Reaction-Diffusion Image Segmentation Reducing Computational Load
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional region extraction processing using reaction-diffusion equations is inefficient due to the high computational load, which increases processing time as the number of pixels to be processed grows.
Innovation Solution
The method involves identifying non-target regions in tomographic images and fixing the brightness values of voxels outside the focused regions, integrating adjacent voxels into nodes, and performing reaction-diffusion calculations only on the focused regions, thereby reducing the calculation amount and processing time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If reaction-diffusion equation calculation is performed for all pixels to ensure accurate region extraction, then segmentation accuracy is improved, but processing time increases significantly
Solution Approach 1:
The patent divides the image processing into two distinct segments: (1) preprocessing step that identifies and marks non-target regions, and (2) reaction-diffusion calculation step that is applied only to remaining pixels. This segmentation allows the computationally intensive reaction-diffusion equation to be executed on a reduced subset of pixels while maintaining overall segmentation accuracy through the preliminary non-target region identification.
Solution Approach 2:
The patent extracts and removes non-target regions from the processing scope before applying the reaction-diffusion equation. By identifying pixels that definitely do not belong to the target organ based on brightness value thresholds, these pixels are taken out of the subsequent calculation process, reducing the computational load while preserving the accuracy of target region extraction.
2Reliability
If reaction-diffusion calculation is performed on all voxels to maintain processing completeness, then region extraction completeness is improved, but computational load increases
Solution Approach 1:
The patent performs preliminary identification of non-target regions before the main reaction-diffusion calculation. By预先 (in advance) marking pixels that fall outside the target organ based on brightness thresholds, the system prepares the data structure to exclude these pixels from subsequent intensive calculations, thereby reducing computational complexity while maintaining extraction completeness.
Solution Approach 2:
The patent applies reaction-diffusion calculation only to the necessary subset of pixels rather than all pixels in the image. By performing partial action on only those pixels that could potentially belong to the target organ, the computational load is significantly reduced while the extraction remains complete for all relevant regions.
Data Source
Figure 1~2
Figure 3A~3B
Figure 3C~5
AI summary
An information processing method includes: converting voxel data to node data in which a voxel, which has a brightness value that is outside a certain brightness value range, is set as a first node, and a voxel, which has a brightness value that is within the certain brightness value range, is set as a second node that has a capability to extract relating nodes based on a neighborhood relationship between voxels; performing, for each second node, a calculation processing to calculate an output value of a reaction-diffusion equation by using a value corresponding to a brightness value of the second node and values corresponding to brightness values of relating nodes extracted from the second node, a predetermined number of times; and determining a brightness value of each second node from the output value of the reaction-diffusion equation after performing the calculation processing the predetermined number of times.