Touch Screen Character Recognition via Stroke Segmentation Graph
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Mobile devices with touch-sensitive screens face challenges in character recognition due to space constraints, where characters may not follow each other sequentially and can overlap or appear before earlier characters, requiring a method to accurately interpret non-sequential and overlapping touch inputs.
Innovation Solution
A character recognition method that creates a stroke-segmentation graph with nodes and edges to detect touch inputs, defines candidate characters with scores, and explores paths through a candidate-character graph to identify the most likely character string, accounting for spatial, temporal, and language constraints to handle multi-stroke inputs and space-constrained environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a traditional keyboard or character input area is displayed on a mobile device, then users can enter alphanumeric characters, but the device requires significant display space which is not available on small screens
Solution Approach 1:
The patent transitions from 2D keyboard layouts to stroke-based input that utilizes temporal sequencing and spatial trajectories. Users draw characters freehand on the screen, converting the input method from discrete key presses in fixed positions to continuous stroke movements that can occur anywhere on the display area, effectively adding temporal and trajectory dimensions to the input process.
Solution Approach 2:
The patent extracts the essential function of character input from the physical keyboard structure. Instead of requiring a predefined keyboard layout, the system extracts only the necessary touch detection and stroke recognition capabilities, allowing character entry through freehand drawing without the constraint of fixed key positions or extensive display real estate.
2Adaptability or versatility
If users are restricted to writing characters in a specific sequential order on the touch screen, then the recognition process is simpler, but the user experience is less flexible and natural
Solution Approach 1:
The patent segments the character recognition process into distinct computational stages: stroke detection, stroke segmentation into candidate characters, graph construction with nodes and edges, path exploration, and score computation. This segmentation allows the system to handle complex overlapping and non-sequential inputs by breaking down the recognition task into manageable, modular processing steps.
Solution Approach 2:
The patent introduces a stroke-segmentation graph as an intermediary data structure between raw touch inputs and final character recognition. The graph with its nodes, edges, and paths serves as a mediator that systematically explores multiple possible character interpretations, allowing the system to resolve ambiguous overlapping strokes through structured path analysis rather than direct pattern matching.
3Measurement precision
If the system only considers sequential stroke inputs, then the processing is faster and simpler, but it cannot accurately recognize characters drawn in non-sequential or overlapping patterns
Solution Approach 1:
The patent performs preliminary actions by pre-defining all possible stroke-segmentation graph structures, candidate character sets, and path exploration algorithms before actual character recognition occurs. The system pre-computes potential character candidates from individual strokes and pre-establishes the graph framework, enabling rapid processing during actual use even when dealing with complex non-sequential inputs.
Solution Approach 2:
The patent implements dynamic path exploration through the stroke-segmentation graph, where the system adaptively traverses different paths based on the specific input pattern encountered. The graph structure allows dynamic adjustment of the recognition process, exploring only relevant paths and candidates based on the actual strokes provided, rather than evaluating all possible character combinations uniformly.
Data Source
AI summary
Some embodiments of the invention provide a novel method for recognizing characters that are input through touch strokes on a touch-sensitive sensor (e.g., a touch-sensitive display screen or a touch-sensitive surface) of a device (e.g., a mobile device, a remote control, a trackpad, etc.). In some embodiments, the sensor has a space-constrained area for receiving the touch input. In some embodiments, the method places no limitations on where the user can write in the space provided by the device. As such, successive characters might not follow each other in the space. In fact, later characters might overlap earlier characters or they might appear before earlier characters.