Aerial Handwriting Recognition Using 2D Grayscale Trajectory Tracking

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional methods for recognizing aerial handwriting require depth images or color images, which necessitate dedicated cameras and result in slow processing speeds, especially on devices with limited processing capabilities like smart watches and smart home devices.

Innovation Solution

The method employs two-dimensional gray-scale images to detect and track the spatial trajectory of a fingertip, utilizing a bounding box tracking method and combining 8-direction feature detection with a deep neural network (DNN) for improved accuracy and reduced computation, enabling faster and more accurate aerial handwriting recognition.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If depth images or color images are used for aerial handwriting recognition, then measurement precision is improved, but processing speed deteriorates and device complexity increases

Engineering Contradiction:
Improvehandwriting recognition accuracyVSAvoidprocessing speed
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent extracts only the essential 2D grayscale image data from what would otherwise require full color or depth information. By converting to grayscale and using 2D images instead of 3D depth images, the system retains sufficient information for fingertip trajectory detection while dramatically reducing processing complexity and speed requirements

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent uses standard 2D camera hardware that is already available in most devices, replacing the need for expensive dedicated depth cameras or specialized imaging equipment. This makes aerial handwriting recognition accessible on common smartphones and tablets without requiring specialized hardware

Inventive Principle:
Principle #27Cheap short-living objects (Disposable)

2Measurement precision

If depth images or color images are used for aerial handwriting recognition, then measurement precision is improved, but device complexity increases

Engineering Contradiction:
Improvehandwriting recognition accuracyVSAvoidcamera system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent makes the existing 2D camera serve multiple functions - it can capture both regular photos and aerial handwriting recognition data using the same hardware component. This eliminates the need for dedicated depth cameras or specialized imaging equipment, reducing device complexity while maintaining recognition capability

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent creates a simplified 2D grayscale representation of the handwriting space that captures essential trajectory information without requiring complex 3D depth imaging. This copied representation is sufficient for recognition purposes and dramatically simplifies the required hardware

Inventive Principle:
Principle #26Copying

3Measurement precision

If conventional aerial handwriting recognition methods are used, then recognition accuracy is maintained, but processing speed deteriorates on devices with limited processing capabilities

Engineering Contradiction:
Improverecognition accuracyVSAvoidprocessing speed
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent segments the image processing task into efficient 2D grayscale frame analysis instead of processing full color or depth images. By dividing the problem into discrete 2D frames that can be processed sequentially, the system achieves faster processing speeds suitable for devices with limited computing power while maintaining trajectory detection accuracy

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3754542B1Method and apparatus for recognizing handwriting in air, and device and computer-readable storage medium
Publication Date: 2024.02.21 BEIJING BAIDU NETCOM SCI & TECH CO LTD
  • EP3754542B1 patent drawingFigure 1
  • EP3754542B1 patent drawingFigure 2
  • EP3754542B1 patent drawingFigure 3~4

AI summary

Provided are a method and apparatus for recognizing handwriting in the air, and a device and a computer-readable storage medium. The method comprises detecting a palm region of a user in a two-dimensional grayscale image and detecting, based on the palm region, a fingertip in the two-dimensional grayscale image. The method further comprises determining a spatial trajectory of the fingertip based on a group of two-dimensional grayscale images, and recognizing the handwriting content of the user on the basis of the spatial trajectory. According to the embodiments of the present disclosure, by using two-dimensional grayscale images to recognize and track a spatial trajectory of a fingertip, the speed of recognizing handwriting in the air can be increased, and the requirement for the processing performance of a device is relatively low, and relatively high accuracy can also be ensured.