Touchscreen Text Input via Gesture Recognition and Direction Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing text input methods on touchscreens, such as those based on Fitts' Law, are limited by the need to connect characters sequentially, resulting in upper speed limits and inefficiencies, particularly when using pointing methods like unistroke or multistroke inputs.
Innovation Solution
A method that combines pointing and recognition techniques by using gestures on an input template, where characters are subdivided into subsets, allowing selection through direction vectors and shape recognition, thereby reducing the need for sequential character connection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If characters are connected sequentially using pointing methods (unistroke或多istroke), then text input can be performed on a touchscreen without a physical keyboard, but the input speed is limited by Fitts' Law due to the path length between characters
Solution Approach 1:
The input method is divided into two distinct phases: a pointing phase where the user positions the input medium on a starting character, and a recognition phase where the system recognizes the gesture pattern (loop, wave, spiral, etc.) to determine the intended character. This segmentation eliminates the need to physically connect to each character, thereby overcoming Fitts' Law limitations while maintaining ease of operation.
2Loss of time
If a Look-Up function is used to provide word completions after inputting a determined number of characters, then some input time may be saved, but the upper speed limit according to Fitts' Law still applies to the characters that must be connected
Solution Approach 1:
The invention extracts the character recognition task from the physical connection process. Instead of requiring the user to connect to each character, the system extracts only the essential information (starting position and gesture pattern) and uses recognition algorithms to identify the intended character, thereby eliminating the time-consuming connection phase while maintaining input accuracy.
3Manufacturing precision
If all characters including vowels and last characters are explicitly input by connection, then complete word input is achieved, but the input path becomes unnecessarily long reducing efficiency
Solution Approach 1:
The system performs preliminary recognition of the gesture pattern during the input process itself. By recognizing the characteristic patterns (loops for closed characters, waves for open characters, spirals for specific characters) as they are drawn, the system can determine the intended character without requiring complete traversal of the character path, thereby achieving both completeness and efficiency.
Data Source
AI summary
The invention relates to a method for the detection of the selection of a character of a character string to be input from a character set on an input surface, wherein the selection of at least one character of the character string is detected by evaluating a direction vector and/or by recognizing a shape of a gesture which is input on the input surface.