Radiation Area Extraction via Candidate Line Combinations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for extracting radiation areas from image data often result in erroneous extraction due to reliance on edge point information alone, leading to incorrect identification of radiation areas, especially when non-edge segments in the image contain straight parts.
Innovation Solution
A method that generates combinations of candidate lines from image data, evaluates these combinations based on their relationships, and extracts the contour lines of the radiation area, reducing erroneous extraction by considering the relationships between candidate lines and using evaluated values to determine the accurate radiation area.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If line segments representing edges of the radiation area are extracted based only on edge point information, then the extraction process is simple, but the radiation area is erroneously extracted when area segments other than edges include straight parts
Solution Approach 1:
The patent introduces candidate lines as an intermediary element between edge points and the final radiation area extraction. Instead of directly extracting line segments from edge points, the method first generates multiple candidate lines that pass through edge points, then evaluates and selects the most appropriate candidate lines based on geometric relationships and evaluation values. This intermediary step filters out false positives from straight parts in non-edge areas while preserving true radiation area boundaries.
Solution Approach 2:
The patent performs preliminary generation of multiple candidate lines through edge points before final extraction. By pre-generating candidate lines with different configurations and evaluating them in advance, the method prepares a set of possible solutions that can be systematically assessed. This preliminary action allows the system to compare multiple hypotheses about radiation area boundaries and select the correct one, preventing erroneous extraction from straight parts in other areas.
2Reliability
If multiple candidate lines are generated and evaluated based on their relationships, then the accuracy of radiation area extraction is improved, but the processing complexity increases
Solution Approach 1:
The patent changes the evaluation parameters from simple edge point presence to comprehensive geometric relationship assessment. The evaluation process considers multiple parameters including the number of edge points on each candidate line, the angular relationships between adjacent candidate lines, and the spatial configuration of line combinations. By using these refined parameters, the system can distinguish true radiation area boundaries from false straight parts, improving accuracy despite increased processing requirements.
Solution Approach 2:
The patent segments the candidate line evaluation process into distinct stages: generating individual candidate lines through edge points, evaluating each candidate line's geometric properties, combining candidate lines into potential boundary configurations, and finally selecting the optimal combination. This segmentation allows the complex evaluation to be broken down into manageable steps, making the increased processing complexity more tractable while maintaining high extraction accuracy.
Data Source
AI summary
A radiation area extracting method for extracting a radiation area from image data includes generating combinations of at least two candidate lines from an area of the image data, an evaluating step of obtaining evaluated values of the combinations of at least two candidate lines, and an extracting step of extracting one of the combinations of at least two candidate lines as a contour line of the radiation area on the basis of the evaluated values.