Related Experiment Video
Updated: Jul 7, 2026
05:05
Four-Dimensional CT Analysis Using Sequential 3D-3D Registration
Published on: November 23, 2019
Joint MAP registration and high-resolution image estimation using a sequence of undersampled images
R C Hardie1, K J Barnard, E E Armstrong
1Dept. of Electr. and Comput. Eng., Dayton Univ., OH.
Summary
This study introduces a novel Maximum A Posteriori (MAP) framework to enhance image resolution and reduce aliasing from undersampled frames. The method jointly estimates registration parameters and the high-resolution image for improved imaging quality.
Area of Science:
- Image processing
- Computational imaging
- Signal processing
Background:
- Detector arrays in imaging systems often lack sufficient density, leading to aliased images, especially in infrared focal plane arrays.
- Previous methods for super-resolution from undersampled frames often require known registration parameters or use unsuitable techniques for aliased images.
Purpose of the Study:
- To present a Maximum A Posteriori (MAP) framework for estimating high-resolution images with reduced aliasing from a sequence of undersampled frames.
- To jointly estimate image registration parameters and the high-resolution image within this MAP framework.
Main Methods:
- A cyclic coordinate-descent optimization procedure is employed to iteratively update registration parameters and the high-resolution image.
- The proposed method addresses severely aliased images by integrating registration parameter estimation into the super-resolution process.
Main Results:
- Experimental results demonstrate the effectiveness of the proposed MAP algorithm in improving image resolution and reducing aliasing.
- Performance is validated using both visible and infrared imaging data, with quantitative error analysis and subjective evaluations provided.
Conclusions:
- The developed MAP framework offers a robust solution for super-resolution imaging from undersampled sequences, particularly in scenarios with severe aliasing.
- Jointly estimating registration parameters and the high-resolution image within the MAP framework leads to superior performance compared to prior approaches.