Mode-Symbol Encoding for Higher Compression Across Data Patterns
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing encoding methods, such as ZRLE, are inefficient when zero is not the dominant value in data, especially at high quality and low quantization rates, limiting their effectiveness in achieving high compression ratios and rapid decoding in various data types.
Innovation Solution
An encoder that identifies a mode symbol in the input data and generates distinct non-mode symbols and runs of the mode symbol, allowing for efficient compression by separating and processing these elements, enabling higher compression ratios and rapid decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If ZRLE encoding method is used, then compression efficiency is improved when zero values are dominant, but compression efficiency deteriorates when zero is not the dominant value
Solution Approach 1:
The patent changes the parameter being encoded from fixed (zero values only) to variable (most frequent symbol). The encoder dynamically identifies which symbol occurs most frequently in the data and uses that symbol for run-length encoding, rather than always encoding zero values. This makes the encoding method adaptable to different data distributions while maintaining compression efficiency.
Solution Approach 2:
The patent introduces dynamic selection of the target symbol for encoding. Instead of a static approach where zero is always the target, the system dynamically determines the most frequent symbol in the current data set and adapts the encoding process accordingly. This dynamic adaptation resolves the contradiction between optimizing for zero-dominant data and handling general cases.
2Productivity
If conventional RLE or SRLE is used, then consecutive runs of symbols are compressed, but efficiency is limited when there are not many consecutive symbols
Solution Approach 1:
The patent changes the encoding parameter from symbol value (what symbol is encoded) to frequency-based selection (which symbol to encode based on occurrence frequency). This allows the system to achieve high compression ratios by targeting the most frequent symbol regardless of whether it appears in long consecutive runs or scattered throughout the data, improving effectiveness across different data patterns.
3Speed
If encoding method focuses on specific dominant value (zero), then decoding is rapid for that case, but general applicability is reduced
Solution Approach 1:
The patent makes the encoding method universal by removing the restriction to zero values only. The same encoding algorithm works for any symbol type (numbers, bits, alphanumeric characters) and automatically adapts to whichever symbol is most frequent in the given data. This maintains rapid decoding performance while significantly broadening applicability to various symbol types and data patterns.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An encoder is provided. The encoder is operable to analyze input data (Dl) to identify at least one mode symbol therein. The encoder is operable to generate data values of a first type including non-mode symbols and data values of a second type including runs of the at least one mode symbol. Moreover, the encoder is operable to generate information that is indicative of a count of the non-mode symbols and information that is indicative of the at least one mode symbol. Furthermore, the encoder is operable to assemble or encode the information that is indicative of the at least one mode symbol, the information that is indicative of the count of the non-mode symbols, the data values of the first type including the non-mode symbols and the data values of the second type including the runs of the at least one mode symbol, to generate encoded data (E2).