Pixelated Detector HDR Counting with Sparsity-Based Image Merging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current detectors are limited in generating high-quality images from pixelated detectors as they must choose between integrating mode for high-flux regions and counting mode for low-flux regions, leading to compromised data quality in either case, especially in applications like 4D STEM and EELS where both high and low flux regions coexist.
Innovation Solution
The method involves generating a hybrid image by high dynamic range counting, which uses a sparsity map to differentiate between low-flux and high-flux regions, applying event analysis and normalization to merge these images, allowing for high signal-to-noise ratio in low-intensity regions while maintaining linearity in high-intensity regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If integrating mode is used for high-flux regions, then linearity is maintained in high-intensity regions, but signal-to-noise ratio deteriorates in low-intensity regions
Solution Approach 1:
The patent divides the image into multiple regions based on a sparsity map: high-flux regions are processed using integrating mode to maintain linearity, while low-flux regions are processed using counting mode to improve signal-to-noise ratio. This segmentation allows each region to be optimized independently according to its flux characteristics.
Solution Approach 2:
Different processing modes are applied to different regions of the image based on local flux characteristics. The sparsity map identifies regions requiring counting mode versus integrating mode, enabling local optimization of image quality metrics appropriate to each region's signal intensity.
2Reliability
If counting mode is used for low-flux regions, then signal-to-noise ratio is improved in low-intensity regions, but linearity deteriorates in high-intensity regions
Solution Approach 1:
The patent segments the image processing approach by region type, applying counting mode specifically to low-flux regions identified by the sparsity map while using integrating mode for high-flux regions. This prevents the linearity degradation that would occur if counting mode were applied globally.
Solution Approach 2:
The processing mode is selected based on local flux characteristics in each region. Low-flux regions receive counting mode treatment for improved signal-to-noise ratio, while high-flux regions receive integrating mode treatment to preserve linearity, optimizing overall image quality.
3Device complexity
If a single mode is chosen for the entire image, then device complexity is reduced, but adaptability to varying flux conditions deteriorates
Solution Approach 1:
The patent implements a dynamic processing approach where the sparsity map automatically identifies flux characteristics in different regions, and the processing mode is adaptively selected for each region. This dynamic adaptation to varying flux conditions improves versatility while maintaining reasonable system complexity through automated classification.
Solution Approach 2:
The processing mode parameter changes based on the flux intensity parameter identified by the sparsity map. Regions are classified by their flux characteristics, and the appropriate processing mode is applied accordingly, enabling the system to adapt to varying flux conditions across different image regions.
Data Source
Figure 1
Figure 2A~2C
Figure 3A
AI summary
The present disclosure relates to an apparatus and methods for generating a hybrid image by high-dynamic-range counting. In an embodiment, the apparatus includes a processing circuitry configured to acquire an image from a pixelated detector, obtain a sparsity map of the acquired image, the sparsity map indicating low-flux regions of the acquired image and high-flux regions of the acquired image, generate a low-flux image and a high-flux image based on the sparsity map, perform event analysis of the acquired image based on the low-flux image and the high-flux image, the event analysis including detecting, within the low-flux image, incident events by an event counting mode, multiply, by a normalization constant, resulting intensities of the high-flux image and the detected incident events of the low-flux image, and generate the hybrid image by merging the low-flux image and the high-flux image.