Medical Image Stitching via Subpixel Feature Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current medical imaging techniques face challenges in effectively stitching multiple images of a region of interest to produce a high-quality, seamless image, particularly in achieving subpixel-level registration accuracy for improved image stitching efficiency and quality.
Innovation Solution
A system and method for image stitching that involves obtaining reference and target images, determining feature points based on preliminary and superior registration accuracy, and generating a stitched image by updating feature points and removing rulers from the images, utilizing a processor and storage device to improve image alignment and quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image stitching methods are used, then the stitching process is simpler and faster, but the registration accuracy remains at pixel-level rather than achieving subpixel-level precision
Solution Approach 1:
The patent segments the feature point matching process into two distinct stages: preliminary matching to establish initial correspondences, and refined matching to achieve subpixel-level precision. This segmentation allows the system to achieve high accuracy without overwhelming computational complexity by breaking down the complex registration task into manageable steps.
Solution Approach 2:
The patent performs preliminary feature point matching before the refined subpixel-level matching. This preliminary action establishes initial correspondences that guide the subsequent high-precision matching process, reducing the search space and computational burden while ensuring final subpixel-level accuracy.
2Measurement precision
If multiple iterations of feature point refinement are performed, then subpixel-level accuracy is achieved, but the processing time increases
Solution Approach 1:
The patent performs refinement iterations only for feature points that require improved precision, rather than uniformly processing all feature points. This partial action approach achieves subpixel-level accuracy where needed while minimizing unnecessary computational overhead for already-sufficient matches, thereby reducing overall processing time.
Solution Approach 2:
The patent implements a feedback mechanism where the system evaluates the quality of feature point matches and selectively applies refinement iterations based on measured accuracy requirements. This feedback-driven approach ensures that computational resources are allocated efficiently, performing additional iterations only when and where needed to achieve the desired subpixel-level precision.
Data Source
AI summary
A method for image stitching is provided. The method may include obtaining a reference image of a first portion of a subject and a target image of a second portion of the subject, and determining at least one pair of feature points based on a preliminary registration accuracy. The first and second portions may at least partially overlap with each other. Each pair may include a reference feature point in the reference image and a target feature point in the target image that matches the reference feature point. For each pair, the method may further include determining an updated pair of feature points based on a superior registration accuracy higher than the preliminary registration accuracy and a neighboring region of the target feature point of the pair. The method may further include generating a stitched image based on the at least one updated pair.