Text Image Symbol Matching Using Dynamic Distance Caching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional pattern matching based coding and decoding systems in bi-level image compression, such as JBIG2, face increased processing time and decreased bit rate due to inefficient matching operations.
Innovation Solution
A dynamic symbol caching method using multiple distances and references to determine symbol matches, where a first distance (XOR) is used for coarse matching and a second distance (WXOR) for fine matching, reducing computational overhead and improving bit rate.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single distance metric (WXOR) is used for symbol matching, then matching precision and compression ratio are improved, but processing time increases and computational complexity increases
Solution Approach 1:
The symbol matching process is segmented into two distinct stages: a coarse matching stage using XOR distance metric and a fine matching stage using WXOR distance metric. The coarse stage performs initial filtering to identify candidate matches, while the fine stage verifies these candidates with higher precision. This segmentation reduces overall processing time by avoiding unnecessary fine-stage computations for clearly unmatched symbols.
Solution Approach 2:
The coarse matching using XOR distance metric serves as a preliminary action that pre-filters the symbol dictionary before performing the more computationally intensive WXOR matching. By performing this preliminary sorting and filtering in advance, the system reduces the number of full precision matching operations required, thereby decreasing total processing time while maintaining high matching precision.
2Measurement precision
If multiple distance metrics are used for symbol matching, then matching accuracy is improved, but device complexity and computational overhead increase
Solution Approach 1:
The computational complexity is segmented and distributed across two functional stages: coarse filtering (XOR) and fine verification (WXOR). The XOR stage handles the bulk of computational operations for initial matching, while the WXOR stage performs only the necessary precise verification for candidate matches. This segmentation reduces overall computational complexity compared to applying only WXOR to all symbols.
Solution Approach 2:
The system applies partial action by using the less computationally intensive XOR metric for the majority of matching operations (coarse matching), reserving the more intensive WXOR metric only for a subset of candidate symbols that passed the coarse filter. This partial application of the more complex metric reduces overall computational overhead while maintaining high matching accuracy for the critical cases.
3Productivity
If comprehensive symbol matching is performed, then compression ratio is improved, but processing speed decreases
Solution Approach 1:
The coarse matching stage performs preliminary identification of potential symbol matches using fast XOR distance computation. This preliminary action creates a reduced set of candidate symbols that require further verification. By performing this initial sorting and filtering before comprehensive matching, the system maintains high compression ratio through thorough verification of candidates while improving processing speed by reducing the total number of full precision matching operations.
Solution Approach 2:
The matching process is divided into a fast coarse matching phase using XOR and a more precise fine matching phase using WXOR. The coarse phase processes all symbols quickly to identify candidates, while the fine phase processes only the relevant candidates in detail. This segmentation enables the system to achieve both high compression ratio (through comprehensive candidate verification) and improved processing speed (by avoiding exhaustive fine matching of all symbols).
Data Source
AI summary
An apparatus of a text image coding and decoding system includes a matching unit to compute a first distance between a symbol of a text image with a reference symbol of a symbol dictionary, to determine whether the symbol matches with the reference symbol according to the first distance and a first reference, and to compute a second distance between the one of the symbols with the reference symbol if the symbol does not match with the reference symbol according to the first distance and the first reference, and to determine whether the symbol matches with the reference symbol according to the second distance and a second reference.