Image Pre-Processing with ROI Visibility Selection for HDR Frame Rates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High frame rate image processing is prohibitively expensive due to the need for complex and costly image processing systems, especially when combined with High Dynamic Range (HDR) techniques that require computationally intense image synthesis.
Innovation Solution
An image pre-processing method that identifies regions of interest, performs visibility measures, and selects a subset of images based on these measures to feed into an image processing pipeline at a lower frame rate, reducing processing power requirements and leveraging image redundancy to select higher quality images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If high frame rate image processing is performed, then image quality and temporal resolution are improved, but system cost and processing complexity increase prohibitively
Solution Approach 1:
The patent performs preliminary actions by capturing multiple images at high frame rate with different exposure times before actual processing occurs. These pre-captured images are then evaluated using visibility measures to select only the most suitable ones for HDR processing, reducing the computational burden while maintaining high temporal resolution
Solution Approach 2:
The patent applies local quality by performing visibility measures specifically on regions of interest (such as specular highlights in eye images) rather than processing entire images uniformly. This selective approach focuses computational resources on critical areas, reducing overall processing complexity while preserving image quality where it matters most
2Measurement precision
If HDR image synthesis is performed on every captured image, then dynamic range is maximized, but processing time and computational power increase significantly
Solution Approach 1:
The patent changes parameters by varying exposure times across multiple captured images and using visibility measures to evaluate image quality. By adjusting exposure parameters and selectively processing only high-quality images, the system maximizes luminance range accuracy while minimizing processing time
Solution Approach 2:
The patent applies partial action by performing HDR synthesis only on a selected subset of images that meet visibility criteria, rather than processing every captured image. This selective approach processes fewer images (excessive action on quality, partial on quantity) while maintaining overall dynamic range accuracy
3Measurement precision
If complex image synthesising techniques are applied to maximise luminance range, then HDR quality is improved, but processing cost and system complexity increase
Solution Approach 1:
The patent performs preliminary visibility assessment on captured images before applying complex HDR synthesis techniques. By pre-evaluating images using computationally simple visibility measures, the system identifies which images are suitable for expensive HDR processing, reducing overall system implementation cost while maintaining luminance representation accuracy
Solution Approach 2:
The patent introduces visibility measures as an intermediary step between image capture and HDR synthesis. This intermediary evaluation layer filters images based on quality criteria, allowing complex HDR processing to be applied only when necessary, thereby reducing system complexity and implementation cost
4Speed
If real-time processing of high frame rate images is performed, then temporal resolution is maintained, but processing power requirements become prohibitive
Solution Approach 1:
The patent extracts and processes only the most relevant images from the high frame rate sequence by applying visibility measures. By taking out only the high-quality images that meet selection criteria for further processing, the system maintains temporal resolution for critical moments while dramatically reducing processing power requirements
Data Source
AI summary
Described herein is an image pre-processing system and method. One embodiment provides a method (500) including: at step (501), receiving a plurality of images captured at a first frame rate, the plurality of images captured under at least two different image conditions; pre-processing the plurality of images by: at step (502) identifying one or more regions of interest within the images; at step (503), performing a visibility measure on the one or more regions of interest; and, at step (504), selecting a subset of the plurality of images based on the visibility measure; and, at step (505), feeding the subset of images to an image processing pipeline for subsequent processing at a second frame rate that is lower than the first frame rate.