Image Search Using Partial Pose and Appearance Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image search technologies face accuracy issues when searching for specific individuals in images, especially when partial information about the person's pose and appearance is not accurately represented in the search query, leading to decreased efficiency in finding desired images.
Innovation Solution
An image processing apparatus and method that acquire and utilize search queries including pose and appearance information, specifically using partial pose and appearance data to search for intended images from a plurality of reference images, enhancing the accuracy of the search process by considering detailed characteristics such as wearing objects and their positions on the body.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If an image search is performed using all available pose and appearance information, then search accuracy is improved, but the labor required to find a suitable search query image increases significantly
Solution Approach 1:
The patent segments the search query image into multiple regions of interest (e.g., head, body, clothing areas) and selectively extracts pose and appearance information from specific regions. This allows the system to use partial but relevant information from the image without requiring manual verification of all image qualities, thus maintaining search accuracy while reducing the labor needed to prepare search queries.
2Ease of operation
If partial pose and appearance information is used for image search, then the labor required to prepare search queries is reduced, but search accuracy may decrease
Solution Approach 1:
The patent applies local quality by assigning different levels of information extraction to different regions of the search query image. Critical regions (e.g., face for identity, clothing for appearance) are analyzed in detail to extract accurate pose and appearance information, while less critical regions are processed with simpler methods. This ensures that the partial information used maintains sufficient accuracy for effective search while reducing overall processing complexity.
3Measurement precision
If comprehensive pose and appearance information is extracted from the search query, then the precision of identifying desired images is improved, but the processing complexity increases
Solution Approach 1:
The patent implements dynamic processing by adaptively adjusting the level of detail in pose and appearance information extraction based on the specific search context and image characteristics. The system dynamically selects which regions to analyze and what level of detail to extract, rather than uniformly processing all images with the same complexity. This reduces average processing complexity while maintaining identification precision when needed.
Data Source
AI summary
To improve accuracy of processing of searching for an image including a desired person from a plurality of images, the present invention provides an image processing apparatus 10 including: an acquisition unit 11 that acquires a search query including pose information indicating a pose of a body of a person and appearance information indicating an appearance of a person; and a search unit 12 that searches for an intended image from a plurality of reference images by using partial pose information indicating a pose of a part of a body of a person in the pose information included in the search query, and partial appearance information being a part of the appearance information included in the search query.