Multi-Mode Image Alignment via Difference Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current inspection systems face challenges in aligning images from multiple modes of an imaging subsystem, making it difficult to use these images complementarily for defect detection and review due to differences in image characteristics, noise sources, and the large number of variable settings, which complicates mode selection and inter-mode image alignment.
Innovation Solution
A system and method that involves generating difference images for specific locations of interest by subtracting reference images from test images in each mode, aligning these difference images, and determining information for alignment, allowing for the alignment of images from different modes to a common design and subsequently to each other.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple modes are used for inspection to detect different defect types, then defect detection sensitivity is improved, but image alignment difficulty increases
Solution Approach 1:
The patent segments the image alignment process into multiple independent steps: (1) aligning each mode's image to the design separately, (2) generating difference images for each mode, and (3) aligning the difference images to each other. This segmentation transforms the complex direct problem of aligning multi-mode images into manageable sequential tasks, resolving the contradiction between using multiple modes and managing alignment complexity.
Solution Approach 2:
The patent introduces difference images as an intermediary element between the original multi-mode images. By subtracting reference images from test images to create difference images, the system creates a standardized intermediate representation that is easier to align across different modes. This intermediary step simplifies the alignment process while preserving the complementary information from multiple modes.
2Adaptability or versatility
If the number of variable settings is increased to accommodate different specimen types, then inspection versatility is improved, but mode selection complexity increases
Solution Approach 1:
The patent implements a universal alignment framework that works across all inspection modes and specimen types. The difference image generation and alignment process is mode-agnostic, meaning the same methodology applies regardless of which specific mode or parameter combination is used. This universality allows the system to handle diverse specimen types and defect characteristics without requiring separate alignment procedures for each mode.
Solution Approach 2:
The patent leverages parameter changes as the basis for creating different inspection modes while maintaining a consistent alignment approach. By allowing flexible adjustment of imaging parameters (such as optical settings, detector configurations) to create specialized modes for different defect types, the system achieves versatility. The alignment process then treats these parameter-varying modes uniformly through the difference image methodology, reducing selection complexity.
3Loss of time
If images from different modes are aligned directly to each other, then alignment speed is improved, but alignment accuracy deteriorates due to image differences
Solution Approach 1:
The patent performs preliminary alignment of each mode's image to the design before attempting to align modes to each other. This preliminary action establishes a common reference frame based on the design, ensuring that all mode images are pre-positioned correctly relative to the specimen's intended geometry. This preliminary step significantly improves the accuracy of subsequent inter-mode alignment while maintaining efficiency.
Solution Approach 2:
The patent replaces direct mechanical/image-based alignment between different modes with a computational substitution approach. Instead of relying on direct feature matching between dissimilar mode images, the system substitutes the alignment process with mathematical operations on difference images and design-based coordinate transformations. This substitution achieves higher accuracy by using computational methods that are less sensitive to the inherent differences between imaging modes.
Data Source
AI summary
Methods and systems for aligning images of a specimen generated with different modes of an imaging subsystem are provided. One method includes separately aligning first and second images generated with first and second modes, respectively, to a design for the specimen. For a location of interest in the first image, the method includes generating a first difference image for the location of interest and the first mode and generating a second difference image for the location of interest and the second mode. The method also includes aligning the first and second difference images to each other and determining information for the location of interest from results of the aligning.