HDR Image Registration Using Pixel Irradiance Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing HDR image processing techniques face challenges in achieving perfect registration across multiple images, leading to issues like ghosting artifacts and miss-registration, especially in dynamic scenes captured by camera phones with CCD or CMOS sensors.
Innovation Solution
A method involving obtaining pixel value to irradiance value mapping curves, selecting a reference image, determining pixel intensity value dependent weighting factors, and registering corresponding mapped pixels using similarity measures like normalized cross correlation and sum of squared differences to combine pixels from multiple images, ensuring accurate high dynamic range image generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple images are taken with different exposure values to capture higher radiance information, then HDR image quality is improved, but motion artifacts and registration errors occur due to camera shake and dynamic scenes
Solution Approach 1:
The patent performs preliminary registration and alignment of multiple captured images before combining them into an HDR image. By pre-aligning the images using feature matching and transformation techniques, the system ensures that corresponding pixels from different exposure images are correctly positioned, preventing ghosting artifacts and registration errors in the final HDR output.
Solution Approach 2:
The patent employs feedback mechanisms by evaluating the quality of image alignment through similarity metrics (such as normalized cross-correlation) and iteratively adjusting registration parameters. This feedback loop ensures that the registration process converges to an optimal solution, maintaining high reliability even in dynamic scenes with camera motion.
2Ease of operation
If existing registration techniques are used to handle dynamic scenes and camera shake, then some level of alignment is achieved, but perfect registration and artifact-free images are not guaranteed
Solution Approach 1:
The patent changes key registration parameters dynamically based on scene characteristics. It adjusts the weighting of different registration metrics (e.g., normalized cross-correlation vs. sum of squared differences) depending on whether the scene is static or dynamic, and modifies the tolerance thresholds for pixel matching. This adaptive parameter adjustment enables precise registration across varying scene conditions without requiring complex manual intervention.
3Loss of information
If multiple images are combined to create HDR image, then radiance information is improved, but ghosting artifacts and miss-registration problems occur
Solution Approach 1:
The patent introduces an intermediary processing stage between image capture and final HDR combination. This intermediary stage performs detailed registration verification using multiple similarity metrics and applies corrective transformations to align images that initially appear misregistered. By acting as a mediator, this stage prevents ghosting artifacts from propagating to the final HDR image while preserving the radiance information from multiple exposures.
Data Source
AI summary
A method for generating a high dynamic range (HDR) image from images of a scene obtained at one or more exposure values is disclosed. In this embodiment, one of the obtained images is selected as a reference image. Further, mapped images are obtained by mapping pixel intensity values to corresponding irradiance values in each image. Furthermore, a pixel intensity value dependent weighting factor is determined. Moreover, a set of images is identified from the mapped images for pixels in the reference image. Also, a set of corresponding mapped pixels is established for the mapped pixels in reference image in the set of images. Further, a similarity measure is computed for the mapped pixels of reference image and corresponding mapped pixels in the set of images. Furthermore, each mapped pixel of the reference image is combined with a subset of its established corresponding mapped pixels in the set of images.