Point Cloud Gesture Positioning via Principal Component Analysis

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The existing technique for recognizing hand gestures based on a learned three-dimensional model of the human body is not stable due to accuracy issues in machine learning, leading to unreliable pointing operations on a display screen.

Innovation Solution

An information processing apparatus and method that generates a point cloud from a range image and performs principal component analysis to detect a vector corresponding to an object, allowing for accurate determination of a specified position on a display screen.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If gesture recognition is performed using learned three-dimensional model data of a human body, then the system can perform pointing operations, but the recognition accuracy is insufficient leading to unstable operation results

Engineering Contradiction:

Improvestability of pointing operationVSAvoidgesture recognition accuracy

Core Design Contradiction:

ReliabilityVSMeasurement precision

Solution Approach 1:

The patent changes the fundamental parameters of gesture recognition by transitioning from learned 3D model data to actual range image data. Specifically, it uses depth information from range images to calculate actual distances between the imaging device and the user's hand, and compares this with the distance to the display screen to determine pointing positions. This parameter change from model-based to measurement-based approach resolves the contradiction by providing both high precision and reliability.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent replaces the machine learning-based recognition system with a geometric calculation system. Instead of using learned models to infer gesture meaning, the system directly calculates spatial relationships using principal component analysis on range image data to extract hand orientation and position, then determines the pointing location through geometric intersection calculations. This substitution eliminates the accuracy limitations of machine learning while maintaining operational capability.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Ease of operation

If learned three-dimensional model data is used for gesture recognition, then the system can identify hand gestures, but the machine learning accuracy limits the precision of position specification

Engineering Contradiction:

Improvegesture-based operation capabilityVSAvoidposition specification accuracy

Core Design Contradiction:

Ease of operationVSMeasurement precision

Solution Approach 1:

The patent replaces the machine learning-based gesture interpretation system with a direct geometric measurement system. By using principal component analysis on range image point clouds, the system extracts the hand's orientation vector and calculates the intersection point with the display screen plane. This mechanical/geometric approach provides precise position specification while maintaining ease of gesture-based operation.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent transitions from 2D image processing to 3D spatial analysis by utilizing range image depth information. The system constructs a point cloud in three-dimensional space, performs principal component analysis to obtain orientation vectors in 3D space, and calculates the intersection with the display screen plane. This dimensional enhancement provides accurate position specification while preserving gesture operation ease.

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This approach enables stable and accurate determination of a specified position on a display screen, improving the reliability of pointing operations by using a range image and principal component analysis.

Implementation Method 1

a point cloud generation unit that generates a point cloud from a range image representing a range equivalent to a distance from a display screen to an object

Methodology Applied

Scientific EffectRange imaging: LIDAR

Implementation Method 2

a position determination unit that performs a principal component analysis on the generated point cloud, to detect a vector corresponding to an object relating to position specification based on an analysis result obtained by the analysis

Methodology Applied

Scientific EffectPrincipal component analysis:

Data Source

PatentUS10088910B2Information processing apparatus and information processing method

Publication Date: 2018.10.02 SONY GROUP CORP

AI summary

Provided is an information processing apparatus including: a point cloud generation unit configured to generate a point cloud from a range image representing a range equivalent to a distance from a display screen to an object; and a position determination unit configured to perform a principal component analysis on the generated point cloud, to detect a vector corresponding to an object relating to position specification based on an analysis result obtained by the analysis, and to determine a specified position, the specified position being specified on the display screen.