Skeleton-Based Image Search for Human Pose Query Input
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems struggle with user-friendly methods for inputting search queries to find images of people in specific states, such as poses and behaviors, often requiring time-consuming preparation of query images and failing to adapt to hidden body parts.
Innovation Solution
An image processing apparatus and method that includes an image acquisition unit, a skeleton structure detection unit, and a feature value computation unit to detect and compute two-dimensional skeleton structures, enabling search for similar states based on feature value similarity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a query image is prepared by manual selection and processing, then the search accuracy for specific poses and behaviors is improved, but the time and operational complexity required for query input increases significantly
Solution Approach 1:
The system creates a synthesized query image by automatically combining a target object image with pose information from reference images, eliminating the need for manual query image preparation. This copying approach allows the system to generate search queries rapidly while maintaining accuracy by reusing proven pose data from the database.
Solution Approach 2:
The system pre-processes and stores pose information from reference images in advance, extracting key pose parameters and storing them in the database. This preliminary action enables rapid query generation during search operations without requiring real-time manual intervention or complex processing.
2Measurement precision
If a complex query image preparation process is used to specify precise poses and behaviors, then the search precision is improved, but the ease of operation deteriorates
Solution Approach 1:
The system copies and reuses pre-extracted pose information from reference images to construct query images automatically. This eliminates the need for users to manually specify complex pose parameters, maintaining high search precision while dramatically simplifying the user interface to simple keyword or image-based queries.
Solution Approach 2:
The system introduces an intermediary synthesized query image that bridges the gap between simple user input and complex search requirements. This intermediary construct automatically incorporates precise pose information from the database while being generated from simple user-provided inputs, hiding the complexity from the user.
3Adaptability or versatility
If traditional image search methods are used without skeleton detection, then the device complexity is reduced, but the ability to detect and search for specific poses and behaviors is insufficient
Solution Approach 1:
The system extracts only the essential pose information from full reference images by detecting skeleton structures and key pose parameters. This extraction approach enables pose-based search functionality without requiring the entire image processing pipeline, reducing unnecessary complexity while maintaining detection capability.
Solution Approach 2:
The system segments the image processing task into distinct modules: reference image processing, pose detection, feature extraction, and query synthesis. This segmentation allows each component to be optimized independently and enables the system to achieve sophisticated pose detection without overwhelming complexity through modular design.
Data Source
AI summary
The present invention provides an image processing apparatus (100) including an image acquisition unit (101) that acquires a query image, based on an input keyword, a skeleton structure detection unit (102) that detects a two-dimensional skeleton structure of a person included in the query image, a feature value computation unit (103) that computes a feature value of the detected two-dimensional skeleton structure, and a search unit (105) that searches, based on a degree of similarity of the computed feature value, for an analysis target image including a person in a state similar to a state of a person included in the query image from the analysis target image.