Radiation Imaging Object Recognition Through Optical Candidate Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing radiation imaging systems face challenges in accurately determining the imaging position of an object due to the presence of non-target individuals or body parts in optical images, necessitating a method to narrow down the object candidates.
Innovation Solution
An information processing apparatus that includes a first recognizing unit to identify a region of interest and a second recognizing unit to identify object candidates in optical images, integrating these results to determine the object of interest, and a consistency judging unit to ensure alignment with imaging orders.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If an optical imaging means captures the entire scene in the field of view, then the imaging scene can be fully captured, but multiple object candidates including non-target individuals appear in the optical image
Solution Approach 1:
The patent segments the object identification task into multiple stages: first identifying all object candidates in the optical image, then filtering these candidates based on spatial relationship with the radiation imaging target. This segmentation allows full scene capture while systematically eliminating non-target objects through multi-criteria evaluation.
Solution Approach 2:
The patent introduces an intermediary filtering mechanism that uses spatial relationship information as a mediator between the optical image and radiation image. By evaluating the spatial correspondence between objects in the optical image and the radiation imaging target, the system identifies and eliminates non-target candidates while maintaining comprehensive scene coverage.
2Ease of operation
If image analysis is performed to determine imaging position, then imaging position can be identified, but accuracy is insufficient across diverse objects and imaging conditions
Solution Approach 1:
The patent implements feedback by continuously evaluating the spatial relationship between identified objects and the radiation imaging target. The system uses this feedback to refine and adjust the object identification process, improving accuracy across diverse objects and imaging conditions through iterative verification.
Solution Approach 2:
The patent changes the evaluation parameters from simple object detection to a multi-dimensional assessment including spatial relationship, object type classification, and correspondence with radiation imaging target. This parameter transformation enables accurate identification across diverse objects and imaging conditions.
3Quantity of substance
If multiple object candidates are present in the optical image, then comprehensive scene information is captured, but it becomes difficult to narrow down the actual imaging target
Solution Approach 1:
The patent segments the object candidate list into potential targets and non-targets by evaluating spatial relationships. This segmentation transforms the difficult task of identifying the single target from multiple candidates into a systematic filtering process based on spatial correspondence with the radiation imaging target.
Solution Approach 2:
The patent extracts and removes non-target objects from the list of object candidates by evaluating their spatial relationship with the radiation imaging target. This extraction process maintains comprehensive scene information while isolating the actual imaging target through criteria-based filtering.
Data Source
AI summary
An information processing apparatus is provided that comprises: a first recognizing unit configured to recognize a region of interest using an optical image including an object, which is obtained in an imaging scene of a radiation image, a second recognizing unit configured to recognize candidate information indicating a candidate of the object which is an imaging-target of the radiation image using the optical image; and a determining unit configured to determine object information indicating the object which is the imaging-target using at least one of recognition results by the first recognizing unit and the second recognizing unit.