Image Processing Apparatus N-Order Energy Minimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods face challenges in efficiently segmenting images using high-order energy functions, particularly due to the difficulty in setting sparse high-order energy, which results in large calculation amounts and inefficient segmentation.
Innovation Solution
An image processing apparatus and method that employs N-order energy minimization by specifying and selecting voxels based on outlines similar to the target region, using energy setting mechanisms to differentiate between target and non-target regions, and applying the QPBO algorithm for efficient labeling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-order energy functions are used for image segmentation, then segmentation accuracy is improved, but calculation amount increases
Solution Approach 1:
The patent segments the high-order energy function into multiple lower-order energy functions by dividing the image into multiple regions and processing each region separately. This allows the use of efficient graph cut algorithms on smaller sub-problems while maintaining the benefits of high-order energy modeling, thereby reducing overall computational complexity while preserving segmentation accuracy.
Solution Approach 2:
The patent transforms the high-order energy minimization problem into a series of lower-order problems by introducing spatial dimensionality through region division. By processing the image in multiple regions rather than as a single high-order problem, the computational complexity is reduced from O(N^k) to O(m*N) where m is the number of regions and N is the number of pixels, while still capturing high-order interactions within each region.
2Measurement precision
If sparse high-order energy is not set properly, then segmentation quality improves, but computational complexity increases
Solution Approach 1:
The patent applies different energy function orders to different regions of the image based on local characteristics. By identifying regions where high-order interactions are most beneficial and applying high-order energy only where needed, the method maintains segmentation quality while reducing overall computational complexity compared to applying high-order energy uniformly across the entire image.
Solution Approach 2:
The patent uses partial high-order energy modeling by applying high-order interactions only to specific critical regions rather than the entire image. This partial application of high-order energy functions achieves sufficient segmentation quality for the most challenging regions while avoiding the computational burden of applying high-order models everywhere, effectively balancing quality and complexity.
Data Source
AI summary
When binary labeling is performed, an outline specification unit specifies a first outline present toward a target region and a second outline present toward a non-target region, and which have shapes similar to an outline of the target region. A voxel selection unit selects an N number of voxels constituting all of the first outline and the second outline. The energy setting unit sets N-order energy when a condition that all of the voxels of the first outline belong to the target region and all of the voxels of the second outline belong to the non-target region is satisfied smaller than the N-order energy when the condition is not satisfied. After then, labeling is performed by minimizing energy.