Body Orientation Estimation Using Left-Right Motion Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating body orientation in images, such as those using motion vectors, often result in incorrect recognition when a person slides sideways or rotates, leading to inaccurate orientation estimation.
Innovation Solution
A method that extracts and compares the moving amounts of left and right sites within a person's region in images taken at different times, using relation information to estimate body orientation based on differences and change amounts, thereby accurately accounting for rotational and lateral movements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If motion vectors are used to estimate body orientation, then the estimation process is simple, but wrong recognition occurs during sideways movement and rotation
Solution Approach 1:
The patent divides the person's body into multiple regions (head region, upper body region, lower body region) and extracts motion vectors from each region separately. This segmentation allows the system to analyze movement patterns of different body parts independently, enabling accurate detection of rotation and sideways movement by comparing relative motions between regions, thereby resolving the contradiction between simple processing and accurate recognition.
2Reliability
If multiple body regions are analyzed to improve orientation accuracy, then recognition accuracy improves, but processing complexity increases
Solution Approach 1:
The patent segments the body into three main regions (head, upper body, lower body) and extracts motion vectors from each region. By comparing the motion vectors between these segmented regions, the system can determine body orientation and detect rotation/sideways movement patterns without requiring complex full-body analysis, thus achieving accurate orientation estimation with manageable processing complexity.
Solution Approach 2:
The patent focuses on analyzing motion vectors from key body regions rather than processing every pixel or detail of the entire body. By selecting specific regions of interest (head, upper body, lower body) and their motion characteristics, the system achieves sufficient orientation accuracy without the excessive computational burden of complete body analysis.
Data Source
AI summary
A method for estimating body orientation includes: executing extraction processing for extracting a person region from images; executing an identification processing for identifying a left site and a right site in the person region; executing calculation processing for calculating a left moving amount and a right moving amount; and executing estimation processing for estimating body orientation in the person region based on a first difference and relation information, the first difference being a difference between the left moving amount and the right moving amount, the relation information including first and second relation information, the first relation information indicating a relationship between the first difference and a change amount when the left moving amount is larger than the right moving amount, and the second relation information indicating a relationship between the first difference and the change amount when the right moving amount is larger than the left moving amount.