Recursive Reparameterization Filtering for Electronic Ink Denoising
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for denoising electronic ink, such as low-pass filtering, often distort the shape of pen strokes and are noticeable to users due to their iterative processing requirements, which makes it impossible to hide the processing from the user, especially when applied to real-time data from hand-held devices.
Innovation Solution
The implementation of recursive reparameterization filtering, which adjusts the sampling rate and performs weighted averaging of secondary points to produce output samples while reparameterizing the trace, allowing for on-the-run denoising without noticeable shrinkage or distortion, enabling real-time processing and rendering of traces.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If low-pass filtering is applied to denoise electronic ink, then noise is reduced, but the shape of pen strokes is distorted and critical fragments are lost
Solution Approach 1:
The patent applies adaptive parameterization by dynamically adjusting the sampling rate and filter coefficients based on local signal characteristics. The system changes parameters such as the number of points to average and the weighting factors according to the detected noise level and stroke characteristics, allowing the filter to adapt to different writing styles and conditions while preserving stroke shape and critical fragments.
Solution Approach 2:
The filtering process is made dynamic through continuous adaptation to changing signal conditions. The system monitors the input signal characteristics and adjusts the filtering parameters in real-time, transitioning from static fixed-parameter filtering to dynamic parameter adjustment that responds to local variations in stroke width, curvature, and noise level.
2Manufacturing precision
If iterative processing is used to reduce shrinkage and distortion, then visual quality improves, but processing becomes noticeable to users in real-time
Solution Approach 1:
The system performs preliminary filtering actions continuously in the background as strokes are being written, rather than waiting for the complete stroke to be finished. By applying filtering incrementally and preparing the processed output in advance, the system achieves high visual quality without requiring noticeable iterative processing delays that would be perceptible to users during real-time writing.
3Reliability
If filtering is applied to hand-held devices with motor tremor noise, then denoising is achieved, but processing delays become noticeable
Solution Approach 1:
The patent segments the filtering process into small incremental steps that are applied continuously rather than as a single large operation. By dividing the filtering task into manageable segments that process small portions of the signal at a time, the system achieves effective denoising of hand-held device noise while maintaining minimal processing delay through progressive refinement.
Data Source
AI summary
A method and system for on-the-run processing of source data is disclosed. In one embodiment, the system includes an input device to trace the moving object, such as a pen, an on-the-run filter, coupled to the input device, to perform denoising of the source data without noticeable shrinkage of the trace using recursive reparameterization filtering, and a storage device, coupled to the on-the-run filter, to compress and store the data.