Handwriting Recognition via Stroke Segmentation and Merging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Electronic devices, especially small form factor devices like smartphones and watch-type wearables, face challenges with accurate handwriting recognition due to overlapping characters, size constraints, and unbalanced strokes, leading to deteriorated recognition accuracy and increased recognition time.
Innovation Solution
An electronic device equipped with a processor and memory that receives handwriting inputs, extracts feature information, merges or separates consecutive strokes based on this information, and performs handwriting recognition, utilizing machine learning algorithms to enhance recognition speed and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If handwriting recognition is performed on overlapping characters in small input windows, then recognition accuracy deteriorates, but recognition speed is required to be fast
Solution Approach 1:
The patent segments overlapping handwriting strokes into individual stroke units and processes them separately through feature extraction and merging operations. This segmentation allows the system to handle complex overlapping patterns by breaking them down into manageable components, improving recognition accuracy without requiring excessive processing time for the entire handwriting input
Solution Approach 2:
The patent performs preliminary feature extraction from handwriting strokes before recognition. By extracting features such as stroke direction, curvature, and intersection points in advance, the system prepares processed data that accelerates the subsequent recognition process, thereby reducing overall recognition time while maintaining accuracy for overlapping characters
2Measurement precision
If conventional handwriting recognition processes all strokes uniformly, then processing time increases, but recognition accuracy is maintained
Solution Approach 1:
The patent applies different processing operations to different strokes based on their local characteristics. Some strokes are merged with adjacent strokes when they belong to the same character, while others are kept separate. This localized quality adjustment optimizes processing efficiency for each stroke individually, improving overall productivity without sacrificing recognition accuracy
Solution Approach 2:
The patent changes processing parameters dynamically based on stroke characteristics. The merging operation adjusts merging thresholds and parameters according to the specific features of each stroke, such as proximity, direction, and curvature. This parameter adaptation allows faster processing for simple strokes while maintaining accuracy for complex overlapping patterns
Data Source
AI summary
According to various embodiments, an electronic device may comprise a display, a memory for storing a machine learning algorithm related to character separation, and at least one processor, wherein the at least one processor is configured to receive a handwriting input via the display; extract feature information between a series of consecutive strokes corresponding to the handwriting input; merge or separate the strokes through the machine learning algorithm on the basis of the extracted feature information; and perform handwriting recognition on the basis of the result of the merging or separation. Various other embodiments may be provided.