Image Search Apparatus Using Segmented Extraction for Orientation Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image selection methods, such as those disclosed in Japanese Patent Application Laid-Open No. 2004-70717, often include unnecessary images like those with subjects facing sideways or at small sizes, leading to increased user effort in selecting desired photographs.
Innovation Solution
An image searching apparatus that reads image data linked with identification information, allows input of search target identification and characteristic images, and uses extraction and comparison sections to filter images showing subjects with predefined outward characteristics, such as facing forward and at a specific size, thereby reducing computation and presenting only relevant images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If image selection is performed based only on identification information, then all images containing the subject are extracted, but unnecessary images such as those with subjects facing sideways or backwards are also included, increasing user selection effort
Solution Approach 1:
The image extraction process is divided into two sequential stages: first extracting images containing the subject based on identification information, then further filtering these extracted images based on outward characteristics. This segmentation allows the system to progressively narrow down results without overwhelming the user with all possible images at once.
Solution Approach 2:
The system performs preliminary filtering by outward characteristics (such as face orientation and size) before presenting images to the user. This preliminary action eliminates unnecessary images in advance, so users only need to review relevant images rather than filtering through all images containing the subject.
2Reliability
If all images containing subjects are extracted, then comprehensive results are provided, but images with subjects at extremely small sizes are also included, forcing users to handle additional selection
Solution Approach 1:
The system introduces additional filtering parameters (outward characteristics such as face orientation, image size, and subject size) to the extraction process. By changing the parameters used for image selection, the system maintains reliability by still extracting all images containing the subject, while simultaneously reducing user selection time by filtering out images that don't meet specific parameter criteria.
3Measurement precision
If outward characteristic comparison is performed on all extracted images, then precise image selection is achieved, but computation amount increases significantly
Solution Approach 1:
The comparison process is segmented into two phases: first comparing outward characteristics only against extracted images (not all original images), then performing detailed analysis only on images that pass the initial filtering. This segmentation reduces the total number of comparisons needed while maintaining measurement precision for the final results.
Solution Approach 2:
The system performs partial comparison by applying outward characteristic filtering to only the subset of images already extracted by identification information, rather than comparing against the entire image database. This partial action achieves sufficient precision for most cases without the excessive computation that would result from comparing all possible images.
4Ease of operation
If multiple filtering criteria are applied, then only relevant images are presented, but the complexity of the searching system increases
Solution Approach 1:
Multiple filtering criteria (identification information and outward characteristics) are merged into a unified two-stage extraction process. The system combines these different types of filters in sequence, where the output of one stage becomes the input of the next, thereby simplifying the overall system architecture while still applying multiple filtering criteria.
Data Source
AI summary
An image searching apparatus comprising:a readout section which reads out image data in which an image of a subject and identification information capable of uniquely identifying the subject are related;an input section for inputting identification information indicating a search target subject and an outward characteristic image of the search target subject;a first extraction section which extracts images to which the identification information is related from the image data read out by the readout section;and a second extraction section which compares the images extracted by the first extraction section with the outward characteristic image to extract images which include the outward characteristic image.