Free Handwriting Text Recognition via Model Data Normalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computing devices face challenges in reliably recognizing and editing text handwriting input in a free handwriting mode due to difficulties in determining when new ink strokes affect previous content, leading to instability and limited editing capabilities.
Innovation Solution
A method that processes text handwriting by generating model data representing text lines, applying transformation functions to normalize input into a structured format, aligning with document patterns, and updating model data to preserve previous recognition states, enabling efficient editing and improved recognition reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If text handwriting is input in a free handwriting mode without constraints, then user flexibility and ease of operation are improved, but text recognition reliability and stability deteriorate
Solution Approach 1:
The patent segments the free handwriting input into multiple text line candidates, each representing a possible interpretation of the handwriting strokes. By dividing the recognition problem into separate text line hypotheses, the system can evaluate each candidate independently and select the most likely interpretation, thereby improving recognition reliability while preserving free handwriting input flexibility
Solution Approach 2:
The patent changes the parameter representation of handwriting input by transforming raw stroke coordinates into multiple possible text line configurations with associated probability scores. This parameter transformation allows the system to handle the variability of free handwriting mode systematically, converting an unstructured input problem into a structured evaluation of multiple hypotheses
2Ease of operation
If text handwriting is input in a free handwriting mode, then ease of operation is improved, but editing capability deteriorates
Solution Approach 1:
The patent performs preliminary action by pre-processing free handwriting input into multiple text line candidates with probability rankings before editing operations are needed. This preliminary structuring creates a foundation that enables subsequent editing operations to work on organized text line hypotheses rather than raw strokes, thereby improving editing capability while maintaining free handwriting input freedom
Solution Approach 2:
The patent introduces an intermediary representation layer between free handwriting input and editing operations. This intermediary consists of multiple text line candidates with probability scores, which serves as a bridge that allows editing operations to manipulate structured text hypotheses while the underlying free handwriting strokes remain unchanged, thus enabling both freedom in input and capability in editing
3Reliability
If multiple text line candidates are generated and processed, then text recognition reliability is improved, but device complexity increases
Solution Approach 1:
The patent applies partial action by generating multiple text line candidates but processing only the top-ranked candidates based on probability scores. Instead of exhaustively processing all possible interpretations, the system focuses computational resources on the most likely text line hypotheses, thereby improving recognition reliability while controlling processing complexity through selective evaluation
Data Source
Figure 1A~1B
Figure 2~3
Figure 4
AI summary
The invention concerns a method comprising: detecting strokes of digital ink (IN) input on a computing device in a free handwriting format (FT1); detecting a text block (BL1) from said strokes; performing text recognition on each text line of said text block, including extracting text lines from the text block (BL1) and generating model data that associate each stroke of the text block with a character, a word and a text line of the text block (BL1); normalising each text line from the free handwriting format (FT1) into a structured format (FT2) to comply with a document pattern (200). The normalisation may comprise for each text line: computing a transform function to transform said text line into the structured format; applying the transform function to the text line; and updating the model data of said text line based on the transform function.