitstream length increases
Solution Approach 1:
The progressive significance map is divided into sum and complementary parts, which are encoded separately for each coding unit. This segmentation allows error propagation control while managing bitstream length by confining errors to specific segments and units, rather than affecting the entire bitstream.
Solution Approach 2:
The patent applies progressive significance map encoding selectively to coding units where error resilience is most beneficial, rather than uniformly to all data. This partial action approach controls error propagation while minimizing the increase in bitstream length by focusing computational resources on critical segments.
Data Source
AI summary
A method, computer system, and a computer program product is disclosed for encoding a sequence of frames as one or more compressed bitstreams, the method includes a processor obtaining a sequence of frames in an order and performing a wavelet transform on each frame of the sequence of frames independently in the order. The processor constructs a significant points list and an insignificant sets list, creating one or more spatial trees comprised of branch nodes of the list of insignificant sets for each frame of the sequence of frames. The processor initializes the significant points list and the list of insignificant sets for each frame and identifies a most significant bit in each frame and for each frame, selects at least one coding unit and creates a progressive significance map for coding. The processor writes a global header and a local header for the compressed bitstreams for each coding unit.